Material Breakthroughs in Smart Food Monitoring: Intelligent Packaging and On‐Site Testing Technologies for Spoilage and Contamination Detection

Despite extensive commercial and regulatory interventions, food spoilage and contamination continue to impose massive ramifications on human health and the global economy. Recognizing that such issues will be significantly eliminated by the accurate and timely monitoring of food quality markers, smart food sensors have garnered significant interest as platforms for both real‐time, in‐package food monitoring and on‐site commercial testing. In both cases, the sensitivity, stability, and efficiency of the developed sensors are largely informed by underlying material design, driving focus toward the creation of advanced materials optimized for such applications. Herein, a comprehensive review of emerging intelligent materials and sensors developed in this space is provided, through the lens of three key food quality markers – biogenic amines, pH, and pathogenic microbes. Each sensing platform is presented with targeted consideration toward the contributions of the underlying metallic or polymeric substrate to the sensing mechanism and detection performance. Further, the real‐world applicability of presented works is considered with respect to their capabilities, regulatory adherence, and commercial potential. Finally, a situational assessment of the current state of intelligent food monitoring technologies is provided, discussing material‐centric strategies to address their existing limitations, regulatory concerns, and commercial considerations.


Introduction
Despite significant technological and regulatory efforts, issues pertaining to food waste and foodborne illness continue to persist globally. [1]With regards to food waste, societal reliance on predicted, static expiry dates is the primary contributor, as it generates the disposal of significant volumes of edible food each year. [2]While efforts to better map out the spoilage patterns of individual food types have been made, spoilage is heavily influenced by environmental conditions, making it a non-standardized process. [3,4]Strategies to monitor spoilage in real-time are currently not applied commercially, specifically because conventional means of food monitoring -such as bacterial culturing, are time-consuming, tedious, and futile on an individual product monitoring basis -a prerequisite for spoilage monitoring given its variable nature. [5]This has yielded significant interest toward the development of sensors that can monitor spoilage-related changes within food in real-time.While food spoilage is mediated by many different biochemical and microbial changes, shifts in biogenic amine concentration, pH, microbial content, and lipid oxidation have been identified as main contributors (Figure 1a). [5,6]These changes heavily influence the organoleptic properties of food, as detailed in Figure 1b.From a sensing perspective, changes in the former two have garnered the most interest, largely due to the diverse means by which these shifts can be tracked using relatively simple reaction cascades.
On the other hand, foodborne illness is largely caused by pathogenic bacteria that contaminate food products at various stages during the food production pipeline (Figure 1c). [7]The consumption of such contaminated foods often results in illnesses with potentially severe prognoses.While several regulatory measures are in place to mitigate the likelihood of such contamination events, pathogenic outbreaks remain rampant.10] Efforts to proactively detect contamination on a commercial scale currently involve time-consuming and expensive processes that require complex sample processing and shipment to external facilities.Common evaluation strategies include chromatographic and spectroscopic techniques, as well as in vitro bacterial culturing and sequencing. [11][14][15] Such sensors target one of two key avenues to real-world implementation: on-site use or in-package integration.On-site use requires sensing platforms with high sensitivity, that can offer output signals within a short period of time.These systems are permitted greater flexibility regarding sample manipulation, as test samples can be readily processed prior to evaluation.Opposingly, in-package integration requires low cost, autonomous platforms that use biocompatible reagents, which can be applied in situ.Such platforms represent a gold standard in food monitoring, as their integration onto packaging materials yields smart food packaging -materials that monitor products over their entire lifespan.
Given that spoilage monitoring requires such constant evaluation, efforts in this space have largely centered upon the creation of in situ platforms.These sensors report shifts in biogenic amine concentration and pH via electrochemical, fluorescent, or colorimetrical means.That being said, some on-site systems have been developed for spoilage monitoring, in hopes of creating commercial scale means of rapid, ultrasensitive testing that can be used as needed.Opposingly, efforts to enable the monitoring of pathogenic contamination in food have yielded both onsite and in situ sensors that offer value at different stages of the food production pipeline.Specifically, on-site platforms are extremely valuable within commercial environments, where rapid, ultrasensitive platforms offer significant value over conventional techniques.On the other hand, in situ platforms -while usually less sensitive, offer a means by which contamination can be monitored within individual products at a consumer level.
The vast majority of food-targeting sensors are composed of sensing agents immobilized onto or within an associated substrate, as free-state sensors are largely unfeasible in this space. [16]Thus, while the capabilities of the incorporated sens-ing agents undoubtedly influence resultant sensor performance, application-informed material selection is equally contributory.To this end, the material properties, mechanical stability, and biocompatibility of candidate substrates represent key considerations that determine their suitability for particular platforms.While a wide range of metallic and polymeric materials have shown applicability in the food monitoring space, a critical review of the real-world viability and optimal use cases of reported substrates is absent.
With consideration toward the rapid progress being made in the smart food monitoring space, this review discusses recent advancements in the development of intelligent sensors designed for on-site and in situ detection of food spoilage and contamination.First, an overview of substrate materials used within smart food sensors is provided and their unique material offerings are established.The use of these material in food sensing is then contextualized through a critical review of recently reported sensing platforms.Specific emphasis is placed on the impact of selected substrates on resultant sensor performance.In parallel, recognizing the plethora of different approaches being explored in food sensing, the limitations of presented platforms are highlighted, to assess their real-world viability.Finally, an assessment of the current state of smart food monitoring technologies is provided and future directions aimed at yielding real-world sensors with regulatory and commercial potential are discussed.

Substrate Materials for Smart Food Sensing Platforms
Materials used in the smart food sensing space can be broadly categorized into metallic, polymeric, or carbon-based in nature, where their vastly different characteristics yield distinctive use cases (Figure 2a).Specifically, while functional surface chemistry or agent entrapment potential is instrumental for all such sensing platforms, other desirable traits are informed by the target application.In general, on-site testing technologies prioritize material contributions to sensing signal transduction to actualize ultrasensitive, rapid target detection.Opposingly, in situ platforms generally target substrates that offer long-term stability in foodrelevant environmental conditions and high biocompatibility, to combat concerns pertaining to material leaching.Emphasis on such traits has yielded the development of various emerging materials produced through modifications to established base materials with favorable characteristics.While the value of such materials is best understood in the context of specific applicationsas described in Section 3 of this review, this section provides an introductory overview to base substrates used in this space and their respective properties (Figure 2b).

Metallic Substrates
While many metallic agents have been established as biocompatible, approval for use within food packaging is stringent. [19]Recognizing that they are also quite expensive, commercial incorporation of such materials as sensing substrates within individual products is largely unfeasible.As such, metallic substrates can be considered best suited for on-site testing technologies, where ) Key contributors to food spoilage and their impact on the organoleptic properties of food.c) Schematic illustrations of the food contamination timeline, and stages at which pathogenic contamination can occur.d) Annual global illness rates caused by respective pathogens obtained from the World Health Organization.Regulatory limits as defined by the Food and Drug Administration for ready-to-eat (RTE) products. [17,18]Infectious dose recorded in literature.Produced using BioRender.
Adv. Mater.2024, 36, 2300875 they offer several optical and chemical properties that can be exploited for rapid, ultrasensitive detection of target entities. [20]old (Au) is by far the most commonly used metallic substrate in this space.[23] Yet, the value of Au substrates is most clearly demonstrated through their ease of functionalization. [24]While Au is largely inert, it is readily functionalized with thiol-modified agents through the formation of Au-S bonds.In this way, well-established self-assembled monolayers of sensing agents can be immobilized onto the substrate without the introduction of a crosslinking agent. [21,25]ese densely packed probe monolayers resultantly contribute to high detection sensitivity.That being said, the stability of this crosslinking approach is somewhat questionable given the relatively weak nature of the Au-S bond. [26]he density of detection probes can be further increased per unit area using the diverse Au form factors available.Specifically, aside from a base flat substrate, micro-and nanoscale structuresnamely wrinkles and flowers, are well-established in literature. [27]uch structural modifications offer higher functional surface area -mediating increased probe density, while also offering antibiofouling properties in certain circumstances, which improves detection performance. [28][31][32] Further, their size confinement effect yields electronic and optical properties that facilitate their use as electrostatic target detectors and colorimetric transducers, respectively. [29]Namely, Au nanoparticles (AuNPs) are well noted for their ability to both quench or enhance the signal of fluorescent molecules, based on their size and specific characteristics of a given system. [30,33]ecognizing the high cost of pure metallic substrates, there has been continued emphasis on the development of hybrid platforms that employ metallic agents as a component of the sensing substrate.Namely, recent works in the food sensing have employed Au and silver (Ag) nanoparticles as substrate components within plasmonic sensors. [34]Similar to Au, Ag offers high plasmon resonances at the visible-near infrared region of interest, making it a suitable substrate for such sensors. [34]Importantly, Ag also shows a propensity for thiol binding, presenting a simple strategy for surface functionalization.Other nanoscale structural forms of Ag have also been employed, with nanodendrites and nanowires both being used for food-related sensing applications for higher surface area and faster electron transfer, respectively. [35,36]The latter innovation has yielded significant improvements in the efficiency of sensing cascades, reducing detection times to less than two hours in some cases. [35][39] Cost considerations have also yielded increasing interest toward aluminum and silicon-based substrates for plasmonic sensors, but such efforts have not been presented in food sensing. [34]oncurrently, there has been a growing interest in the use of metal-organic frameworks (MOFs) for food biosensing.MOFs are hybrid structures of metal ions conjugated to organic ligands. [37,40]Based on the component type and number of ions, as well as the structure of the organic ligand, these MOFs have unique selectivity for target analytes based on shape and size.Typically, these MOFs are composed of lanthanide ions or other fluorophores, that exhibit strong, sustained signaling upon target binding. [41]MOFs have been explored extensively in the sensing space as their structure is easily fine-tunable and modifiable, which permits optimization of structural porosity and target specificity. [40]Extensive work has been conducted to synthesize MOFs responsive to cues of spoilage and contamination in food.In the intelligent packaging space, MOFs are convenient as they can be synthesized on packaging materials themselves to make nanocomposite films, often with limited intervention. [42]These MOFs act as nanofillers within the film matrix, enhancing the structural performance of the polymer while acting as analyte receptors and signal transducers.Optimized incorporation of these MOFs into various materials is ongoing to maximize their sensing functionality within resultant intelligent films.

Polymeric Substrates
Principally, polymeric substrates dominate the food sensing space, owing to their low cost and high tunability. [34]Such substrates are frequently used for both on-site sensors and in situ package sensors.As such, considerations toward material viability are largely defined by the target application.Most prominently, biodegradable polymers -such as cellulose, starch, and chitosan, and plastics -such as polystyrene and polyethylene, have been used in this space. [20,43]nterest in biodegradable polymers is largely driven by environmental concerns surrounding conventional food packaging materials, making such substrates a focus within in situ detection platforms. [44]Such polymers exhibit high biocompatibility and can be embedded with biosensing agents relatively easily due to their tunable porosity. [43]However, they have historically offered poor stability in food-relevant conditions, limiting their real-world applicability. [45]To this end, the introduction of various modifications has produced several emerging substrates that are detailed in Section 3. Briefly, recently developed materials have aimed to strike a balance between environmental resilience -largely through the maximization of mechanical and thermal stability, and sensing performance, wherein the incorporation of sensing agents within polymer substrates has reduced film strength at times. [46,47]In parallel, efforts have also been made to improve the stability of polymer-entrapped, volatile sensing agents.[50] With regards to on-site detection platforms, cellulose paper-based substrates have been commonly employed, owing to their low cost and versatility. [51]Here, poor stability is of limited concern given the significantly shorter use time.
Despite their detrimental environmental impact, conventional packaging plastics are often favored for in situ applications due to their high environmental stability and barrier capabilities. [52]irect incorporation of biosensing agents onto such polymers can prove comparatively tedious however, requiring surface activation steps using various physical or chemical strategies that offer varying degrees of consistency -especially when applied on a commercial scale. [53]Emerging platforms that employ such substrates have thus developed means by which probe immobilization can be simplified and the sensitivity of the resultant platforms can be maximized.Such strategies are also detailed atlength in Section 3.

Carbon-Based Substrates
Carbon-based substrates are heavily used in on-site electrochemical biosensors, where they function as effective signal transducers.Compared to metallic substrates, carbon substrates offer high surface impedance for electrochemical detection at a much lower cost. [54]The use of carbon is further substantiated by its successful deployment within screen-printed electrodes, which have been noted for their compact form factor and high detection sensitivity. [20,43]Efforts to further improve the performance of these substrates have yielded significant focus toward carbon nanotube substrates, which offer high surface area, antifouling properties, and rapid electron transfer -characteristics that all act to improve detection sensitivity. [55]Other nanostructures -namely nanohorns, have also been explored due to similar advantages. [56]While carbon substrates offer limited functional surface chemistry, conventional crosslinkers can be used for the biofunctionalization of these substrates with target-specific recognition molecules. [55]The pairing of such substrates with metallic particles has also yielded improvements in biofunctionalization. [43] Concurrently, graphene-based substrates have also established themselves as keystone carbon-based substrates for biosensing applications.Alongside their high mechanical strength, favorable surface properties, and high electron transfer capabilities, these substrates are cost-effective and optically transparent. [57]This latter property mediates compatibility with various optical transduction strategies.Such optical strategies can be further supported by the inherent quenching activity of graphene, which has been exploited in fluorescence sensing. [57]From a manufacturing perspective, laser-induced graphene producers have found the use of graphene as a base substrate very attractive.In fact, laser-induced graphene has been formed on existing food packaging in recent works. [58]Outside of conventional layered graphene-based substrates, the use of graphene quantum dots is also increasing, owing to their tunable photoluminescence and photostability, enabling them to act as transducers. [59]Further, combinatory approaches that pair such agents and metallic nanoparticles have yielded materials with unique properties. [60]

Material Developments in Smart Food Sensing Platforms
With growing demand for improvements in food safety, biogenic amine, pH, and pathogen monitoring have all experienced significant material-driven progress in detection performance, mechanical stability, and commercial potential in recent years.In this section, we detail emerging platforms, with specific consideration toward their real-world viability.

Biogenic Amine Monitoring
An important mediator of spoilage in food is the growth of bacteria and other microorganisms. [4]This growth is influenced by the environmental conditions in which food is stored, including temperature and humidity.Food handling can also introduce extrinsic bacteria into food that then alter the spoilage process.The tracking of bacteria-mediated changes in food has been identified as a potential avenue for the real-time monitoring of food quality.Specifically, tracking compounds produced by the proteolytic breakdown of animal tissue has garnered significant interest.These entities are often toxic when consumed at high levels, further substantiating their use as a detection target.Biogenic amines are one such class of organic by-products, formed by the decarboxylation of amino acids and other nitrogenous compounds. [61,62,63]Histidine, tyrosine, lysine, and ornithine are common sources of biogenic amines during spoilage, forming histamine, tyramine, cadaverine, and putrescine, respectively (Figure 3a). [61]mportantly, not all bacterial species contain the decarboxylase enzymes necessary to produce these amines, meaning that spoilage and contamination are not universally correlated with their presence. [64,65]Thus, the application of biogenic amines as an indicator of food quality is limited to certain food products, and to the bacteria that typically contaminate them.To this effect, fish and other seafood have been identified as a prime target for biogenic amine monitoring.These products exhibit high concentrations of histidine within their muscles, which is converted into histamine -an agent that can induce poor clinical outcomes, by microbes naturally present within the fish. [66,67]Similarly, dairy products also exhibit high biogenic amine formation as they begin to spoil, with tyramine being the most prevalent amine in these foods. [68]urrent methods to measure biogenic amines in food include gas and high-performance liquid (HPLC) chromatography, as well as capillary electrophoresis, which all identify specific amine compounds based on their unique structural properties. [68]Alternatively, real-time quantitative PCR (qPCR) can also be used to detect the increased expression of decarboxylase enzymes in processed food samples. [69]Efforts have been made to improve laboratory techniques for analyte detection, but these methods involve complex equipment, extensive sample processing, and readout analysis, making them unfeasible for on-site, in situ detection.While these methods are generally useful for the detection of biogenic amines, an ideal sensor for food monitoring would exhibit high sensitivity and specificity to biogenic amines present within complex food matrices, without the need for sample processing.A variety of systems have been explored under this premise, based on their selective reactivity with different basic/nitrogenous organic compounds.Three common platforms used within this space are nanoparticles, reactive organic compounds (ROCs), and biorecognition probes with specificity for biogenic amines.These platforms interact with biogenic amines via electrostatic, covalent, and conformational interactions that transduce into detectable signals (Figure 3b).While they are produced within the food matrix as a by-product of protein degeneration, biogenic amines are quickly emitted from food as gaseous compounds as they are volatile in nature.As such, an important physical consideration in the design of these platforms includes the ability of materials to sequester and retain these volatile amine molecules.

Nanoparticle-Mediated Detection
Nanoparticles have become a staple in the biosensing space due to their diverse physical and chemical properties, ease of functionalization, and high surface area-to-volume ratio. [70]These are all properties that have been leveraged in the development of electrochemical and colorimetric biogenic amine sensors.Colorimetric detection usually involves the induction of nanoparticle aggregation in response to target binding, resulting in an observable change in color.For real-world application, an observable colorimetric threshold can be set in accordance with by-product levels indicative of food spoilage, but the qualitative nature of this approach limits system accuracy.In contrast, electrochemical sensors use electrodes or other probes to measure impedimetric and conductive changes in response to nanoparticle-target binding.Within electrochemical biogenic amine sensors, targeted binding of these compounds permits their quantification in real-time.Both systems have been explored heavily in recent years and offer unique benefits and drawbacks, that tailor their use toward varying circumstances. [71]With regards to colorimetric sensing, AuNPs are used extensively, owning to their vivid transition from deep red to purple when shifting from an aggregated to non-aggregated state (Figure 3c.i).Such an approach has been applied most commonly in solution, where the free movement of both the target and the nanoparticles maximizes binding events.For biogenic amine de-tection, these binding events are most easily induced via electrostatic interactions that occur between positively charged target amines and negatively charged citrate groups functionalized onto the nanoparticles.These interactions induce aggregation via various hydrophobic interactions or the electrostatic binding of multiple nanoparticles to a single target molecule, which brings them closer together in proximity.The precise mechanism of individual systems is often left unclear.El-Nour et al. used this strategy in the development of a histamine sensor for chicken, where binding to the target resulted in AuNP aggregation. [72]The resultant sensor demonstrated a limit of detection of 0.6 μm but required solid test samples to be homogenized prior to evaluation.Unfortunately, target detection via electrostatic interactions means that the system does not respond to the target with high specificity.Rather, the nanoparticles can react with any similarly charged compounds present in the test matrix, leading to false positives.The existence of a plethora of organic compounds and proteins in complex food matrices makes this a significant concern.In an effort to create a colorimetric histamine sensor with higher specificity, Lapenna et al. investigated unmodified AuNPs, that lacked the typically incorporated citrate groups. [73]They found that the aliphatic amino group and the imidazole ring of histamine were both capable of binding to the gold surface.As such, histamine can act as a crosslinker between AuNPs in the absence of capping agents such as citrate.The resultant sensor was assessed in wine and showed a limit of detection of 0.2 μm.Ultimately, while the colorimetric nature and high sensitivity of these platforms are intriguing, their in-solution nature and need for sample homogenization prevents their use in situ (Table 1).
Aiming to eliminate processing and homogenization steps, many recent works have developed biogenic amine-targeting intelligent noses. [74,75]These intelligent noses often use nanoparticle-based reaction cascades to monitor gas production, which can lead to colorimetric signals in a solid-state format.Further, some approaches use closed systems to retain volatile amines for prolonged amounts of time -a strategy that offers improvements in detection sensitivity.By overcoming the need for direct contact with food, these platforms eliminate concerns pertaining to sensor surface fouling and limit the likelihood of cross-reactivity with non-target compounds.There is significant diversity in the mechanisms these systems use to yield accurate signals, but food-targeting systems demand specific emphasis on use of ease to enable widespread deployment.
Under this premise, Gholampour et al. (2021) developed a colorimetric intelligent nose that employed a cadmium (II) ionembedded polyethylene-glycol (PEG) and polylactic acid (PLA) copolymer that released the immobilized ions in response to gaseous pH increases induced by the accumulation of vaporized amines. [76]PEG-PLA copolymers are commonly used in controlled-release systems, with slow degradation of the copolymer over time leading to the release of nanoparticles or drugs into a biological matrix. [77,78]Such a system offers control over the rate of Cd 2+ ion release into foods over time to interact with anionic amines.This triggers color changes that correlated with increases in amine vapors.To make their system applicable to in situ detection in food, they incorporated their polymer into agarose hydrogels, which permitted the permeation and increased retention of trimethylamine vapour into the detection system.The system showed successful colorimetric detection of spoilage at room temperature, but minimal change was observed after 24 days at 3 °C.The spoilage state of this sample was not specified, and thus the efficacy of the system remains unclear.While intriguing, the system is quite complex, limiting its commercial viability.Further, the toxic nature of cadmium would pose a significant hurdle for real-world application.
Similarly seeking to create an alternative intelligent nose, Tseng et al. (2017) developed a nanoparticle-based paper sensor for the detection of putrescine and spermidine. [79]Here, a mold was used to print silver (Ag) and Au nanoparticles onto paper substrates using reversal nanoimprint lithography.Nanoparticle immobilization onto solid substrates to create films for biosensing, rather than using them in an in-solution format reduced the level of user interference required.This strategy yielded over 85% transfer efficiency from the mold to the substrate.Using lithography to imprint the nanoparticles onto papers is cheap and efficient, as it eliminates the need for chemical modifiers.In the resultant sensing platform, interactions between the immobilized nanoparticles and the targeted amines led to a measurable wavelength dip shift as measured by localized surface plasmon resonance (LSPR).This same shift did not take place when the sensors were exposed to other common gases in the air.This system was evaluated as a spoilage detection platform for salmon and demonstrated a limit of detection of 13.8 ppm.From an application perspective, the authors argued that this system is well situated as a rapid, low cost, on-site detection tool.However, the need for LSPR measurements is likely to severely hinder widespread commercial use and eliminates the possibility of consumer level monitoring.This is common with the use of nanoparticles as a biosensing platform, as they do not elicit significant visual signals upon aggregation when immobilized onto materials.
Ultimately, the use of nanoparticles for biogenic amine detection is largely driven by their high surface area, ease of functionalization, and contributions to signal transduction within both colorimetric and electrochemical platforms.In comparison to other platforms, nanoparticles are also cost-effective as they are relatively easy to synthesize and functionalize. [22,32]Moreover, sensing cascades can be designed to activate nanoparticle aggregation at a predefined biogenic amine concentration, without any user interference. [72]Yet, while some recent works have shown an ability to differentiate target biogenic amines from chemically similar non-target compounds, nanoparticle-based systems offer limited specificity.Further, the incorporation of nanoparticle-based platforms in situ is difficult to implement due to regulatory hurdles.While some nanoparticles are FDA-approved for usage as therapeutics and drug-delivery agents for disease treatment, their use in consumables requires further investigation.As such, concurrent research has applied alternative agents that -in some cases, offer more specific interactions with biogenic amines, to maximize sensor accuracy.

Reactive Organic Compound-Mediated Detection
Various reactive organic compounds (ROCs) carry inherent reactivity with biogenic amines, offering a promising avenue for their detection.This reactivity can be non-specific, and mediated generally by positive charge, or can be specific for the structures of distinct biogenic amines.Changes in the electron distribution and the creation of new reactive by-products following ROCbiogenic amine reactions enable detection via electrochemical, colorimetric, or fluorometric transduction.A benefit to the use of such compounds is the relative ease with which they can be finetuned and modified to increase their target selectivity.While enhanced reactivity with amines and other basic compounds could also lead to an increase in background signals, incorporation into intelligent noses can limit such effects.Prevalent ROCs used for the detection of biogenic amines include pyrylium salts, donor-acceptor conjugation structures, metal-organic framework structures, and metallocomplexes.
Pyrylium salts are aromatic compounds that react with biogenic amines to derive pyridine, or similar conjugated analogs.This conversion changes the fluorescence profile of the compounds, enabling their use as chameleon labels (ChLs) -labels that change in color or fluorescence upon interacting with biological molecules.One study recently explored the use of pyrylium salts in crafting ChLs to detect putrescine and trimethylamine. [80]n this platform, the salts were deposited onto silica gel-coated aluminum strips to create amine-responsive ChLs.Other convenient material candidates for the development of ChLs include organic polymeric matrices such as cellulose, that readily absorb reactive probes.These labels underwent fluorescence quenching over time owing to increased binding of target amines.A LOD of 0.1μm was obtained for both biogenic amines in a gaseous state.When stored with fish samples at room temperature, the platform demonstrated an 85% quenching efficiency in 18 h, validating its use as an in situ monitoring system.Qi et al. (2021) developed a similar fluorescent sensor by immobilizing a pyrazine compound (PD-6) onto edible gelatin films. [81]Due to the entrapment of the sensing agent within the substrate matrix, a high density of the compound could be introduced relative to 2D surface immobilization.The sensor exhibited strong changes in fluorescence in response to target amine exposure, while displaying minimal reactivity with non-target amine-based compounds.These films were also reusable, as the addition of conjugate acids reversed biogenic amine binding and mediated the reproduction of the PD-6 derivative.Ultimately, pyrylium salts offer an easy-toincorporate tool for biogenic amine detection.However, considering that fluorescence signaling is difficult on a consumer scale, colorimetrically signaling ROCs have been studied in parallel.
To this end, donor--acceptor (D--A) conjugation structures are similar compounds that can be co-opted for use in intelligent films, changing in fluorescence or color in response to a variety of toxins and compounds present in food.Prosthetic substitution of D--A compounds can be used to induce changes in responsiveness to specific compounds.Duan et al. evaluated the use of a D--A structure substituted with benzoic acid (DPABA), which exhibited sensitivity to both solid and gaseous state amines released from fish during spoilage. [82]The formulation of DPABA ChLs involved simply soaking cellulose films in DPABA solution.In response to target exposure, DPABA exhibited changes in both color and fluorescence.When incorporated into packaging in the form of food ChLs, these sensors showed a limit of detection as low as 0.3-0.6 ppm of ammonia, fading in color from red to pink, and then to yellow in response to decreased freshness.A toxicity assay confirmed that DPABA had no cellular toxicity at the concentrations being used in the developed system.However, further toxicity studies are required for reactive compound-incorporating ChLs, including time-course studies that evaluate leaching into foods.The practicality of this system is hindered by the limited selectivity of DPABA though, which exhibits a degree of crossreactivity with non-target agents present within food.The evaluation of alternative prosthetic groups that offer higher specificity represents a viable next step, warranted by the colorimetric transduction these compounds offer.
[85][86] These compounds are composed of coordinately bonded organic ligands and metal ions, that collectively form complex multi-dimensional structures.When these compounds are paired with porous coordination polymers, hybrid polymers with distinct reactive properties are produced (Figure 3c.ii).These properties are easily tunable via manipulation of the constituent MOFs with different metals ions and organic groups.The use of MOFs for biogenic amine detection is substantiated by the structural voids they possess, which enable entrapment of gaseous compounds.Retention of these compounds does not permanently alter MOF structure, allowing each framework to sequester volatile compounds multiple times for enhanced detection.Based on their nature, these MOFs can undergo changes in luminescence, fluorescence, and electric potential upon target binding, offering flexibility in system design.Lanthanides are commonly used as the metallic element of MOFs as they can undergo photobleaching-resistant increases in luminescence. [41]With regards to biogenic amines in particular, these MOF-incorporating polymers have demonstrated responsiveness to amines present in both solid and gaseous states with high sensitivity.Modifications to the MOF structure using prosthetic ligands that selectively bind to the imidazole rings or aromatic groups of biogenic amines can be used to fine-tune system performance.Namely, Xu et al. ( 2017) developed amine-detecting cellulose hydrogels composed of methyl red (MR)-conjugated europium MOFs (EuMOFs). [87]Both MR and Eu ions exhibited fluorescence shifts in response to the presence of biogenic amines, owing to a pH-induced energy transfer.This was paired with a pH-induced colorimetric shift of MR.System evaluation using histamine gas demonstrated a limit of detection of 100 nm.That being said, the specificity of this platform remains unclear.Further, the use of pH indicator compounds such as MR raises concerns pertaining to toxicity.While the platform does not require direct contact with the food matrix, implementing such compounds within food packaging is unlikely to be regulatorily approved.
Rather than using such pH-responsive compounds, Yao et al. (2021) described a similar tetrahedron cage composed solely of lanthanide ions and a tris-beta-diketone ligand, the shape of which was optimized for the permeation of biogenic amines. [88]his system used the stabilizing ligand within the MOF structure itself to engage in covalent target binding, which subsequently enabled an inherent luminescent shift.Using spin-coating, a film composed of this MOF was produced, demonstrating an optical response to amine exposure within seconds.While this rapid response offers significant promise, the ligand binds target molecules through a simple carbonyl-amine reaction.The generalized nature of this reaction makes the system susceptible to false positives induced by other volatile nitrogenous, non-target small molecules.
A MOF-based platform with higher specificity was developed by Jindal et al. ( 2020), who used a cadmium MOF that exhibited fluorescence quenching upon reacting with biogenic amine vapors.Rather than using luminescence, this approach co-opts changes in autofluorescence that take place with chemical bonding.This shift is induced by the deprotonation of imidazolium cations within the MOF structure. [89]The structure showed ultrasensitive detection toward aliphatic biogenic amines to the order of 56 ppb owing to their higher compatibility with the system's reaction mechanism.A degree of affinity was also observed against aromatic biogenic amines, permitting generalized biogenic amine detection.Ultimately, such studies that fine-tune MOFs with more specific reactivity to distinct biogenic amine subtypes are required for the development of more applicable platforms.
The last commonly used ROC-based biogenic amine detection strategy involves the use of metallocomplexes.These are heterocyclic compounds that carry natural reactivity with positively charged amine-based species, which ligate their metal groups to change compound structure.In relation to biogenic amine detection specifically, Sahudin et al. recently developed sensing films by conjugating Zn to salphen, N, N'-phenylenebis (salicylidene) -a Schiff base, to form stable metal complexes. [90]These structures were then immobilized onto transparent silica microparticles, which bonded selectively to histamine vapor secreted from food and subsequently exhibited an increase in autofluorescence.These microparticles are both anionic and hydrophilic in nature, improving the ease of functionalization of the Schiff base, as well as the retention of cationic amines.When immobilized onto glass slides, these particles achieved a detection limit of 4.4 pm with histamine-spiked shrimp, increasing up to 73% in fluorescence after exposure.Importantly, the Zn-salphen complexes demonstrated binding selectivity for histamine and biological amines over other common amine-containing organic compounds, ensuring strong signal correlation with spoilage.The system also demonstrated increased selectivity toward histamine -an aromatic amine, relative to other aliphatic amines, ensuring a degree of specificity toward the target structure.Lai et al. ( 2021) applied a similar strategy to create fluorescent nanofilms that enable non-destructive, direct-contact detection of spoilage in food.They derived nanofilms on glass by condensing a Calix [4] pyrrole derivative (CPTH), which promoted metallocomplex self-assembly with a tetraphenethylene derivative (TPEBA) at an air/DMSO interface.TPEBA exhibits innate fluorescence that quenches after binding to volatile amines. [91]To increase system specificity toward aliphatic amines, the films were nanostructured, and yielded a limit of detection of 0.89 ppm.The effect of the nanostructures on specificity remained unclear however, as the study did not evaluate sensing performance in the presence of non-target, volatile organic compounds, which could competitively bind the nanofilms.
A variety of chemical probes thus exist that can be leveraged as intelligent noses for biogenic amine detection.D--A conjugation structures, MOFs, and similar metal-complex platforms carry the advantage of being highly tunable in terms of shape, specificity, chemical nature, and signal transduction, offering several avenues for performance optimization.Large-scale manufacturing strategies for MOFs as biosensors are currently being developed.A significant benefit of using these chemical probes is that they can often be recycled and reused within biosensing films. [92]Typically, treatment with acids and bases can cause a restoration of the structure of the parent compound, permitting repeated use.The primary limitation to their use for in situ monitoring thus becomes their long-term retention within packaging films.Recognizing that specificity toward subtypes of target biogenic amines remains the most prominent limitation of existing systems, such optimization represents the focus of future works in this space.

Probe-Based Detection
Like the aforementioned chemical agents, many biological macromolecules also carry inherent affinity for biogenic amines.Examples of such molecules include enzymes and antibodies, and functional oligonucleotides, all of which can be used in biogenic amine detection as biorecognition probes.Such systems have been explored extensively, owing to their high long-term stability, specificity, and selectivity.
DNA aptamers are a class of functional oligonucleotides that selectively bind specific molecular targets (Figure 3c.iii). [93]In contrast to antibodies, which offer similar functionality, aptamers are cheaper to synthesize, as they do not require the generation of hybridomas or the use of animals.They are also less prone to degradation and highly specific in their target recognition abilities, with limited cross-reactivity to related molecules. [94,95]uch probes can also be multiplexed within biosensors with relative ease, enabling the detection of an array of biogenic amines in parallel. [96]Considering these advantages, Lerga et al. described the derivation of H2, a DNA aptamer with high binding affinity for histamine, as quantified by its low dissociation constant (K D = 3-34 nm). [97]The authors subsequently incorporated the developed aptamer into a solution-based, on-site platform for histamine detection.By pairing biotinylated aptamers with streptavidin-horseradish peroxidase, colorimetric detection limits of 18 pm and 76 pm in buffer and urine were reported, respectively, with little competitive binding from other biomolecules.The efficacy of this system is somewhat reduced by the immobilization of histamine onto magnetic beads, as this restricts their orientation, limiting their interactions with aptamers.Nonetheless, the system remained highly sensitivity, showing high applicability as a sensor for liquid test matrices.Unfortunately, the use of such a reaction cascade for solid food matrices would likely prove unfeasible.Seeking to evaluate solid samples nonetheless, the same authors (2020) immobilized the H2 aptamer onto AuNPs to colorimetrically detect spoilage in fish samples via NaCl addition. [98]While the system showed no in situ value, given that it required substantial sample homogenization and processing prior to testing, this study sought to substantiate its use within complex food matrices.In this system, dispersed AuNP-H2 complexes -appearing red, withstood nonspecific NaClmediated aggregation due to the presence of aptamers.Upon biogenic amine exposure, the aptamer bonded selectively to histamine and the free AuNPs aggregated, appearing blue.A limit of detection of 8 nm was reported.
In another instance, Dwidar et al. used a previously derived A1-949 RNA-histamine aptamer to develop a fluorescent biogenic amine monitoring system with a detection limit of 1 μm. [99]They first constructed a fluorescently labeled L-DNA version of their aptamer, more stable than its RNA counterpart.This was then coupled with a complementary oligonucleotide that contained a quencher which was released following substrate binding, yielding a fluorescent signal.The A1-949 aptamer is less sensitive than the H2 aptamer with a dissociation constant of 370 nm, contributing to the system's lower limit of detection.Fluorescent aptamer labeling is more convenient than an HRP-conjugated system, as it does not require the addition of any external reagents.Again however, this system required sample homogenization and processing prior to testing, making it unfeasible for in situ use.
In contrast to DNA, the advantage of using certain peptides is the abundance of existing candidates that react with aminebased compounds physiologically.Antibody probes are one such peptide-based biorecognition element for monitoring amine production in food.Namely, Zeng et al. recently used histaminebinding antibodies within a lateral flow assay aimed at monitoring fish products. [100]The sensor was capable of detecting 0.25 mg kg −1 of histamine within complex test matrices within 15 min.Alternatively, Parate et al. developed a similar chip-based sensor consisting of histamine-specific antibodies conjugated to interdigitated electrodes for electrochemical detection of the target. [101]These sensors demonstrated a limit of detection of 30 μm in tuna broth.Ultimately though, these platforms have limited in situ applications, as they require a liquid interface to be able to capture histamine ions for effective detection.They do, however, offer some perspective on the viability of antibody-based lateral flow and electrochemical sensing, respectively, for on-site food monitoring.
Aside from amine-specific antibodies, other peptides -namely monamine and diamine oxidase enzymes, also exhibit biogenic amine responsiveness.These enzymes are useful as they not only bind to biogenic amines as antibodies do, but rather produce reactive by-products that can be used for detection.These enzymes also have many free carboxyl and amine groups, which mediate their functionalization onto surfaces with limited impact on enzyme activity.Under this premise, Kacar et al. immobilized both enzymes onto indium tin oxide nanoparticles, which were then coated onto electrodes to develop amperometric amine sensors. [102]These electrodes displayed high sensitivity toward biogenic amines -particularly putrescine, histamine, and cadaverine, via Prussian blue-mediated reduction of the H 2 O 2 by-product formed by the oxidase enzyme.Further, the sensor showed effective detection when tested with homogenized fish and cheese samples.Importantly, while the functionality of using amine oxidase enzymes can be called into question because of non-specific reactions with non-target aminobased biological compounds, the study filtered for biogenic amine-triggered reactions by employing a low potential.Similarly, Torre et al. (2019) described the cross-linking of diamine oxidase onto a screen-printed electrode using bovine serum albumin and glutaraldehyde, which showed responsiveness to histamine present within liquid fish extracts. [62]While some interference was seen from other amines -namely phenylethylamine and spermine, the high prevalence of histamine within spoiling fish samples was noted to limit the impact of non-specific signals.Another electrochemical enzyme-based approach was presented by Vanegas et al. (2018), who coated graphene surfaces with copper nanoparticles conjugated with diamine oxidase for biogenic amine detection. [103]They used a liquefied fish paste to show highly selective detection of histamine, with a limit of detection of 11.6 μm.Seeking to the biogenic amine detection capabilities of these enzymes with one another, Boka et al. (2012) compared their efficacy for amperometric electrode sensing. [104]They showed that diamine oxidase can be used to measure total BA content because of its limited specificity, monoamine oxidase could be used for tyramine, tryptamine, and phenylethylamine, and putrescine oxidase could be used to selectively measure putrescine in isolation.Ultimately, the use of these systems continues to be limited by the questionable in situ compatibility of such reaction cascades and the need for complex readout equipment.
A variety of biosensing approaches continue to be explored for the detection of biogenic amines in food as markers of spoilage.Strategies for using biogenic amines to detect spoilage should vary between different types of foods.For some foods, specific detection of certain biogenic amines that are produced only during spoilage is convenient.Multiplexed detection is an advantageous strategy for monitoring spoilage in other foods, where parallel sensing would provide a more comprehensive assessment of spoilage.The synthesis of multiplexed biogenic amine sensors is possible with highly specific and selective biorecognition probes, such as aptamers or MOFs.Given the magnitude of food waste globally, the need for such real-time, food quality sensors is well-established.While many of the discussed platforms are well-situated as on-site detection tools, in situ detection remains the gold standard to allow for monitoring throughout the entire food production and consumption pipeline.Many of the biogenic amine-based sensors lack in situ compatibility in their current form due to issues surrounding system cost and detection specificity, but significant advancements have been made toward this objective, as detailed above.Concurrent evaluation of other systems that offer increased viability for in situ spoilage monitoring has yielded significant focus on pH sensors.

pH Monitoring
As an alternative to tracking highly specific entities, pH monitoring has been used as a more generalized means of real-time food quality assessment. [105]pH changes in food are induced by an accumulation of metabolic products such as lactic acid, and the aforementioned biogenic amines, largely formed through bacterial growth and protein degradation.[107] Importantly, given that bacteria actively contribute to changes in food pH, real-time pH detection can be considered to account for adverse changes in microbial population introduced by poor storage conditions and contamination -a characteristic that increases the utility of such platforms.Concurrently, environment-induced fluctuations in pH can directly promote microbial growth and compromise the integrity of biomolecules within food itself, promoting rancidity. [106,108]Such events make the tracking of pH itself advantageous.
Yet, as opposed to biogenic amines or bacteria, where elevations directly correlate with deteriorating food quality, the relationship between pH and food quality is much more variable.Specifically, food products exhibit diverse baseline pH values, and the direction in which spoilage and contamination shift these values is not universal (Figure 4a).For example, fruit exhibits a decrease in pH over time due to an accumulation of organic acids during the ripening process.Contrarily, meat displays an increase in pH, largely induced by the production of bacterial by-products.This variability means that each food product must be evaluated independently to characterize correlations between spoilage and pH.Until now, studies seeking to report such relationships have largely focused solely on fruits and meat due to their perishable nature and high dietary consumption.
Further, considering its generalized nature, correlating pH values to the point at which a given food product becomes inedible has proven difficult.To substantiate the use of pH-based sensors as real-time food quality indicators, the establishment of well-defined spoilage thresholds is necessary.Nonetheless, the plethora of food-targeting pH sensors reported thus far offer a significant improvement over existing industry-standard, static expiry dates, by allowing users to gain some insight into the general quality of a given product in real-time.These pH sensors have historically involved the use of chemical dyes that undergo color changes as pH fluctuates.While several agents can be used for this purpose, an important consideration in food monitoring is the biocompatibility of the dyes, to ensure safe applicability onto food samples.To this end, non-toxic, organic, dye-based compounds are used. [109,110]

Anthocyanin-Based pH Sensors
Innately present within different types of fruits, flowers, and vegetables, anthocyanins are a class of such non-toxic, dye-based compounds that function as pigmenting agents (Figure 4b). [109,110]These compounds are typically pH responsive, which they transduce via a color change.Most often, they are red at acidic pH levels (<7), purple under pH-neutral conditions, and blue at basic pH levels (>7).This color scheme is not universal however, with many anthocyanins exhibiting notably different signal transduction spectra.Importantly, the pH response signals of anthocyanins are often reversible, enabling sensor reusability.This has been demonstrated within various studies, where anthocyanin-embedded packaging that was stored on spoiled food samples reverted to their initial color state following open air incubation.Concurrently, anthocyanins also boast innate antimicrobial and antioxidant capacities -characteristics that have yielded interest into their value as therapeutic candidates for cardiovascular disease and neurodegeneration. [111,112]ithin the scope of food packaging, these antioxidant and antimicrobial properties make anthocyanins viable candidates for food preservation -a growing area of research, but outside the scope of this review.
115] All of these sources have been found to yield anthocyanins with good pH responsiveness in buffer.The use of anthocyanins and other such organic dyes are also convenient for pH detection as they enable reagentless sensing, and intrinsically change color in response to pH changes.Within more complex environments however, these compounds have exhibited poor stability.Thus, recent efforts in the food monitoring space have centered upon maximizing stable anthocyanin incorporation into packaging films, to enable in situ detection.Specifically, anthocyanins have been successfully incorporated into polymeric films composed of natural polymers such as cellulose, chitosan, and various starches, as well as synthetic polymers such as polyvinyl alcohol (PVA). [115]Synthesis of these films is simple, and involves blending of component starches or polymers, followed by mixing with plant-extracted anthocyanins. [114]Stabilizing agents such as glycerol or other plasticizing agents are also added prior to casting.Film preparation is usually conducted at low humidity and temperature, to limit film hydration and subsequent anthocyanin degradation. [114]While natural polymers have been explored heavily due to their biodegradability and limited toxicity, their use within in situ pH sensing systems is further driven by their chemical properties (Table 2).
Specifically, an important consideration in the development of pH-monitoring films is low susceptibility to fluctuations in pH.Significant changes in pH can compromise the integrity of some materials, impairing their ability to act as functional sen-sors.To this end, organic polymers are relatively resistant to such degradation, thus acting as a stable interface for anthocyanin-pH interactions. [47,116]The exploration of such a diverse range of materials has also been necessitated by the need to limit the rapid degradation of anthocyanins, to induce improvements in pH responsiveness.
Simultaneously, it is important to consider the effect anthocyanin incorporation has on the stability of the polymer matrix.To this end, the swelling index (SI), water solubility (WS), and tensile strength (TS) of developed films are commonly assessed, to consider their viability for in situ use over the duration of a given product's lifespan (Figure 4c). [117]Low SI and WS values are imperative in ensuring that films do not expand exponentially over time within fluid-rich food matrices.Such expansion risks leeching of incorporated anthocyanins, which can result in ambiguous and inconsistent colorimetric signals. [118]Unfortunately, anthocyanin incorporation often induces higher SI and WS values because of their propensity for interactions with water molecules. [119]The existence and magnitude of such interactions depend on the chemical nature of a given anthocyanin. [120]nthocyanins that exhibit defined color changes while exhibiting relatively lower affinity for water are thus optimal.Efforts to further improve system stability are focused on film material optimization.Concurrently, TS provides a measure of the mechanical strength of a given film -a consideration that is vital for realworld applicability, given the manner in which food packaging materials are handled.
Material Optimization of Anthocyanin Films: Most recently developed anthocyanin films have reported polymer blends that yield properties superior to the constituent materials alone.To this end, cellulose-chitosan films have shown significant promise, owing to improvements in the mechanical and chemical characteristics of resultant films.Namely, Alizadeh-Sani et al. created a matrix composed of chitosan nanofibers and methyl cellulose via a solution casting approach that entrapped saffron petal anthocyanins within the polymeric matrix. [121]Compared to a pure cellulose substrate, the study noted a significant reduc-tion in moisture content within the film following the incorporation of chitosan nanofibers and anthocyanins.This was attributed to the replacement of water molecules by these entities within the film matrix.While a promising development, these hybrid films offered limited anthocyanin carrying and retention capacity due to the larger size of the constituent polymers.Concerns surrounding anthocyanin retention are further amplified by the system's reliance on physical entrapment for anthocyanin immobilization.
Comparatively, PVA can form direct interactions with embedded anthocyanins.Such interactions stabilize their immobilization within resultant films and limit interactions with water molecules in the environment, which can accelerate their degradation.Under this premise, Kuswandi et al. created red cabbage anthocyanin-embedded bacterial cellulose-PVA films, via a simple protocol that involved the immersion of cellulose films in a 1% PVA-anthocyanin solution. [122]As expected, anthocyanin incorporation alone yielded an increase in the SI and WS, but this was counteracted by PVA, resulting in properties comparable to the cellulose base film.Unfortunately, PVA incorporation did not yield any improvements in TS and resultantly failed to counteract the plasticizing effects of anthocyanins.As such, a reduction in TS was observed from 65.2 to 39.5 MPa.These sensors were used to detect pH decreases in milk colorimetrically, displaying a transition from blue-gray to pink as the milk spoiled.The authors argued that the improved anthocyanin stability afforded by these sensors qualified their use for in situ spoilage monitoring, as the films could be embedded onto milk and juice bottle caps.Such a set-up would limit mechanical stress on the film, limiting concerns pertaining to the decreased TS.
Operating under a similar premise, Vo et al. successfully used anthocyanin-embedded, PVA-infused chitosan films for the pH monitoring of pork samples. [123]Sodium tripolyphosphate (TPP) was added into the polymer matrix to improve film strength via the induction of increased intermolecular interactions.While the optimized 35/65 PVA to chitosan blend offered some improvements in performance, the TS of the developed film remained suboptimal.Ultimately, further material interventions are needed to overcome the poor mechanical strength of such films.
Compared to chitosan, starch films offer significantly worse mechanical properties.Nonetheless, studies seeking to improve their material properties for food packaging applications are widespread. [124]These efforts are largely driven by the inexpensive nature of the material.Here, PVA has also been used, namely by Zhang et al. ( 2020), who detailed the production of anthocyanin-embedded packaging films composed of cornstarch and PVA. [125]While the resultant film exhibited strong pH monitoring performance when applied to shrimp samples, its TS was only 7.3 MPa.Anthocyanins were seen to slightly increase the strength of these films -likely due to the replacement of water within the film matrix, but only to a maximum of 11.3 MPa.The films also exhibited very poor stability, with a WS of 40.3%.As such, widespread use of starch films for food packaging requires changes in film composition outside of PVA to induce significant improvements in strength and stability.
Ultimately, film material optimization has been shown to increase the overall stability and functionality of in situ pH sensors substantially.However, resultant sensors are still limited by the sub-optimal film properties and poor anthocyanin stability, specifically over extended periods of time and in varying environmental conditions.This has substantiated the need for secondary film modifications that improve film and/or anthocyanin stability, yielding more reliable in situ detection platforms.
Chemical Modifiers: While the induction of denser film matrices is the most common strategy for the development of more stable films -as presented in many of the aforementioned studies, modifications that induce improved chemical compatibility within the film matrix have also been explored.Namely, Eze et al. incorporated riceberry phenolic extract (RPE) into anthocyanincontaining chitosan films as a means of increasing hydrophilic interactions (Figure 5a.i) and thus promoting hydrogen bonding with chitosan. [126]Increased cross-linking was seen within these modified chitosan films, effectively decreasing WS and SI.In addition, improved preservation of the stored food was also observed, due to the reduced porosity and permeability of the developed film, which limited the introduction of moisture, oxygen, and other external gases into the packaging.This phenomenon was supported by the innate antioxidant activity of RPE.
Seeking a similar chemically-induced improvement in physical stability, Wen et al. stabilized 2,2,6,6-tetramethylpiperidine-1oxyl (TEMPO)-oxidized bacterial cellulose films with thymol, to increase the resultant film's TS and hydrophobicity -the latter of which also minimized sensor surface fouling. [127]Subsequent functionalization with anthocyanins from purple potato extract showed pH responsiveness as expected.TEMPO oxidation introduced anionic carboxylate crosslinkers (Figure 5a.ii) into the cellulose films, enhancing their associations with embedded anthocyanins, thus improving physical stability.These films were functional at pH sensing for shrimp in situ when applied as a packaging material over stored shrimp samples at room temperature for 6 hours and extended shelf-life preservation, which was attributed to the film surfaces' resistance to microbial colonization and water vapor permeability.
Such chemical approaches have also been applied to improve anthocyanin stability in response to environmental stimuli.Heat degradation is of concern given that various food products undergo thermal processing post-packaging.Such treatment causes the polymerization of anthocyanins, which impairs their ability to respond to pH and causes dye browning.The latter may be interpreted as false positives within colorimetric pH monitoring.Qin et al. reported that the heat instability of anthocyanins can be reduced via coupling with Maillard reaction products of whey protein isolates (WPI). [128]Maillard reactions change the glycation state of WPIs, allowing them to readily conjugate with other compounds.When such WPIs were coupled with anthocyanins, reduced anthocyanin degradation and better color stability were observed.Similarly, Wu et al. ( 2021) developed gellan gum packaging films embedded with anthocyanins stabilized by heat-treated soy-protein isolates.The films demonstrated enhanced stability and lowered water vapor permeability, owing to the increased intermolecular interactions.Evaluation of the film with shrimp samples confirmed strong in situ pH monitoring capabilities.
Ultimately, chemical modifiers are a promising avenue for film optimization given the diverse range of improvements they provide.That being said, they only offer limited improvements in terms of anthocyanin stability, another key factor for in situ use.
Anthocyanin Nanoencapsulation: Efforts to protect anthocyanins from environmental disturbances have been explored by non-chemical means as well, using approaches that both induce improved film stability and limit the leaching of anthocyanins from carrier films.To this end, the nanoencapsulation of anthocyanins has garnered significant interest, largely through the use of protein-polysaccharide complexes (Figure 5a.iii). [129]These complexes prevent the heat-based degradation of anthocyanins, enhancing their potential use in long-term pH monitoring.Many variations of these food-grade complexes have been evaluated for their ability to act as protective nanocarriers for anthocyanins, while maintaining their ability to interact with the greater food matrix. [130]One such strategy involved the incorporation of anthocyanins into ovalbumin-propylene glycol alginate nanocomplexes, which were deposited onto PVA and glycerol matrices to make intelligent films. [131]This approach increased the stability of embedded anthocyanins, resulting in improved performance against pH buffers.That being said, the study noted that alginate nanocomplex incorporation increased water vapor permeability, yielding an increase in the film's WS, from 12.26% to 29.34%a shift that compromises the long-term stability of the film.
To better understand the effect of anthocyanin nanoencapsulation on film stability, Qin et al. tested anthocyanins incorporated into sago starch films stabilized with PVA, both in a free and nanoencapsulated form. [132]Both forms were immobilized into the starch matrix via hydrogen bonding.While both films maintained the pH-responsiveness of anthocyanins and demonstrated effective real-time monitoring of fish, nanoencapsulation was found to improve the performance of the resultant film across several metrics.Specifically, decreases in water vapor permeability and moisture content, as well as improvements in TS were observed.Given this finding, modifications to the polysaccharides used to develop nanoen-capsulating films have also been explored to improve system performance.In a recent work, Cheng et al. (2022) developed a "dual-modifying" cassava starch films incorporated with red cabbage anthocyanin. [133]These "dual modifications" refer to multiple chemical modifications that increase electrostatic interactions between anthocyanins and starches, inducing improved immobilization.Several modifications were tested, including the addition of oxidized hydroxypropyl starch, acetylated di-starch phosphate, and oxidized acetylated starch, which all increased the anionic strength of the starch substrate, resulting in increased electrostatic interactions.All these modifications increased film stability, water resistance, and tensile strength.Modified cassava starch films were also thicker than native, starch-only films because of the increase in intermolecular interactions, which also led to increased density and rigidity.However, the potency of the colorimetric responses was not changed by any modification, as native and modified films had similar biosensing capacity.
Micro-and Nanoscale Modifiers: Lastly, micro-and nanoscale modifiers have been explored extensively. [134,135]While most of these modifications involve particles, Koshy et al. recently reported the use of carbon nanodots (CNs) as a packaging film modifier (Figure 5a.iv). [136]CNs are organic, biocompatible entities that are often used as fillers within pH monitoring films, inducing improvements in film stability.In their study, they embedded CNs alongside anthocyanins into starch films for realtime food monitoring.Despite the incorporation of CNs, the resultant film had a relatively low TS of ≈10 MPa.Similarly, minimal improvements in stability were observed, with CN incorporation yielding a WS of 27.2%.While these films showed good detection of pH changes occurring in pork during spoilage, the subpar mechanical properties of starch films were not overcome through the incorporation of CNs.The incorporation of CNs into anthocyanin-embedded chitosan and cellulose packaging films is yet to be explored.It is possible that CN incorporation into these film matrices may yield more substantial improvements in film properties.
On the other hand, micro-and nanoparticles have been explored heavily as both anthocyanin carriers and as filler materials that induce improvements in film stability.With regards to anthocyanin carrier particles, chitosan-cellulose microparticles have been explored for their anthocyanin loading and retention capacity.Specifically, Wang et al. contrasted cellulose as an anionic cross-linking agent with sodium tripolyphosphate (TPP) when coupled with chitosan in the form of microparticles. [137]They found that the chitosan-cellulose particles exhibited high loading efficiency, as indicated by evenly distributed anthocyanins throughout the particles, while the chitosan-TPP particles only displayed anthocyanin on their surfaces.The improvement in distribution induced by cross-linking with cellulose increased retention of anthocyanins over time.This stability was facilitated by interactions between negatively charged cellulose and positively charged chitosan.While effective, these interactions simply entrap anthocyanins.The lack of anthocyanin immobilization raises concerns surrounding long term stability.
As an alternative, chitosan-based anthocyanin nanoparticles have also been applied as anthocyanin carriers. [138]Alongside high retention capacity through surface immobilization of anthocyanins, this system demonstrated high thermal protection of the dyes by minimizing their hydration from the surrounding environment.Protection from pH fluctuations and light was also reported.With regards to the latter, the study found a 20% loss in anthocyanin functionality after 10 days when incorporated onto the chitosan nanoparticles, as opposed to an 86% loss with free anthocyanins stored in the same conditions.Further work is required to investigate whether the induction of such effec-tive anthocyanin protection compromises the pH-sensing capabilities of the anthocyanins.Similar approaches investigated thus far have signaled minimal limitations.
The use of particles as filler materials largely centers upon inducing a denser film matrix, yielding improved stability.In a recent study, Zhang et al. used TiO 2 nanoparticles as a filler material within anthocyanin-embedded chitosan films for food-based applications. [139]The incorporation of TiO 2 alone did not significantly alter the WS of the chitosan film.However, chitosan films incorporated with both TiO 2 nanoparticles and black plum anthocyanins showed reduced WS in comparison to either agent alone.This improvement was attributed to interactions between the nanoparticles and hydrophilic anthocyanins, which prevented hydrophilic interactions with water molecules.

Other Compounds Used for pH Monitoring
Aside from anthocyanins, a variety of pH-responsive organic compounds have also been explored for integration into intelligent food packaging, although at a lower frequency.Current approaches have largely moved away from the use of chemical pH indicators, such as methyl red and bromocresol blue, because of their toxicity.That being said, Kim et al. overcame this issue by incorporating bromocresol blue into highly absorbent materials that reduced the likelihood of leaching. [140]The interface between the compound and food samples, however, was maintained to allow for effective sensing.In their approach, the indicator was embedded into filter papers and stabilized with PVA.Polyethylene terephthalate film was used as a physical barrier to limit compound leeching into the bulk matrix.While it is difficult to assess the extent of leeching of the indicator into the meat and any subsequent cytotoxicity that results, the study reported no visual migration of the purple dye onto a chicken matrix after 10 days of storage.Ultimately however, given the increasing prevalence of organic, non-toxic pH-responsive compounds, this approach has limited applications for in situ detection for food, where biocompatibility is a paramount concern.The search for non-toxic alternatives has yielded several viable compounds.
In particular, compounds native to plants have shown promise -namely, curcumin, alizarin, and betanin (Figure 5b).Curcumin is well-documented for its pH-responsive ability, and curcumin nanoparticles have been explored outside the food space as well, namely in improving signal readout from pHresponsive ELISAs. [141,142]Curcumin has been used heavily in the development of intelligent packaging films, as both a pHresponsive compound and a film stabilizer. [124]Typically found in the turmeric plant, curcumin gains a phenoxide group upon interaction with increasing concentrations of hydroxyl ions -as seen within meat spoilage, yielding a color change.Moreover, curcumin functions as a film stabilizer when complexed with other pH-responsive compounds, because of its hydrophilic nature.This function was reported in a recent study, that showed that a combination of curcumin and anthocyanins yielded a packaging film with improved mechanical stability than either agent in isolation. [143]However, the mechanism by which this combination of agents yielded improved performance was not reported.
Compared to curcumin, alizarin offers an additional benefit in that it can be used as a reusable pH indicator.This is due to the reversibility of the compound's protonation statethe parameter that influences the color of the compound.Ezati et al. incorporated alizarin dye into cellulose-chitosan films. [144]iven alizarin's hydrophobic nature, its incorporation yielded a decrease in WS from 16.13% to 12.89%.The SI of the developed film remained unchanged relative to a base cellulose-chitosan film.Interestingly, a decrease in film stability was noted when stored at room temperature.From a functionality standpoint, these films exhibited responsiveness from pH 2 to 11 -a range that ensures widespread applicability with different food products.Tests with beef samples stored in the fridge confirmed effective pH detection in accordance with sample spoilage.In related works, the same research group showed that alizarin can also be incorporated into cellulose and chitosan films, offering similarly promising results. [144]eeking to evaluate how such alternative compounds compare with anthocyanins, Etxabide et al. compared anthocyanins, curcumin, and betanin -another natural pigmenting agent, based on their pH-responsiveness and stability under both light and heat exposure. [145]The study found that in isolation, curcumin pigment was the strongest of the three as a colorant but lost its color intensity when under light exposure for prolonged periods of time.Opposingly, anthocyanins and betanin retained adequate intensity for up to a month.The long-term stability of these colorimetric agents is crucial for in situ detection to ensure viability throughout a food product's lifespan.With consideration toward the high background turbidity of food products, this brings the reliability of curcumin into question.Anthocyanins and betanin did exhibit less thermostability under warmer conditions, but this was considered to be a low priority consideration given that various material strategies have been shown to induce improved thermal stability, as detailed previously.Following extensive evaluations, Etxabide et al. concluded that anthocyanins remain the most viable class of agents for in situ pH-responsive intelligent packaging, owing to their dramatic color changes, wide pH detection range, and amendable stability. [145]However, while a largely comprehensive study, this work only analyzed the viability of tested compounds in solution, not in a film embedded format.To this end, Tirtashi et al. showed that immobilization and entrapment of these compounds influences their pH responsive properties, both in accuracy and magnitude of color change. [146]iven the diverse range of materials being used to prepare in situ films and the wide range of compounds being explored for their pH-responsive behavior generalized claims surrounding the viability of a given entity have thus proven difficult to support.
Ultimately, existing in situ pH sensors offer significant promise as spoilage monitoring systems, but their use continues to be limited by poor dye and film stability, preventing their widespread commercial use.As most pH sensors are designed by incorporating reactive plant-based dyes into simple organic polymers, they are cheap and cost-effective for spoilage detection, and are also convenient for commercial usage as they lead to obvious visual changes without requiring complicated signal readout methods. [114]As they are relatively simple in design, they are also easy to mass produce at a relatively low cost, and to store in liquid-free environments to prevent dye release and degradation.Nanoencapsulation of dyes further increases the long-term stability of these films during storage.Alongside this, there also exists the potential to multiplex different types of anthocyanins with similar color-responsiveness in pH labels, to enhance sensor performance.However, the relationship between pH and spoilage lacks the objective consistency that is offered by biogenic amine or pathogen monitoring, which raises some concerns regarding the viability of such systems.Intrinsic differences in the degree of pH fluctuation between foods during spoilage, as well as other compounds produced during shelf storage, can also make pH unreliable as a metric for spoilage.Resultingly, there continues to be a defined need for studies that evaluate this relationship comprehensively on a large sample scale.Lastly, given that food spoilage is typically accompanied by discernible color and odour changes, it is imperative that such sensors report spoilage at a timepoint where consumption is harmful, but not apparent.Evaluation of how well existing pH monitoring films identify this timepoint is necessary.Nonetheless, pH monitoring of food continues to be an active area of research with frequent developments that yield confidence toward commercial applicability in the future.

Pathogen Monitoring
The contamination of food products by pathogenic bacteria usually arises from improper practices during livestock handling, food processing, and product packaging. [16,147]Timely identification of food contamination is paramount in minimizing the incidence of food-borne illness.However, contamination detection typically requires multiple days, as it involves food processing, culturing, and genomic analysis to evaluate the presence of pathogenic bacterial strains.All these steps cannot be done in situ and require export outside of processing facilities and commercial grocery stores -an expensive and time-consuming process.Moreover, some bacterial strains can enter a viable but nonculturable (VBNC) state that prevents accurate identification and quantification via typical microbiological culture methods. [148]As such, there is a growing demand for in situ biosensors for bacterial detection that enable the on-package detection of pathogenic contamination within a few hours of sample incubation, with minimal user processing.This section discusses recent advancements in this space.
Functional biorecognition probes are the predominant bacterial detection platform, due to their high sensitivity and specificity. [149]Various types of probes can be used to detect bacterial contamination in food by harnessing their natural binding affinity for bacterial compounds.In comparison to the detection of fluxes in biogenic amine concentration and pH, pathogen detection demands significantly higher specificity, to ensure that non-pathogenic microbial populations that are naturally present in food products do not induce false positive signals.Candidates for the development of biorecognition probe-based sensors for in situ pathogen detection are thus chosen primarily due to their high target binding specificity.Promising probes include bacteriophages, functional oligonucleotides, and antibodies (Figure 6a).These probes are incorporated into detection platforms that signal pathogenic contamination through diverse transduction methods.

Bacteriophage-Based Probes
Bacteriophages, commonly referred to as phages, are bacteriatargeting viruses with intrinsic specificity to their host species,  which makes them reliable biorecognition probes. [150]Upon binding to the host bacterium, lytic phages replicate exponentially in host cells until they induce lysis, which destroys the bacterium and allows for further phage propagation within nearby host cells. [151]Because of the natural predatory relationship that exists between bacteriophages and bacteria, these viruses act as a promising type of probe for application in food biosensing -an application that is further supported by their biocompatible nature.With consideration toward real-world applicability, bacteriophage production is easily scalable, and the viruses demonstrate high stability under diverse environmental and chemical conditions, as discussed by Ahovan et al. in a recent review. [152] predominant approach for developing bacteriophage-based biosensors involves phage incorporation into electrochemicalbased platforms that exhibit detectable changes in electrical current upon binding target proteins.For example, Quiton et al. developed an impedimetric electrochemical biosensor that exhibited changes in impedance following phage-mediated binding of Salmonella cells onto graphene-modified screen-printed electrodes (Figure 6b.i). [153]Phage immobilization is accomplished by pre-activating carboxyl groups on screen-printed electrodes prior to incubation with bacteriophage.Interestingly, this leads to sufficient immobilization of the bacteriophage onto the electrodes in the right conformation, as this sensor demonstrated a limit of detection of 12 CFU mL −1 following a 40-minute incubation with test samples.In such a sensor, impedance of electron transfer takes place upon phage-bacteria binding, leading to a change in signal output.The system showed specificity toward S. typhimurium -the most common source of foodborne Salmonella infection, over other non-target Salmonella strains.Mechanistically, this was due to phage specificity for strains of Salmonella, as there were no changes in impedance upon incubation with other common pathogens.Yet despite the high sensitivity and specificity of the developed platform, the widespread applicability of the system was limited by its reliance on electrochemical impedance spectroscopy, a complex lab-based technique performed by trained experts.Similarly, Wang et al. ( 2022) also explored Salmonella detection using AuNPs to enhance the sensitivity of an electrochemical bacteriophage-based sensors. [154]uNPs were absorbed onto gold disk electrodes to increase the sensor's specific surface area and conductivity (Figure 7a).Subsequently, these nanoparticles were crosslinked with SEP37 Salmonella bacteriophages using cysteamine.This probe showed a limit of detection of as low as 1 CFU mL −1 within contaminated chicken.However, higher bacterial concentrations that increased bacterial lysis rates impeded sensor performance due to the release of intracellular ions from cells, which alters the electron transfer potential of the system.
In another study, Zolti et al. described the immobilization of a P100 bacteriophage onto carbon nanotubes with high sensitivity for Listeria monocytogenes, detecting as low as 8.4 CFU mL −1 of bacteria. [155]The bacteriophages were first functionalized with quaternized polyethylenimine which then enabled chargedirected immobilization onto the carbon nanotubes, increasing interaction and infection with target bacteria.This approach is used to immobilize bacteriophages as polyethylenimine is positively charged at neutral pH, enabling electrostatic interactions with the bacteriophage head. [156]They reported a similar detection strategy for E. coli and other pathogenic bacteria as well.Interestingly, these sensors efficiently retain phage and allow them to retain their performance for up to two weeks.Similar to other studies, they show that the high specificity of bacteriophage strains for species-specific bacterial receptors leads to low cross-reactivity with other bacteria.Other surface modifications have also been coupled with phage to help enhance the sensitivity of electrochemical biosensors.
Seeking to develop a platform with more applicability in the food industry, Sultan et al. incorporated bacteriophages onto a millimetre-wave based antennae sensor, which was used to detect E. coli within produce wash water (Figure 6b.i). [157]Rather than using phages for detection, they instead use antennae to detect changes in resonant frequency that result with increases in phage replication, which take place in the presence of bacteria.This strategy also allows for very sensitive detection at low levels of contaminant pathogen, with phage fluctuation detectable within 40 min from sample incubation.However, detection would work significantly better with higher bacterial levels, which lead to more profound production of phages.The use of wash water instead of food for the substrate to detect changes in resonant frequency reduces the presence of competitive signals that could lead to false positives.While antenna-based detection is extremely sensitive to changes in resonant frequency, direct application of bacteriophages to food is a non-viable approach for reagentless sensing and for prolonged monitoring.An ideal approach would incorporate these bacteriophage into food packaging to enable monitoring without interfering with packaged food products.
Overarchingly, electrochemical-based bacteriophage systems offer excellent sensitivity and specificity, ensuring rapid, accurate detection.Efforts are now being made to optimize such platforms, as the stability of bacteriophages on electrochemical surfaces remains questionable.Yet, while valuable tools, sensor measurements largely require the use of complex apparatuses that are not readily adoptable to in situ environments, limiting the applicability of such platforms to processing facilities.Consumer-level monitoring requires an in situ platform with signal readouts that are easily acquired and interpreted.
To this effect, fluorescent and luminescent transduction systems are being explored to address some of the drawbacks within conventional electrochemical bacteriophage sensors.Namely, Wisuthiphaet et al. paired a genetically engineered T7 phage that overexpresses alkaline phosphatase (ALP) following successful E. coli infection. [158]A non-fluorescent ALP substrate was then introduced into test solutions.Cleavage of this substrate by ALP yielded a fluorescent precipitate that was tracked optically.This platform was shown to successfully detect 10 4 CFU mL −1 in various beverages.Ultimately, the use of such homogenous test matrices raises concerns regarding this system's viability for complex food samples, especially given its limited sensitivity.
Similarly, Nguyen et al. engineered Salmonella-responsive bacteriophages with NanoLuc, a luciferase enzyme, that produced luminescent signals following successful infection of target cells. [159]This platform was able to detect as low as 1 CFU Salmonella in 25 g of inoculated ground turkey and 100 g of powdered infant formula after as low as 2 h, owing to the high op-tical intensity of NanoLuc.However, it should be noted that the turkey and infant formula samples were enriched for 7 and 16 h, respectively, and the samples were subsequently homogenized prior to testing.Enrichment and homogenization of samples limits the in situ application of this sensor, but the use of engineered bacteriophage luminescence represent an interesting avenue that could be explored within the in situ space.Overarchingly, such synthetic bacteriophage-based sensors suggest easier in situ incorporation compared to pure electrochemical platforms.One current limitation to their usage is that they require the addition of external reagents to produce signals as these phages are usually engineered with enzymes that require substrates to generate signals.Engineering phages with genetic circuits that use substrates found in food, or ones produced as an effect of bacterial contamination itself, would increase the convenience of using these sensors.Biosensors designed around the premise of phage replication could also use phages engineered to express reporter molecules themselves, as replication in the presence of bacteria would lead to an increased positive signal.Such genetic circuits are also advantageous because they can be used to facilitate multiplexed detection of contaminants within samples.By engineering different bacteriophage strains with unique reporter molecules, contamination by a number of prevalent pathogens can be analyzed.
While these studies have shown that bacteriophages act as prime candidates for the detection of pathogenic bacteria, few studies have manipulated them into functional, intelligent materials that can be applied to food.Bacteriophages are fairly easy to propagate, making them easy to mass produce. [160,161]owever, this production process also has significant risks for biosafety, as continuous propagation of phages through the bacterial strains they infect favors the evolution of resistant strains of these bacteria. [161]Beyond production, the storage of these phages post-production is also challenging.Unlike other probes, bacteriophages are composed of complex assemblies of protein and genetic material, making them more difficult to store longterm. [160,161]The recent development of new approaches to engineering bacteriophages into functional nanomaterials themselves warrants further exploration into how these probes can be deployed to create intelligent packaging for food monitoring. [162]

Oligonucleotide Probes
DNA and other nucleic acid-based nanotechnologies have been increasingly explored for their biomedical applications, from clinical therapeutics to biosensing.The characteristic advantage of using such probes lies in the intrinsic programmability of nucleic acids, which can be used to synthesize oligonucleotides that display high binding affinity and specificity toward a diverse range of ligands.There are two main types of oligonucleotide probes: 1) aptamers, which are single-stranded oligonucleotides that fold into secondary structures with high binding affinity to target biomolecules, and 2) DNAzymes, which are oligonucleotides that exhibit catalytic activity upon binding to a target. [94,163]These are both synthesized via systematic evolution of ligands by exponential enrichment -a methodical approach involving the selection of responsive nucleic acid probes from large DNA libraries that are exposed to targets. [94]Both aptamers and DNAzymes can be functionalized via their DNA backbones using a variety of different prosthetic groups that enable their use in fluorescent, colorimetric, and electrochemical-based detection systems. [164]These nucleic acid probes have been explored in the development of pathogen-detecting biosensors, by coupling their binding activity with a signal reporting cascade.These probes have then been immobilized onto various sensing substrates.Evaluation of existing probes is well-documented in literature: Li et al. ( 2019) reviewed electrochemical aptamer-based pathogen detection methods and Khan et al. ( 2021) reviewed the development of DNAzyme-based biosensors for widespread applications, with a defined focus on surface immobilization. [95,163]ptamer-based pathogen detection is not as common as DNAzyme-based methods given that a greater number of pathogen-responsive DNAzymes have been reported thus far.For the aptamer-based sensors that do exist, a key consideration is signal transduction.While DNAzymes have enzymatic, reactive activity that enables the activation of a reporting entity following target-induced cleavage, DNA and RNA aptamers only have binding affinity.As such, aptamer-based sensors must co-opt structural changes to the aptamer upon target binding, with colorimetric, fluorescent, or electrochemical additions to the DNA backbone. [99]Once coupled with signaling entities, aptamers are powerful detection tools.
For many years, antibodies have been used in the biosensing space for the development of lateral flow assays such as the pregnancy test, which incorporates the specificity of antibody binding with a colorimetric signal provided by AuNPs.Given their high binding affinity, aptamers can also be used to develop such nanoparticle-based assays for pathogenic bacteria (Figure 6b.ii).Tasbasi et al. described the development of a lateral flow assay to identify Listeria monocytogenes. [165]They synthesized silica nanoparticles entrapped with TMB, a substrate for HRP, and capped them with Listeria-specific aptamers.When these sensors are exposed to as low as 10 2 cells, the aptamer is released, leading to a color change in the sensor upon the addition of HRP.Reagent entrapment is efficient as it limits the degree of external tampering required, as such reactions typically require sequential addition of HRP and TMB.Typically, colorimetric systems that incorporate TMB in active food packaging are limited by the high reactivity of redox reagents.The aptamer-gated silica particles developed in this work provide an innovative incorporation strategy.Future investigation into the efficacy of this system when coupled with food is necessary for in situ detection.
A study by Hasan et al. described the development of a Salmonella-specific DNA aptamer used for electrochemical pathogen detection in chicken, with a limit of detection of 55 CFU mL −1 and high specificity. [55]They modified their aptamer with an amine group, which permitted immobilization onto carbon nanotube surfaces that were enriched with carboxylic acid groups.An impedimetric electrochemical readout was obtained upon binding of bacterial proteins in samples of homogenized chicken.However, given that this device requires electrochemical impedance spectroscopy to measure aptamertarget binding, this system is faced with many of the same commercial challenges previously described for electrochemical platforms.Accordingly, aptamers with commercially translatable fluorescent mechanisms have been developed for pathogen detection.Namely, Tian et al. ( 2021) immobilized Alicyclobacillus acidoterrestris-specific aptamers onto magnetic nanoparticles to isolate target bacteria within juice samples, and subsequently used A. accidoterrestris-binding antibodies labeled with fluorescent quantum dots for fluorescence visualization. [166]The study reported a limit of detection of 10 3 CFU mL −1 within 90 min, with high specificity.While the optimization of such a system for in situ use is yet to be reported, the incorporation of similarly designed antibody-nanoparticles into nanocomposite packaging films may offer a platform of notable value.
Concurrently, DNAzymes have been applied within a wide array of pathogen-detecting systems. [167]Their recognition and reporter molecules, these oligonucleotides offer simple reaction cascades in which they are the only application in the food industry has only recently been explored and is an intriguing application of this technology.Given that DNAzymes can be designed with a built-in reporter, allowing them to offer dual functionality as both recognition and reporter molecules, these oligonucleotides offer simple reaction cascades in which they are the only functional entity. [163]This standalone nature has inspired direct incorporation onto a wide array of substrates, to develop functional packaging materials.To this end, a key consideration is the technique by which DNAzymes are deposited onto the sensor surface, as proper DNAzyme deposition is critical for sensor stability and subsequent in situ use.
One simple technique involves depositing DNAzyme solutions onto the target substrate and drying to induce noncovalent attachment.Qin et al. (2021) utilized this method to develop a fluorescent DNAzyme sensor responsive to Pseudomonas aeruginosa, PAE-1. [168]They dried the DNAzymes onto polystyrene boards to create stable, non-covalently immobilized sensors.This system demonstrated high selectivity for P. aeruginosa based on cross-testing against seven other bacterial species.Furthermore, it boasted a limit of detection of 10 2 CFU mL −1 after 10 min of incubation with contaminated black tea.Concurrently, the viability of this deposition method was also assessed by Ali et al. in a platform designed for in situ use. [169]This study showed that DNAzymes can be incorporated onto paperbased sensors alongside stabilizing agents to create sensors for in situ E. coli detection.Specifically, they deposited solutions of an E. coli-fluorogenic DNAzyme dried onto wax-coated nitrocellulose paper stabilized with trehalose and pullulan sugars (Figure 7b).Lysozymes were also printed onto the paper sensors in parallel to lyse target bacteria.This increased target availability for DNAzyme binding, thus triggering faster detection.The resultant platform demonstrated a limit of detection of 10 2 CFU mL −1 after 30 min.Successful detection was also observed within diluted apple juice and milk samples. [169]Lastly, Ma et al. (2021) developed a fluorescent Aeromonas hydrophilaresponsive DNAzyme, which they called DAh1, immobilizing it onto polystyrene using trehalose/pullulan stabilizers and drying to deposit the probes onto the surface. [170]DAh1 sensors remained stable for 6 months in comparison to freshly prepared sensors and were used to obtain a limit of detection of 36 CFU mL −1 in milk.Ultimately though, while DNAzyme deposition via drying is fast and requires no additional reagents, the lack of covalent immobilization threatens the integrity of the sensor when applied to complex, solid food matrices.This concern is not addressed within the presented studies, given that the reported sensors were assessed solely in liquid test matrices.Considering the harsh handling and environmental conditions food products are exposed to, other DNAzyme deposition techniques that offer better stability and detection in complex food matrices have gained traction for in situ pathogen detection.
Comparatively, covalent DNAzyme immobilization using chemically modified DNAzymes and/or substrate surfaces yields more robust and durable sensors that are better fitted for in situ applications compared to simple drying.In a foundational study, Yousefi et al. (2018) used epoxy-mediated immobilization to develop an in situ-compatible E. coli-responsive DNAzyme patch with a limit of detection of 10 3 CFU mL −1 in buffer within 4 h (Figure 6b.iii). [16]Specifically, 5' amine-modified DNAzymes were deposited onto epoxy-coated cyclo-olefin polymer substrates.Importantly, these flexible substrates were also incubated with unprocessed, contaminated solid food samples for 2 h, and demonstrated successful detection of 1 CFU mg −1 E. coli on contaminated lettuce and beef samples.[176] An increase in DNAzyme sensor sensitivity was observed in milk using this approach, with successful detection of 10 2 CFU mL −1 E. coli within an hour.This improvement was attributed to lubricant-mediated redirection of DNAzyme-binding entities away from the sensor substrate, and toward the oligonucleotides, yielding an increase in signalinducing binding events.This indicated the scope for DNAzymebased sensors to have anti-biofouling, repulsion, and signal amplification features incorporated within an in situ device.
A few concerns still exist in relation to the incorporation of DNAzymes in situ, much of which centers upon their dependency on specific metal ions for catalytic functionality.This can become a hindrance to DNAzyme-based sensor designs as it becomes important to incorporate these metal ions as aqueous solutions, which is a challenge when working with physical food samples.The platform used by Yousefi et al. addressed this obstacle by pre-treating food samples with a non-toxic magnesium ion-containing reaction buffer solution. [16]While effective, such treatments can influence the organoleptic properties of given food products.As such, better means of buffer incorporation remain a need in this space.A related concern was well-established by Qin et al. during the development of their PAE-1 sensor, where they noted that DNAzyme performance peaks at certain metal ion concentrations. [168]As such, while some highly effective DNAzymes exhibit excellent sensitivity and specificity, requirements for high concentrations of metals ions for functionality limits their viability in the food sector.Increasing the presence of these ions past food-safe thresholds introduces regulatory concerns pertaining to toxicity.The metal ion requirement of a given DNAzyme thus must be considered during probe selection.Lastly, pH can also be a restrictive factor for DNAzyme activity, as it can cause DNA denaturation or trigger metal ion precipitation.However, while certain foods like meat do increase in pH during spoilage, these increases are usually insignificant toward DNAzyme functionality.
Yet, while DNAzymes effectively detect and report pathogens, they often lack the high target binding affinity of aptamers -a property than can yield higher sensitivity and specificity.As such, some works have coupled these two probes to harness both the binding activity of aptamers and the catalytic activity of DNAzymes.For example, Liu et al. constructed a coupled DNAzyme-aptasensor detection system for the detection of Cronobacter sakazakii in liquid solution. [177]This sensor used a split-DNAzyme-aptamer that assembled in the absence of C. sakazakii to produce a DNAzyme-mediated colorimetric signal.In the presence of C. sakazakii, the aptamer's affinity for the bacteria prevented DNAzyme assembly, resulting in no development of color.This study reported a limit of detection of 1.2 CFU mL −1 .Since most C. sakazakii-related deaths result from infant formula, this system holds value despite its lack of viability as an in situ system, as it can be used to evaluate liquid test matrices with ease.Nonetheless, the split-DNAzyme-aptamer system is evidently promising for the detection of other pathogens in solid foods, making in situ-centric optimization a valuable area of future work.That being said, turn-off platforms yield an increased risk of false positives induced by colorimetric systems experiencing substrate depletion, and fluorescent probes experiencing photobleaching, which is inevitable during long-term monitoring.Transitioning toward a turn-on platform is thus a potential next step.
Overall, the customizability, sensitivity, and practicality that DNA-based systems offer makes them strong platforms for in situ food contamination detection.With growing efforts to improve sensor stability for long-term use, alongside the discovery of probes with higher sensitivity and specificity, the value of such systems can be expected to increase.

Antibody-Based Probes
Analogous to DNA probes, antibodies are advantageous because they can be generated with high specificity and selectivity for certain target bacteria.As peptides, their abundance in free reactive amino and carboxyl groups can be used for covalent linkage to a variety of sensing platforms.This represents a key advantage over DNA-based probes, which are artificially functionalized with limited reactive groups.Another advantage of antibodies is that they do not typically require external reagents to maintain their secondary structure.However, their widespread use is limited by several key disadvantages.Namely, antibodies are complex proteins that offer less environmental stability, showing denaturation in response to harsh changes in pH and temperature.Antibodies are also less refined compared to DNA-based probes, the latter of which undergo multiple rounds of in vitro selection to ensure optimal binding efficacy.Lastly, antibodies are more expensive to synthesize than DNA-based probes, which decreases their commercial viability.Despite these drawbacks, some antibodybased sensors have been developed using colorimetric and electrochemical transduction methods.
In a recent work, Zeng et al. (2022) demonstrated that antibodies specific to Yersinia enterocolitica, a pathogen commonly found in raw pork, can be used to develop a paper-based colorimetric sensor to directly monitor contamination in meat samples. [178]his method involved linkage of antibodies with colloidal gold nanoparticles to develop a lateral-flow based assay on paper strips for contaminant detection (Figure 6b.iv).A secondary detection antibody was subsequently introduced to produce a colorimetric signal, yielding a limit of detection in the range of 10 2 -10 3 CFU mL −1 .The system successfully detected the target bacteria in both milk and pork samples down to a limit of detection of 5 CFU mL −1 , but only following sample enrichment with growth media and after incubations of 4 to 6 h.Assessment of contaminated samples without growth media was not presented, limiting proper understanding of sensor functioning at real-world levels of contamination.
Recognizing that target pathogens are often located within solid food matrices, Kim et al. used a slightly different approach that involved the incorporation of E. coli-specific immunoglobulin G (IgG) antibodies into a microneedle-based platform (Figure 7d). [179]Specifically, porous silk microneedle substrates were functionalized with polydiacetylene liposomes that underwent colorimetric changes in response to mechanical stressors such as binding events between the immobilized antibodies and target E. coli cells.Owing to the porous nature of the needles, capillary action mediated the collection of test solution onto the surface of the functionalized microneedle patch.Paired with the ease of colorimetric readout, this platform's ability to sample pathogens from within food samples rather than just the surface presents an intriguing approach for future works to explore.These microneedles also exploit encapsulation of their colorimetric dye to limit the degree of external manipulation required for sensor functioning.High antibody specificity also ensured no cross-reactivity with non-target bacteria, based on evaluations with Salmonella-contaminated fish.However, a comprehensive assessment of the platform's limit of detection was not completed.
Antibodies can also be used for impedance-based detection in other ways, usually following gold nanoparticle-based conjugation.For example, Vu et al. ( 2021) developed an electrochemical sensor by modifying screen-printed electrodes with AuNPs, which were then further conjugated to E. coli specific antibodies. [54]The resultant system exhibited a limit of detection of 15 CFU mL −1 within 30 min E. coli-specific antibodies introduced specificity, which reduced AuNP-mediated nonspecific reactivity.However, basal responses to other bacteria still existed.Greater potential for interference exists by bacteria that are closely related to E. coli but are not pathogenic strains themselves.Using a similar approach, Chiriaco et al. ( 2018) developed a portable microfluidic platform for the detection of Listeria monocytogenes. [180]Specifically, gold microelectrodes within the microfluidic chip were functionalized with L. monocytogenesspecific antibodies using protein A -a peptide that mediates immobilization orientation, thus maximizing functionality.The resultant system demonstrated a limit of detection of ≈6 CFU mL −1 in artificially spiked milk samples.This study, however, went a step further, addressing issues pertaining to on-site electrochemical detection.Specifically, this work employed a portable electrochemical impedance spectroscopy device that was connected to a portable impedance analyzer for measurements -a setup that may be applicable on a commercial scale.Unfortunately, the sensor is limited to liquid test matrices, due to its microfluidic nature.Nonetheless, further development of platforms compatible with such portable technologies may enable the widespread commercial use of such highly sensitive electrochemical sensors in the food industry.
Antibodies offer high specificity and easier immobilization compared to DNA-based probes.Antibodies are also widely produced for the development of commercial biosensors, such as pregnancy tests, and as such are easy to translate.However, they are less researched due to their disadvantages which include their high cost and difficulties pertaining to functionalization due to their complex structural nature.Despite these disadvantages, interesting antibody-based sensing platforms are being developed.These works require improvements for in situ package detection but are nonetheless intriguing devices with scope for future developments.
Ultimately, these three main biorecognition probes offer high sensitivity and specificity, making them good choices for use within food contamination detection sensors.Consequently, these probes offer significant promise in terms of actualizing in situ contamination detection for specific pathogens, given the large body of work supporting future developments.

Situational Assessment and Future Directions
Despite the material-driven development of a plethora of food spoilage and contamination sensing systems -each with its own set of benefits and drawbacks, no platform has achieved widespread commercial acceptance as of yet.A wide range of issues continue to hinder real-world viability of individual platforms -such as inadequate sensitivity, lack of specificity, insufficient material stability, poor in situ compatibility, and cost.The objectives of future works will resultantly center upon addressing these shortcomings.

Technical Assessment and Future Directions for Food Spoilage Monitoring
While promising, the real-world applicability of biogenic amine sensors remains uncertain, given the unclear quantitative relationship that exists between biogenic amine levels and food spoilage.Specifically, while the production of these compounds is indicative of spoilage, threshold values that indicate the specific point at which a given product has expired are difficult to establish.Such thresholds are necessary for the effective elimination of food waste.Future works that seek to define these values will significantly substantiate the use of such technologies by mediating the design of sensors that exhibit positive signals at the point at which a given food product becomes inedible.pH-based monitoring similarly struggles with a lack of concrete threshold values indicative of spoilage, as various biochemical changes in food can induce fluctuations in pH, making the relationship between pH and spoilage difficult to define.As such, future studies that comprehensively evaluate this relationship within different foods are needed.While severe changes in pH can confirm food spoilage, this is meaningless to the consumer as such changes are paired with alterations in odour and visual properties that are easily observed.Better understanding the relationship between spoilage and both biogenic amine concentration and pH may also substantiate the need for ultrasensitive detection -an objective that will be supported by highly responsive probes such as aptamers.
As it currently stands, monitoring biogenic amine production remains the best indicator of the spoilage state of many different kinds of foods in real-time owing to a more concrete quantitative relationship.Intelligent noses represent perhaps the most promising class of biogenic amine sensors because of their high sensitivity, tunable specificity, in situ compatibility, and vapourresponsive nature -which eliminates the need for food sample manipulation.The diverse range of material approaches applied in this space yields several avenues of future exploration.Next steps pertaining to the design of in situ pH sensors are more ambiguous given that most platforms reported thus far continue to suffer from the same issues -limited film strength and poor dye stability.Nanoencapsulation of pH responsive dyes is currently positioned as the most promising avenue for platform improvements.On a different front, Riaz et al. ( 2019) recently showed that wild-cabbage-sourced anthocyanins can be crosslinked to poly-2hydroxyethyl methacrylate contact lenses to detect pH changes in the eye. [181]Given that improvements to cellulose, chitosan, and starch films have yielded limited improvements, the exploration of such alternative polymers may be justified.
A second consideration for pH sensor design pertains to a growing focus on multi-functional intelligent food packaging systems.pH monitoring platforms are well-positioned to contribute to this space given that many organic pH responsive dyes and film modifiers offer relevant secondary functions.Specifically, the antioxidant and antimicrobial functions of anthocyanins may have significant implications for food preservation and contamination response.Ma et al. ( 2019) reviewed the possible implications of these dyes for pathogen control in food. [112]A few recent studies have brought increased focus toward this preservative measure.Namely, Chen et al. ( 2020) developed a pH-responsive cellulose film through the incorporation of purple sweet potato anthocyanin that exhibited strong antimicrobial properties, with up to 99.99% reduction of E. coli and L. monocytogenes. [182]Further development of such systems is warranted.
While many of the discussed secondary modifiers also offer a degree of antimicrobial activity, films may also be embedded with antimicrobial agents directly.For example, Mustafa et al. (2020) explored the incorporation of propolis extract (PE) into PVA/starch films cross-linked with boric acid. [183]Embedded PE exhibited antimicrobial activity against both E. coli and methicillin-resistant Staphylococcus aureus.PE incorporation was also shown to prolong the shelf-life of milk, as the antibacterial activity exhibited by the films was sufficient to delay milk spoilage.Perhaps more intriguingly though, bacteriophages present a class of safe, antimicrobial agents that can also be used to preserve foods and prevent contamination.These entities are already used in the food space commercially, and several studies have reported their encapsulation into films for improved stability and performance.The incorporation of bacteriophages into pH-responsive films may offer significant benefits.
While many developed platforms appear promising within a controlled, experimental context, commercial application will require consideration toward real-world environment-induced variations in signaling performance.This is likely to incite the comprehensive analysis of sensing interferences within diverse food products, multiplexing of different probes for improved reliability, and a focus on reagentless design.This latter consideration is especially important, as many of the complex sensing cascades presented in this work require a range of reagents that may influence food spoilage.Finally, with improvements in sensor design, artificial intelligence, image analysis, and machine learning are expected to become key pillars in spoilage monitoring.Through the neural network-mediated training of intelligent signal processing systems, the risk of false positives can be expected to be negated to commercially acceptable levels.86]

Technical Assessment and Future Directions for Pathogen Monitoring
The need for improved real-time pathogen detection in food is well-established given the high prevalence of foodborne illness worldwide.The aforementioned developments have led to substantial gains in both on-site and in situ detection platforms for targeting bacteria within complex food matrices in real-time.The objective of future works will center upon addressing remaining shortcomings -namely improving the sensitivity of in situ platforms.
Making improvements to in situ sensitivity has proven difficult considering that most conventional signal amplification strategies do not have the necessary characteristics for incorporation.Ideally, they must function autonomously with no user interference, use biocompatible reagents, and be able to contribute to the reaction cascade while in an immobilized, solid-state form.As such, future efforts must rather look toward improving base probe sensitivity and incorporating novel signal amplification strategies.
Using bacteriophage as a sensing probe for bacterial detection in food has been reported sparingly, offering significant flexibility in terms of pathways for improvement.With more established oligonucleotide and antibody probes, one simple approach aimed at improving probe sensitivity involves modifying the means by which they are incorporated onto a sensing substrate.Specifically, improper orientation or restrictive immobilization can induce low functionality, yielding poor sensor sensitivity.This premise was explored by Liu et al., who non-covalently immobilized E. coli-responsive DNAzymes into graphene sheets and found that such a system permitted single cell detection -partially due to the free nature of the DNAzyme. [187]There are a plethora of different immobilization strategies that can be used for probe integration, each with unique characteristics that make them optimal for particular platforms.Such incorporation strategies have been reviewed by our group previously. [163]With the integration of new sensing substrates, novel immobilization strategies can be implemented.Alternatively, whole cell biosensing represents another viable avenue of exploration, with proven efficacy within other biosensing applications.To date, use in food sensing has been limited to very few studies. [188]Their complex nature and reliance on living bacteria likely limits these platforms to on-site detection, but their excellent performance garners future interest.
While many studies have evaluated bacteriophages, nucleic acids, and antibodies as probes for bacterial detection, they have so far been evaluated in isolation.As different probes carry distinct reactivity for different species of bacteria, these probes can be coupled with one another to create multiplexed sensors for bacterial detection.This is especially true for nucleic acids and antibodies, which can be functionalized with different types of fluorescent and colorimetric signal molecules, allowing for efficient multiplexing.In contrast, bacteriophages cannot be so easily chemically functionalized without permanently modifying their survival and their ability to detect bacteria. [189]However, genetically edited bacteriophages encoding different types of visible reporters present one possible path forward.Additionally, multiplexing is also possible using a mix of these different probes.As bacteriophages induce bacterial lysis, they cause the release of bacterial proteins, which may aid in nucleic acid probe and antibody-mediated detection.
Considering that bacterial contaminants cannot be detected in vapor -as biogenic amines and pH often are, but rather require direct contact with the sensing substrate, the detection of contamination in situ is limited to the food surface to which they are applied.Most of a given food product is thus left untested.Microneedle-based biosensors present an interesting opportunity to the development of intelligent packaging because of their unique ability to penetrate the biological matrix of solid foods, allowing them to collect more robust liquid test samples -which are drawn out by capillary forces, that are representative of both the food surface and the underlying matrix.Aside from the study by Kim et al. (2020), microneedles have also been used in the food industry in the context of allergen detection by sampling DNA from complex food matrices. [179]Given their ease of fabrication and use however, this represents perhaps one of the most defined future directions for pathogen sensing work.A variety of distinct compounds -such as anthocyanins and antibodies, could also be functionalized onto the interior and exterior surfaces of these microneedles to elicit measurable signals.
On a broader scale, many of the presented in situ sensing systems require access to defined volumes of liquid test samples and have reaction cascades that require buffer solutions.For example, DNAzymes, which require metal ions to properly adopt a secondary structure, require the addition of metal-ion reaction buffers to sensors for their proper functioning. [16]It is optimal that biosensors that are incorporated into packaging materials are able to function without external reagents as they can act as sentinels for bacterial growth and spoilage over time.Novel approaches are required to design reagentless biosensors for this purpose, or to allow the incorporation of such buffers into packaging materials.A packaging system that mediates collection of released juices and meat purge onto a sensing interface and facilitates stable incorporation of biocompatible buffers would thus substantially increase the reliability of many developed platforms.Printing probes onto organic polymers that also embed required reagents provides another potential alternative.With improved consistency, artificial intelligence can be better applied in food sensing, as complementary signal reading and processing devices would be easier to calibrate.
Finally, while spoilage and pathogen monitoring largely remain separated in research given the complex nature of each in isolation, it is anticipated that combined technologies will soon emerge.The need for such multiplex systems is apparent, given that only with spoilage and pathogenic contamination considered together, can we attain a comprehensive assessment of food safety.With continued emphasis on new avenues of detection, recognition toward areas of overlap is expected to drive such work forward.Material-driven innovation can be expected to be at the center of such efforts, as the incorporation of two different sensing cascades into a signal platform will require various physical, chemical, and biological consideration.To this end, smart material advances that enable the development of both in situ and on-site platforms for use at the production, distribution, and consumer levels will define food safety going forward.

Regulatory Considerations
Significant barriers continue to limit the translation of the presented in situ platforms from promising prototypes to approved products.Specifically, while many of the platforms discussed in this paper have significant potential in terms of their sensing ability, they incorporate materials that are not approved for foodcontacting applications and are potentially harmful to the environment.Few studies have evaluated the extent of material leaching and the impact that these materials may have on human health and the environment.Overarchingly, regulatory approval of on-package labels and other on-food sensors usually involves ensuring that the leaching of immobilized compounds into foods stays below a threshold of 0.5 ppb. [190]It is unclear if the presented in situ biosensors meet these guidelines.
The regulatory requirements for approving new nanomaterialbased materials for intelligent food packaging applications are stringent, with comprehensive clinical and environmental toxicology profiles required by many governmental agenciesnamely the Food and Drug Administration (FDA) and the European Union (EU). [191,192]EU regulations go a step further, requiring packaging nanomaterials to be listed as ingredients on food products, given the possibility of leaching. [191]Improper regulatory management has led to the recall of packaging materials in the past, making such considerations a priority during commercial development. [192,193]Regulations mediating relevant performance metrics are also important to consider when evaluating the design and efficacy of intelligent food monitoring systems.In terms of microbial contamination, critical thresholds for pathogenic bacteria have been set by various global agencies.Namely, the FDA limit for generic strains of E. coli is often as low as 10 2 CFU g −1 . [194,195]Pathogenic E. coli strains such as O157:H7 are limited to even lower thresholds, as they pose a significantly greater threat to human health. [196]Similarly, levels of Salmonella contamination in pork are as low as 10 CFU g −1 . [197]With regards to chicken, cooked products have a zero tolerance 0 CFU g −1 threshold.The rational design of ultrasensitive biosensors that consider these thresholds can inform the creation of relevant intelligent sensors for food safety.As mentioned previously, such thresholds have not been set for spoilage, yielding ongoing societal reliance on pre-determined expiry dates that lack accuracy.
Finally, regulatory approval of performance claims requires significant consideration toward sensing interference.This includes thorough evaluation of developed sensing platforms within diverse food product to eliminate concerns pertaining to food matrix-induced false positives.The ability of these platforms to function throughout the food distribution pipeline, which involves significant variations in temperature, humidity, and duration of storage, is also of particular importance when considering claims centered around real-world viability (Table 3).

Table 3.
Recently developed sensing platforms for pathogen identification in food.

Commercial Considerations
Yet, while a plethora of promising on-site and in-package sensing platforms have been presented here, commercial success in this space has been scarce.To this end, Table 4 highlights some of the presented platforms that have made some progress toward commercialization.While the specific factors that hinder the realworld use of a given platform are specific to its design, overarching barriers include limited real-world consideration, high cost, and consumer hesitancy.

Limited Real-World Consideration
While the aforementioned Table 4 highlights select platforms that have positioned themselves as commercially promising products, many platforms in the food space do not adequately account for real-world considerations.Namely, food products are regularly oversimplified, treated as homogenous matrices with sensing studies.In reality, a single product offers several different microenvironments -all with varying properties that can undermine biogenic amine, pH, and pathogen monitoring. [198]This makes in situ platforms suited for real-world use extremely difficult to design, drawing attention toward on-site sensing platforms that can incorporate a product homogenization step prior to testing.Alongside failing to enable individual product monitoring, such measures do not account for the diverse farming and processing practices employed today, that yield compositional differences among individual food products. [198]ven with all such factors considered, the controversy surrounding the existence of consistent biogenic amines and pH baseline and spoilage values complicates real-world use.While many works have reported baseline and spoilage values, other studies have found large degrees of variation within individual products. [199]Such variations make the universal calibration of such platforms difficult.Ultimately, there exists a defined need for studies that comprehensively evaluate large numbers of food products sourced from different supplies, to provide a concrete situational assessment.In the event that baseline and spoilage values are largely inconsistent, the presented platforms can still offer significant value, but will require significant modifications in their reporting structure.For example, percent change in signal may be more useful than objective values.Of course, all such efforts will be supported by the anticipated development of more sensitive and specific sensing cascades, as detailed in Sections 4.1 and 4.2.The development of such performance metrics will also help identify technologies with the greatest potential in a currently crowded area of study.
With regards to pathogen detection, the use of artificially spiked food products lowers the translatability of the results to the real world.Specifically, the distribution of bacterial contaminants on whole food products following real-world contamination has not been comprehensively studied.As such, artificial contamination involves the homogeneous distribution of bacterial targets across the surface of a test matrix -a protocol that makes consistent detection more attainable.Contaminated food samples sourced from industry must thus be incorporated in future studies.

High Cost
The commercial viability of many developed food sensors is also hindered by unclear value propositions.In situ platforms in particular, can be expected to add noticeable cost to food packaging -an industry in which very minimal increases in cost are often considered too dramatic to implement.While increased packaging costs can be partly justified through the prevention of food recalls, it has been estimated that in situ sensing platforms would account for over 50% of the cost of such smart food packaging. [200]The food industry expects that packaging should not account for more than 10% of the cost of goods sold, meaning that such a significant markup is unrealistic. [200]Ultimately, developing a compelling commercial case for in situ sensing platforms involves a greater focus on the cost of resultant platforms -a shift that has been observed within some recent works.Further, there exists a need for accurate estimations of the expected economic benefit of incorporating such sensors into packaged foods, to ensure that the cost of implementation is significantly lower than the anticipated savings.Such efforts must also account for the start-up costs associated with the development and mass manufacturing of such sensors.
On-site sensors offer a stronger commercial case as a replacement for culture-based methods that are expensive and logistically complex -largely owing to their need for off-site processing and reliance on sophisticated laboratory equipment.Acting as a first line of defense, such platforms can function as rapid tests that can replace some -if not all, culture-based testing protocols.Here, several presented works have shown promise, owing to their rapid detection and high sensitivity. [84,170,179]Given that testing is done periodically on select food products, a compelling cost-benefit ratio is much more achievable with such platforms.Nonetheless, minimizing cost is instrumental, to justify the significant capital required to develop and mass produce such platforms.Concurrently, ensuring that the approaches used for the fabrication of a given platform are suitable for commercial-scale manufacturing is also a defining consideration.

Consumer Hesitancy
Numerous studies have evaluated the willingness-to-pay (WTP) of consumers for in situ food monitoring technologies. [201,202]onsumers have a higher WTP and general interest in such products when they pose a greater societal benefit with minimal risks toward health.To this end, labels embedded with reactive, foodderived agents such as anthocyanins are especially compelling.While various synthetic nanoscale agents have been categorized as food-safe, perceptions surrounding their incorporation in food remain largely negative. [201,202]A mix of consumer education and technological adaptation in response to consumer feedback can be expected to yield improvements in this space.
Concurrently, the environmental implications of such packaging platforms are of great interest to consumers. [202]The integration of biodegradable materials such as cellulose and chitosanas discussed in this review, can thus be expected to aid in consumer acceptance.In comparison, the use of metals and plastics can generate runoffs that pose a threat to both environmental and human health.For platforms that require such agents, devising sustainable strategies for disposal are essential for large-scale distribution.Considering such factors proactively when designing food monitoring technologies is a necessary shift in this space, to ensure that more developed platforms achieve commercial success.

Figure 1 .
Figure 1.Current state of food spoilage and food contamination.a) Schematic illustration of food spoilage.b) Key contributors to food spoilage and their impact on the organoleptic properties of food.c) Schematic illustrations of the food contamination timeline, and stages at which pathogenic contamination can occur.d) Annual global illness rates caused by respective pathogens obtained from the World Health Organization.Regulatory limits as defined by the Food and Drug Administration for ready-to-eat (RTE) products.[17,18]Infectious dose recorded in literature.Produced using BioRender.

Figure 2 .
Figure 2. Commonly used substrate materials for food monitoring.a) Materials used as substrates within food sensors.b) Summary of relevant properties and use cases of common substrate materials.Primary application refers to situations in which the described material is most commonly applied but should not be considered all-encompassing.Produced using BioRender.

Figure 3 .
Figure 3. Strategies to detect biogenic amine accumulation in food.a) Structures of common biogenic amines produced during food spoilage.b) Interactions that mediate the detection of biogenic amines.c) Mechanisms for signal transduction in response to biogenic amine binding: (i) colorimetric nanoparticle aggregation, (ii) fluorescence shifts within metal-organic framework, and (iii) signal-inducing conformational changes within probes.Produced using BioRender.

Figure 4 .
Figure 4. Intelligent monitoring of pH changes in food.a) Opposing pH changes in different subtypes of foods, mediated by distinct processes.b) Anthocyanins, the predominant organic food dye used in pH-mediated monitoring of food spoilage, undergo chemical reactions that result in dramatic color shifts.c) Various organic materials used to derive intelligent pH-monitoring films, and parameters for optimization to enhance performance in situ.Produced using BioRender.

Figure 5 .
Figure 5. Physical properties of intelligent pH-responsive films used to monitor spoilage.a) Common modifications made to films that alter their mechanical and thermal stability, including: (i) the addition of hydrophilic and (ii) anionic compounds to facilitate cross-linking of film materials, (iii) nano-encapsulation of pigments to increase long-term stability and prevent heat-or light-mediated degradation, and (iv) nanofillers that increase film density and strength.b) A comparison of the properties of different pH-responsive organic dyes used in intelligent packaging.Produced using BioRender.

Figure 6 .
Figure 6.Biorecognition probes used for on-site pathogen detection and in situ food monitoring.a) The common types of probes that are used to identify bacteria -antibodies, bacteriophages, DNAzymes, and aptamers.b) Various probe-based strategies to detect bacterial contamination, including: (i) phage-mediated identification, transduced via amperometric monitoring of target-binding, or phage replication-induced resonant frequency shifts, (ii) aptamer-mediated recognition and resultant nanoparticle aggregation, (iii) DNAzyme-mediated detection, resulting in fluorescence changes, and (iv) antibody-based recognition, inducing colorimetric transduction via the use of dye-loaded, antibody-gated liposomes integrated into microneedles.Produced using BioRender.

Figure 7 .
Figure 7. Material strategies for the improvement of foodborne pathogen sensors.a) Material structuring to induce higher probe density.b) Material encapsulation for the preservation of probes over time in response to environmental stresses.c) Antifouling coatings for the prevention of non-specific adhesion on the sensing surface.d) Form factor manipulation to increase sensor contact with test matrix.Produced using BioRender.

Table 1 .
Recently developed platforms for biogenic amine detection in food samples.

Table 2 .
Relevant properties of base film materials and modifiers for in situ pH sensing.−11 gm m −2 s −1 Pa −1 )

Table 4 .
Food monitoring technologies in the commercial pipeline.