Advancements in Airborne Viral Nucleic Acid Detection with Wearable Devices

Wearable health sensors for an expanding range of physiological parameters have experienced rapid development in recent years and are poised to disrupt the way healthcare is tracked and administered. The monitoring of environmental contaminants with wearable technologies is an additional layer of personal and public healthcare and is also receiving increased focus. Wearable sensors that detect exposure to airborne viruses can alert wearers of viral exposure and prompt proactive testing and minimization of viral spread, benefitting their own health and decreasing community risk. With the high levels of asymptomatic spread of Coronavirus Disease 2019 (COVID‐19) observed during the pandemic, such devices can dramatically enhance the pandemic response capabilities in the future. To facilitate advancements in this area, this review summarizes recent research on airborne viral detection using wearable sensing devices, as well as technologies suitable for wearables. Since the low concentration of viral particles in the air poses significant challenges to detection, methods for airborne viral particle collection and viral sensing are discussed in detail. A special focus is placed on nucleic acid‐based viral sensing mechanisms due to their enhanced ability to discriminate between viral subtypes. Important considerations for integrating airborne viral collection and sensing on a single wearable device are also discussed.


Introduction
Wearable sensing devices are enhancing human health and enabling personalized healthcare by providing detailed metrics on personal physiological parameters, such as heart rate and body temperature, [1,2] which can correlate to general health status and predict disease onset.Wearable sensors are small, flexible, and are designed to be used throughout an individual's daily activities.Thus, unlike traditional, bulky, hospital-bound medical DOI: 10.1002/adsr.202300061devices, they can provide continuous monitoring of important physiological parameters in a non-invasive or minimally invasive fashion. [3][6] Since these parameters can change before an individual becomes symptomatic, these strategies can enable early detection of illnesses like Coronavirus Disease 2019 (COVID-19).[6] With improvements in flexible substrate materials and sensor technology, the applications of wearable sensors are expanding to include tumor size monitoring, [7] hydration, [8] and skin-mimicking tactile sensing for prosthetics. [9]Besides monitoring changes in physical parameters like touch, motion, and temperature, next-generation wearable sensors are also being developed to identify disease-signaling biomarkers in human fluids like sweat and interstitial fluid. [2]These advancements will have an impact on telehealth, human performance monitoring, and public health measures.
While significant research has looked at improving the range of physiological parameters that can be detected by wearable sensing devices, environmental monitoring is also important from a healthcare perspective.Exposure to environmental stressors like air and noise pollution can significantly increase an individual's risk of developing illness. [10]For example, extended exposure to airborne particulate matter less than 2.5 μm in diameter (PM 2.5 ) is associated with an increased risk of myocardial infarction, [11] lung cancer, [12] and diabetes. [13]The burden of airborne pollution on mortality is even estimated to be higher than acquired immunodefiency syndrome, malaria, and tuberculosis combined. [14,15]Thus, being able to monitor environmental exposure to pollutants using wearable sensors is valuable as it can inform users of their risk of developing illness based on their personal exposure level.Users can then respond to this information by proactively take measures to minimize their risk of developing illness.In a similar manner, the COVID-19 global pandemic has highlighted the need for wearable sensors that monitor exposure to specific airborne viral particles.COVID-19 spreads through airborne and droplet transmission [16,17] with evidence suggesting that significant transmission occurs from presymptomatic or asymptomatic individuals. [18,19]Since a large percentage of these COVID-19 cases are likely to go undetected, [20] this represents a significant gap in our efforts to combat the virus.Wearable sensors could help address this gap by informing wearers of viral exposure so that they can proactively seek testing and take precautions to minimize viral spread before symptoms indicate possible infection.
[23] These devices have demonstrated the ability to detect low concentrations of airborne viruses.However, their focus on exhaled aerosols means that these sensors can only be used to diagnose COVID-19 in sick patients, rather than providing continuous, environmental monitoring of airborne viral exposure.Thus, they cannot be used to minimize viral spread by informing wearers of possible infection.[26] Accurately monitoring these viral particles at such low concentrations requires that the particles are collected from the air and concentrated or amplified before detection of the virus takes place.Therefore, any wearable sensor that monitors exposure to airborne viral particles must integrate three sequential unit operations: airborne particle collection, virus nucleic acid detection (primarily RNA for respiratory viruses but also DNA for a large class of adenoviruses), and reporting.
There has been significant research into these three unit operations in isolation.However, few studies integrate these into a single wearable device for airborne virus detection.To help facilitate the development of these integrated wearable sensors, we review recent research into airborne viral particle collection, wearable sensing, and reporting.Throughout, we discuss the benefits and limitations of different methods for carrying out these unit operations as it pertains to airborne viral monitoring in a wearable format.We also provide guidance on future research directions that should be pursued to help actualize these wearables.Since viral sensing is a broad research area, we primarily focused our review on technologies for airborne viral ribonucleic acid (RNA) or deoxyribonucleic acid (DNA) detection (vs protein surface markers) as nucleic acid-based detection methods can discriminate between viral subtypes. [27]This is particularly advantageous as the viral subtype can determine infectivity and the severity of infection. [28,29]While several recent reviews provide an in-depth investigation of each unit operation, [30][31][32][33] to the best of our knowledge, none focus on integrating these unit operations into a single wearable sensor for airborne viral monitoring.Thus, our review fills an important gap in the literature with the intended goal of spurring future work.Although achieving continuous monitoring of airborne viral nucleic acids will be challenging, the realization of this goal could help improve our collective ability to combat viral spread by informing individuals of infection risks so they can take appropriate preventative actions.

Device Requirements for Wearable Airborne Viral Nucleic Acid Detection
Wearable sensors that report on the presence of airborne viral nucleic acids must carry out three sequential unit operations: sample collection, analyte sensing, and reporting (Figure 1).][36][37] The goal of the first stage is to collect enough of the target analyte, in this case, viral RNA or DNA, for detection by the chosen analyte sensing method.Current nucleic acid detection methods require the nucleic acids to be suspended in a liquid buffer. [21,38]However, airborne viral RNA or DNA is present in airborne viral particles at concentrations much lower than the limit of detection of current detection methods. [30,39,40]Thus, airborne viral particles must be collected and enriched from the air during sample collection.Lysis and extraction of the viral RNA or DNA into a liquid buffer follow so that the second stage, analyte sensing, can occur.During this stage, a biorecognition element binds the target of interest.Current biorecognition strategies for nucleic acids commonly center on the hybridization of the target sequence to a short complementary primer or probe such as in polymerase chain reaction (PCR) or clustered regularly interspaced short palindromic repeats (CRISPR), [21,38] respectively.The goal is to have the biorecognition element bind with high specificity and sensitivity to the target analyte.This enhances target recognition at low concentrations while also minimizing false positives due to nonspecific binding interactions with off-target molecules. [2,41,42]The complementary base pairing of nucleic acid sequences through Watson-Crick interactions are conducive to highly specific binding interactions, with probes being able to discriminate between closely related sequences. [41,43,44]However, since target analyte recognition through binding cannot be directly monitored, it needs to be transduced into a user-readable format.This occurs during the final stage of viral RNA or DNA detection, and reporting, with analyte detection being commonly transduced into a user-interpretable colorimetric or fluorescent signal. [21,23]hile many tools and methods exist for carrying out the required unit operations for airborne viral RNA detection, it is challenging to integrate these technologies into a single wearable device.Wearable sensors have several important design requirements that must be met to ensure usability and general performance. [2,45]First, they must be small, flexible, and durable so that they are compatible with wearable use. [21,22,46,47]Second, their energy requirements must be low enough to be powered by batteries or other types of wearable power sources. [48,49]They also need to have rapid detection times and high levels of accuracy.Specifically, for monitoring airborne viral RNA, it is desirable that the detection time of the sensor is rapid enough to alert wearers of viral exposure before they become infectious.This can enable the wearer to seek testing and take appropriate precautions to prevent viral spread.It is likely also preferable that the sensor only informs the user of exposure when it poses a transmission risk.However, the relationship between levels of airborne viral RNA exposure and transmission risk remains unclear since samples positive for viral RNA are not always culturable, likely due to the sample collection method used. [38,50,51]The challenging part of designing wearable sensors for airborne viral RNA detection is that the design requirements can be conflicting.For example, the requirement to keep the device small prohibits the use of large pumps for air sampling.This lowers the amount of air that can be sampled which can increase collection time. [38,51]o our knowledge, only one study has constructed a wearable sensor for airborne viral RNA monitoring that reasonably satisfies these design requirements. [21]For airborne monitoring of viruses more generally, two recent studies also built sensors satisfying the aforementioned design requirements, but their sensors targeted viral spike proteins. [22,23]Besides wearability, rapid detection time, and accuracy, other design considerations like ease-ofuse and privacy are also important but are not our focus.Further discussions of the design requirements for wearable devices, including cognitive and emotional requirements, can be found in the recent review by Francés-Morcillo et al. [45]

Aerosol Collection
Existing strategies for collecting viruses and other pathogens from the air and concentrating them for downstream characterization can be divided into two broad classes, those relying on active air sampling and those relying on passive (diffusive) air sampling (Figure 2). [30,52,53]Both collection mechanisms have advantages and disadvantages to integrating them into wearable sensors for the detection of airborne pathogens.Recent studies have primarily relied on active air samplers to study the spread of airborne viruses such as SARS-CoV-2 and Influenza A, both in hospital and commercial settings. [38,50,54]This is due to the ability of active air samplers to sample large volumes of air quicker than passive air samplers.23] Here, we provide an overview of both sampling methods and discuss their current research with a special emphasis on the benefits and limitations of using them in wearable sensing devices.More in-depth discussions of active and passive air samplers and their various applications can be found in several recent reviews. [30,31,55]

Active Aerosol Sampling
Active air samplers function by using mechanical means, such as internal fans, blowers, or pumps to force air into a collection apparatus at various flow rates. [58,61]Typical flow rates range from 3.5 to 500 L min −1 (Table 1).For reference, humans breathe at a rate of 6 L min −1 . [62]This use of forced airflow is what allows active air samplers to monitor larger volumes of air more quickly than passive air samplers and is what distinguishes the two collection methods.Once air enters the collection apparatus, aerosol particles of a targeted size are captured and enriched in a collection medium for downstream testing and characterization, usually PCR-based detection but can also include sample culturing and genetic analysis. [30,38,50]The sizes of captured aerosol particles, enrichment mechanism, collection medium, and other characteristics are summarized in Table 1 for recent studies.These characteristics can vary widely between samplers and can significantly affect sampler performance.Typically, performance for active samplers is determined by measuring the collection efficiency, defined as the fractional amount of particles captured by the sampler upon entering.[65] Thus, one needs to ensure that a sampler has a high collection efficiency for the particle size of interest.For viral collection, it is not clear what the optimal particle size is for collection.While viral particles are small (≈0.1 μm), [66] they  A-C) and passive (D-F) air samplers.A) Picture of IOM Inhalable Sampler.Reproduced with permission. [56]opyright 2023, SKC Inc.. B) Portable cyclone sampler (scale added; adapted under the terms of the CC BY 4.0 license. [57]Copyright 2021, The Authors).C) Sioutas Personal Cascade Impactor Sampler with Leland Legacy pump being worn by a Cambodian poultry worker (Reproduced under the terms of the CC BY 4.0 license. [58]Copyright 2022, The Authors, published by Wiley).D) PDMS-based wearable clip sampler for monitoring personal exposure to SARS-CoV-2 (Reproduced with permission. [52]Copyright 2022, The Authors).E) Electrostatic dust collector (Reproduced with permission. [59]Copyright 2008, American Society for Microbiology).F) ElectroStatic Sampler for Air (ESSA) an electrostatic precipitation sampler (Reproduced under the terms of the CC BY-SA 4.0 license. [60]Copyright 2019, Taiwan Association for Aerosol Research).
are present in larger aerosols particles. [30]Two recent studies were able to achieve similar levels of positive SARS-CoV-2 samples for various air fraction sizes from <2.5 μm to >10 μm. [38,50] better understanding of the optimal size fractions to target for monitoring infectivity risk could lead to the design of more efficient collection devices.
Other metrics besides collection efficiency can also be used to judge the performance of active samplers.Recently, Bhardwaj et al. [39] reported sampler performance using a potentially more useful metric called the enrichment ratio, defined as follows: The enrichment ratio relates the collection efficiency, , flow rate, Q A , sampling time, t, and collection volume, V L , to give the concentration of target aerosol particles in the collected sample relative to the concentration of the sampled air.Thus, it directly sets the required limit of detection (LOD) for any post-collection characterization method by the following relationship where C V is the initial particle concentration in the air: Being able to directly relate the enrichment ratio to the LOD is more informative for the design of wearable sensors that both col-lect and detect airborne viruses or other pathogens.It may also be informative to normalize the enrichment ratio by sampling time as this more clearly shows both the speed and level of enrichment.While high enrichment is required for adequate detection of viral particles, it needs to happen quickly enough for reasonable detection times if wearable devices integrating airborne collection and pathogen sensing are to be practical.
Of the active samplers reviewed, the majority relied on centrifugal force, impaction, or filtration as the mechanism for enriching the target aerosol particles (Table 1).Both cyclone-based samplers and impaction-based samplers rely on inertia to enrich target particle sizes. [30,57]In cyclone-based samplers, centrifugal force guides particles with sufficient size and inertia toward a collection medium at the bottom of the device.Lower inertia particles cannot travel all the way down and instead become adhered to the device walls.In impactors, aerosol particles are accelerated toward a collection substrate where particles of sufficient size and inertia hit the substrate and become trapped.When the collection substrate is a liquid, it is called impingement.Filtration-based samplers on the other hand rely on their filters to collect and enrich particles above a certain size threshold.Samplers utilizing all three collection methods have been shown capable of achieving high collection efficiencies for a wide range of particle sizes (Table 1).However, the direct comparison of the performance of the different sampler types for enriching airborne viruses is complicated by the fact that few studies [38,74] compared the  Bhardwaj et al. [67] Portable electrostatic particle concentrator N MS2 Virus

SKC Personal Button
Sampler with SKC Y SARS-CoV-2

Gelatin filter
Cylindrical: (<80 nm -0.9 μm) [65] Pivato et al. [  [73] 8-24 h 500 a) Devices are not always uniformly shaped so the reported device size should be viewed as an approximation. b) Collection efficiencies weren't always measured.In this case, when possible, we reported collection efficiencies from other studies that used the same or similar sampler.Since the collection efficiency could change based on flow rate and other parameters, these values should be taken as representative approximations.performance of different samplers under the same set of experimental conditions.Instead, studies typically only tested the performance of one sampler and differed in the target analyzed, detection method, and whether real or artificially generated air samples were collected. [38,54,58,61,63,67]One recent study that carried out a direct comparison of different sampler types did so in a nursing home setting.In this study by Linde et al., [38] aerosol samples were collected from confirmed COVID-19 patients rooms using an impingement-based sampler, a two-stage cyclone-based sampler, and a filtration-based sampler.The presence of SARS-CoV-2 in the samples was then determined by a two-gene reverse transcription quantitative polymerase chain reaction (RT-qPCR).
All three samplers achieved similar positive sample rates (50-70%) indicating that they have similar performance in real-world settings.The volume of air sampled by the impingement-based sampler was 40% lower than the other samplers though which indicates this sampler may have had better performance overall.However, the number of samples collected was too small to draw any definitive conclusions and only one sampler of each type was evaluated.Clearly, further research is required to compare the airborne virus collection ability of the different air sampling mechanisms more rigorously.This knowledge could help inform which collection mechanism to pursue for integration into wearable sensors.Besides collection performance, other metrics important to consider for integrating active air samplers into wearable devices include the sampler size and collection time.All three sampler types have been made small enough to be applied in personal air monitoring.For example, Horwood et al. [58] recently used a Sioutas Personal Cascade Impactor Sampler to measure the exposure of poultry workers to avian influenza viruses (AIVs).The sampler was worn by the workers while they completed their daily activities and the samples were tested for the presence of AIVs using reverse transcription polymerase chain reaction (RT-PCR).Through their investigation, they found that avian influenza viruses could be reliably detected in collected aerosol samples during periods of peak virus circulation.Similarly, Santarpia et al. [61] used the filtration-based SKC Personal Button Sampler to test for the presence of SARS-CoV-2 virus in healthcare settings and also reliably obtained positive samples from collected aerosols.While these studies collected their aerosols on a solid substrate, direct collection in liquid media is also commonly employed. [38,63,67]This is preferable for integration into wearable sensors as aerosols collected on solid substrates need to be extracted into a liquid buffer for PCR and related detection methods. [38,58,75]The personal air samplers used in both studies also require the use of an external personal pump which can be quite large relative to the size of the personal samplers (694−1400 cm 3 for pump vs 20-204 cm 3 for personal sampler).Although power requirements are not currently a prominent design metric for active samplers, power considerations will become more important as active sampling strategies are integrated into wearable sensors and miniaturized.

Passive Aerosol Sampling
Passive air samplers have been applied less frequently to the collection of airborne viruses than active air samplers.Unlike ac-tive air samplers, passive samplers do not require forced airflow and instead let aerosol particles settle and adsorb onto a collection substrate like polydimethylsiloxane (PDMS) or filter-paper under ambient air conditions. [52,53,60]The lack of a pumping unit tends to make passive air samplers simpler to use and operate than active samplers and they tend to be smaller.For example, setups as simple as electrostatic cloth attached to a folder and a small PDMS pad in a wearable clip have been used as passive air samplers. [52,59]Small size and simplicity are obvious advantages of using passive air samplers instead of active samplers in integrated wearable sensors for detection of airborne viral RNA.However, since passive samplers utilize much lower airflow rates, they suffer from long collection times.For example, in the study by Linde et al., [38] a passive electrostatic dust collector (EDC) was used to monitor the presence of SARS-CoV-2 in a nursing home.The EDC obtained similar positive sample rates as the active samplers simultaneously investigated in the study.However, the collection time for the EDC was 2-4 weeks, much longer than the active samplers which had collection times ranging from 1 to 6 h.Similarly, Angel et al. [52] used a PDMS-based passive sampler that can be affixed to a shirt collar to monitor SARS-CoV-2 exposure in Connecticut residents, including restaurants and healthcare workers.While the authors were able to detect SARS-CoV-2 in 8% of collected samples using reverse transcription droplet digital polymerase chain reaction (RT-ddPCR), the sample collection time was 5 days.Since the pre-symptomatic period for SARS-CoV-2 is estimated to range from <1 to 4 days, [76] this implies that subjects could be infectious with SARS-CoV-2 before symptoms or a positive sample from the collector would indicate that they should be tested.This is a significant current limitation to using passive air sampling methods for monitoring exposure to viral RNA.One potential solution is to incorporate electrostatic precipitation in passive sampling.This method uses collection electrodes to generate electric fields that charge and direct aerosol particles to the collection substrate. [60]Recently, Imani et al. [60] demonstrated that the collection range of the electrodes combined with ambient air flow could allow for volumetric flow rates between 150 and 360 L min −1 , rivaling those achieved by active samplers.However, the collection efficiency is lower than more established methods and no testing was performed using viral samples in real-world settings.

Integration into Wearable Devices
Both active and passive air sampling strategies have benefits and limitations to integrating them into wearable devices.While active samplers have low collection times and high collection efficiencies, their size poses challenges to using them in a wearable format (Table 1).Similarly, while passive samplers can achieve sizes small enough for use in wearables, their collection periods can be too long (order of days). [38,52][23] All the devices in these studies rely on active sampling to collect virus particles for downstream detection.To avoid the issue of having to carry a pumping unit, these devices are cleverly built into facemasks and rely on the lungs to exhale aerosol particles directly onto the inside of the facemask Figure 3. Images of integrated wearable sensors for collecting and sensing airborne viral particles.A) Facemask integrated with a collection pad for sampling exhaled viral particles.This collector can be interfaced with a dot blot assay (B) for simultaneous detection (scale added; adapted with permission. [23]Copyright 2021, American Chemical Society).C) Integrated facemask and nanoscale impedance immunosensor for detection of viral particles in exhaled aerosols (Reproduced with permission. [22]Copyright 2021, Elsevier B.V.).D) Diagram and photograph of integrated facemask collection and CRISPR-based sensing platform for detecting the presence of SARS-CoV-2 viral RNA in exhaled aerosols (Reproduced with permission. [21]opyright 2021, Springer Nature BV).
where they are collected.Specifically, Soto et al. [23] integrated their facemask collector with a dot blot-based antibody assay to enable specific and colorimetric detection of the SARS-CoV-2 S1 spike protein.Viral spike proteins were also targeted in Xue et al. [22] who interfaced their facemask collector with an antibodybased nanoscale impedance immunosensor.Nguyen et al., [21] on the other hand, developed a facemask-based device for detecting SARS-CoV-2 RNA by integrating freeze-dried cell-free reactions for viral lysis, signal amplification, and CRISPR-based target detection with a lateral flow assay.Τhe detection systems used in these studies will be discussed in more detail below, but, from a collection standpoint, they demonstrated suitable performance.Soto et al. [23] demonstrated that virus particles sprayed on the mask are adhered to the collector and detectable using their dotblot assay.Xue et al. [22] and Nguyen et al. [21] both demonstrated sensing of their target after collection during simulated breathing experiments with collection times under 30 min.
While these studies represent important steps toward developing wearable sensors for airborne viral RNA detection, they still need to demonstrate efficacy under real-world testing conditions and they only detect the presence of the virus in exhaled air rather than monitoring personal exposure to the virus from the environment.Thus, while they are useful for diagnostic purposes, they cannot inform users whether they have been exposed to the virus and are at an increased risk of infection.Monitoring viral exposure from the environment is desirable as it could inform wearers of the infection risks posed by different activities.In theory, this could allow wearers to tailor personal protective measures to the risk level of the activity being engaged in.To realize monitoring of viral exposure, active or passive sampling strategies that collect environmental aerosols instead of exhaled aerosols will need to be integrated into wearable sensors.Along with the personal active air samplers mentioned previously, backpack-based active samplers [77] are promising targets for future integration efforts.Passive sampling strategies should also be explored.

Sensing and Reporting Mechanisms
The sensing and reporting components of a wearable biosensor typically consist of a recognition element (e.g., complementary sequence probe) that binds the target analyte and an optical, electrochemical, or mechanical transducer that converts the binding event into a measurable signal. [78]The reporting mechanism in wearable biosensors for viral nucleic acid detection can be categorized into two main classes: optical and electrochemical signal transduction.Optical signal transduction involves the use of light to measure the detection signal, including fluorescence, luminescence, and colorimetry.[81] Currently, the quantitative polymerase chain reaction (qPCR) assay (optical transduction) is the gold standard for validation of new nucleic acid diagnostic methods and benchmarking their time to detection, sensitivity, and selectivity. [82]For protein diagnostics, enzyme-linked immunosorbent assay (ELISA), another optical transduction method, is a widely used and well-established technique for protein detection and quantification. [82]Sensing methods can further be classified by their amplification requirement of the target molecule.Methods presented in this paper are summarized in Table 2.

Amplification-Based Strategies
Target amplification is often necessary to increase the sensitivity of a diagnostic and can be performed prior to or alongside signal production (Figure 4).Traditional polymerase chain reaction (PCR) amplification requires thermal cycling equipment, which can be more difficult to miniaturize but can be accomplished with slightly more bulk, power, and cost. [91]However, work is being done to overcome these limitations.As an example, Cheong et al. [92] designed a point-of-care (POC) device capable of performing rapid PCR in 17 min.The device's heating system uses magneto-plasmonic nanoparticles to complete a PCR cycle from 58 to 90 °C in 8.91 s, allowing the device to perform a complete PCR in <10 min.The device also enables reverse transcription to be performed on the same sample in 5 min, with signal detection taking ≈3 min.The magnetic core of the nanoparticles allows for the separation of the magnetic nanoparticles from the fluorescence dye, which generates qPCR signals.The device has a LOD of 3.2 gene copies μL −1 .Due to limitations of polymerases that require thermal cycling, isothermal amplification methods such as loop-mediated isothermal amplification (LAMP), recombinase polymerase amplification (RPA), [93] and rolling circle amplification (RCA) are increasingly applied to wearable biosensors. [94]hese are summarized next.
Regarding recombinase polymerase amplification (RPA) and reverse transcription, recombinase polymerase amplification (RT-RPA) has also emerged as a popular method for DNA and RNA amplification, respectively, in wearable sensing applications.One of the primary advantages of RPA over PCR, is its isothermal nature, which eliminates the need for thermal cycling and allows for simpler, more portable devices.This feature allows RPA to be incubated using the lower temperatures of the human body (30-37 °C), making it ideal for wearables.As an example, a wearable RPA-enabled biosensor with a bandage-like design (Figure 4A) for the real-time detection of Zika virus was developed by Yang et al. [89] which was heated using the body; it was capable of amplifying DNA within 10 min.The total DNA amplified by RPA was monitored using SYBR-Green-1 fluorescent dye and a UV-handheld light and achieved a LOD of 10 copies μL −1 .
Additionally, RPA has been demonstrated in a freeze-dried format, thereby increasing the shelf life of reagents and facilitating storage and transportation. [21,93]Yang et al. [90] integrated microneedles (MN) with RPA in a wearable device (Figure 4C) for efficient collection of cfDNA from interstitial fluid samples.Their device demonstrated a collection efficiency of 42.9% for Epstein-Barr virus cfDNA at a concentration of 5 copies μL −1 and 95.4% at a concentration of 50 000 copies μL −1 .After the collection stage, the cfDNA was amplified using RPA, and the resulting amplicons were detected using a Ru complex acting as an electrochemical probe.The current intensity in the differential pulse voltammetry (DPV) experiment changed in response to the detection of cfDNA, and the current change was modeled.The LOD for the assay was reported to be 1.1 copies μL −1 .
Two recent RPA-based viral detection works include first, the work of Kong et al. [87] in which a wearable wristband device

Time to detection
Carr et al. [83] SARS-CoV-2 viral RNA Amplified DNA is used to trigger a cell-free reaction that produces a reporter protease.The degradation of a gelatin film by the protease is then monitored using a wireless resonant sensor.
RT-RPA 100 copies μL −1 8 h Cordray et al. [84] Plasmodium DNA Antibody-labeled gold nanoparticles are used to produce a colorimetric signal in the test line of a lateral flow paper device.This signal is triggered by the detection of RPA products.
RPA 5 copies μL −1 55 min Deng et al. [85] miRNA-21 A solution-gated graphene transistor that produces a shift in the Dirac point voltage upon hybridization with immobilized DNA probes on the Au gate electrode.
-1 0 −20 m 3-5 min Dou et al. [86] N. meningitidis DNA Fluorescent signal is generated by consuming quenching manganese ions from a calcein fluorophore in the presence of a target in the LAMP reaction.

LAMP
≈3 DNA copies (or 7.4 fg) Gao et al. [47] miR-4484 Graphene FET sensor with immobilized ssDNA that can produce an electrical signal upon hybridization with a target nucleic-acid.
-1 0 f m 2 0 m i n Kong et al. [87] HIV-1 DNA Fluorescence signal that corresponds to total amount of amplified DNA is monitored.A cell-phone system performs the signal reading.
RPA 102 copies mL −1 24 min for amplification Li et al. [88] 2019-novel coronavirus (nCoV-N) plasmid Colorimetric LAMP to produce a visible color change upon the detection and amplification of a target DNA sequence.

RT-LAMP -80 min
Nguyen et al. [21] SARS-CoV-2 viral RNA Target biomolecule triggers a cell-free reaction coupled with RPA that activates the Cas12a protein.The protein cleaves the quencher from the reporter fluorophore, generating a fluorescence signal.RPA 2.7 fm 2 h (30 min for collection and 1.5 h for sensing on facemask) Xue et al. [22] COVID-19 Spike protein Antibody-protein interaction causes an impedance signal change when the protein binds to the surface of a nanowire array.
-7 p f u m L −1 5 min Yang et al. [89] Zika virus fragment DNA cloned into pUC57 plasmid Fluorescence signal is generated when a fluorescent molecule binds to amplified DNA.RPA 10 copies μL −1 10 min Yang et al. [90] Epstein-Barr virus cell-free DNA The biosensor device uses a differential pulse voltammetry (DPV) experiment, where the current intensity changes in response to the detection of a target after amplification.RPA 1.1 copies μL −1 25 min (Figure 4B) was designed for the detection of HIV-1 DNA using a PDMS microfluidic system.To amplify the signal, RPA was employed along with Fluorescein amidites (FAM) as a fluorophore for labeling the primers.The FAM-labeled primers were designed to produce a fluorescent signal upon cleavage of the fluorophore dye from the primer during the polymerization process.The level of fluorescence signal was shown to be proportional to the amount of total target DNA in the sample, which increased as RPA progressed.The sensor demonstrated a linear response ranging from 102 to 105 copies mL −1 and a LOD of 100 copies mL −1 within 24 minutes of the RPA reaction.Second, Cordray et al. [84] developed a lateral flow detection device (Figure 4D) coupled with RPA amplification for the detection of Plasmodium DNA.Although this device is not wearable, the compact size of it makes it possible for wearable applications.The device consists of a slider that transports the sample between the zones for RPA, buffer exchange to prevent false-positive results and lateral flow assay.The FAM-and biotin-labeled product of RPA reacts with gold nanoparticles functionalized with anti-FAM in the test line to produce a colorimetric signal in the lateral flow assay.They achieved a LOD of 5 copies μL −1 within 55 min of adding samples to the device with visual inspections.Reverse transcription is a crucial step in RNA detection systems such as RT-PCR, RT-RPA, and RT-LAMP that allows for the amplification of viral RNA, which composes the bulk of respiratory viruses such as SARS-CoV-2.This step involves converting RNA into complementary DNA (cDNA) molecules, which are then amplified by the traditional RPA reaction.Furthermore, Representative amplification-based virus nucleic acid detection strategies suitable for wearables A) Target specific RPA reaction in a wearable bandage-like device to detect Zika virus genomic DNA.Fluorescent dyes were used to generate a signal and were detected using a portable UV-torch (Reproduced with permission. [89]Copyright 2019, Elsevier B.V.).B) Wearable wristband device for the detection of HIV-1 DNA using a PDMS microfluidic system; RPA was employed to produce a fluorescent signal upon cleavage of dye from the primer (Reproduced with permission. [87]Copyright 2019, Elsevier B.V.).C) Microneedle wearable device for collection of cell-free DNA (cfDNA) from interstitial fluid (ISF) samples; RPA was used to amplify the target nucleotide and a ruthenium (Ru) complex was used to generate electrochemical signals from the presence of target nucleotide (Reproduced with permission. [90]Copyright 2020, Wiley).D) A lateral flow detection device coupled with RPA amplification for the detection of Plasmodium DNA (Reproduced under the terms of the CC BY 4.0 license. [84]Copyright 2015, The Authors, published by Springer Nature).E) The SARS-CoV-2 mail-insensor workflow begins by using RPA amplicons of viral RNA to turn on the Toehold Switch, which starts a cell-free reaction that produces a reporter protease.The degradation of a gelatin film by the protease is monitored using a wireless resonant sensor to transfer resonance frequency shift data.(Reproduced with permission. [83]Copyright 2022, American Chemical Society).
RNA is generally less stable than DNA and present in lower concentrations in samples, making it more challenging to detect.Therefore, rapid and efficient reverse transcription is necessary for the sensitive and specific detection of RNA targets in diagnostic assays, such as RT-RPA. [95]Two recent examples of RT-RPA for SARS-CoV-2 detection use freeze-dried cell-free expression (CFE) to accomplish a wearable-suitable strategy.The first by Nguyen et al. [21] (Figure 3E) uses CRISPR/Cas [96] in which the RPA reaction triggered by the target dsDNA amplifies the RPA amplicons and activates Cas12a-gRNA complexes.This complex in turn causes the activation of the Cas12a protein.Simultaneously this protein utilizes its trans-ssDNase activity to cleave a quencher domain from ssDNA fluorophore probes to generate a fluorescence signal.They investigated the ability of the integrated CRISPR-based specific high-sensitivity enzymatic reporter unlocking (SHERLOCK) sensors for detection of SARS-CoV-2 viruses from exhaled-aerosols using a facemask.This face mask consisted of 3 major parts: a hydration reservoir; a wax-patterned μPAD which contains all freeze-dried compo-nent zones for lysis, RT-RPA and Cas12a SHERLOCK; and a lateral flow strip for reporting.They reached a LOD of 500 copies (17aM) of SARS-Cov-2 RNA within 1.5 h of sensor activation.A second example of RT-RPA with freeze-dried CFE components is our group's development of a mail-safe, paper-based system for SARS-CoV-2 detection [83] (Figure 4E) in which the viral RNA is first collected and converted to DNA off of the sensor.When applied to the sensor, the amplified DNA binds to a toehold switch and initiates a cell-free reaction to express a protease (subtilisin BPN′) that degrades a gelatin film monitored by a wireless, resonant sensor.This work achieved a LOD of 100 copies μL −1 and was capable of being read through a mail-safe, opaque envelope without having to open the sample.
Regarding LAMP, Soto et al. [23] developed a wearable facemask for viral sample collection which uses the LAMP strategy for signal amplification.They evaluated their sensor capturing efficiency by qualitatively comparing the amount of virus in a stock solution, the amount of virus recovered from the mask after application, and controls using reverse transcription loop-mediated isothermal amplification (RT-LAMP) coupled with a colorimetric readout.The researchers also evaluated their wearable collector system by testing it with an aerosol spray of 15000 heat-inactivated SARS-CoV-2 viruses.By using RT-qPCR, they estimated that around 585 viruses were captured.They also performed a dot blot assay on a redesigned sensor to investigate the ability of their sensor to detect SARS-CoV-2 surface protein using a colorimetric readout. [23]In addition to this study, work by Dou et al. [86] has also demonstrated the potential of using LAMP in wearable sensors.Their PDMS/paper biosensor is coupled with microfluidic channels and performs LAMP on the paper fluid zones.The paper-based detection zone is made of nitrocellulose paper and contains LAMP amplification reagents that are specific to the target DNA.When the sample flows through the detection zone, any target DNA that is present in the sample will bind to the primers and initiate a LAMP reaction.Detectable fluorescence signal is generated by depriving quenching manganese ions from calcein fluorophore.They were able to reach a LOD of 3 copies of target DNA per LAMP zone in 45 min and at 63 °C.
Finally, regarding RCA, Ali et al. [97] developed a paper stripbased biosensor to perform rolling circle amplification (RCA) upon detection of a target synthetic ssDNA.In summary, poly(Nisopropylacrylamide) microgels conjugated with DNA1 is ligated with DNA2 in the presence of a target DNA3 as a bridge for ligation, producing a DNA1-DNA2 ligated product.The newly ligated DNA1-DNA2 product will act as a primer to start an RCA reaction with the circularized template to make long, concatemerized ssDNA molecules.A fluorescent-labeled DNA probe is added for signal transduction.The reported LOD of this work is 100 pm and the time until a detectable signal is 140 min.

Non-Amplification Based Sensing Strategies
Non-amplification-based biosensors (Figure 5) have a few advantages over the techniques just reviewed.One advantage is that they can provide a rapid response and do not require the time and complexity associated with enzymatic amplifications, making them more readily suitable for "real-time," wearable applications.They also do not require specialized equipment such as a thermal cycler [98] and can be simpler to use, with lower cost and lower power requirements.Furthermore, non-amplificationbased biosensors may be less susceptible to interference from background signals and non-specific binding in certain used cases. [99]This is because random binding events cannot initiate an amplification cascade that subsequently produces a false positive signal.However, the sensitivity and specificity of nonamplification based biosensors is usually lower than those of amplification based biosensors, making it difficult to detect low levels of analytes or distinguish between closely related molecules.This presents a challenge to detect the low concentrations of airborne viral particles.
One promising class of sensors that does not require amplification is field-effect transistor (FET) sensors.FET sensors can detect analytes in a label-free manner, eliminating the need for fluorescent or colorimetric reagents and optical readers.Their high sensitivity removes the necessity of a prior target amplification which improves their detection speed, ease of use, and re-duces false positive results from non-target amplification. [99,102]he most common FET sensors for wearable applications are graphene-based FET sensors due to their mechanical properties (e.g., flexibility), as well as high surface area, high carrier mobility, and low noise [103] which makes them ideal for detection of biological targets like nucleic acids, proteins, and cells. [104]The working principle of this sensor is based on the affinity of the target molecule to the graphene surface, which changes the local electrostatic potential and thus the electrical conductivity of graphene.This change in conductivity can be measured by a nearby gate electrode in an FET configuration. [105]ne promising FET sensor for use in wearable RNA sensing is a flexible graphene field-effect transistor (GFET) recently developed by Gao et al. [47] (Figure 5A).This sensor operates based on the change in the position of the Dirac point voltage caused by the hybridization of target miRNA molecules with immobilized ssDNA probes on a graphene layer.The sensor demonstrated a LOD of 10 fm and a detection time of 20 min when targeting miR-4484, a potential biomarker for increased expression of MMP-21 in systemic sclerosis patients. [106]While the authors performed a selectivity study by comparing the sensor's response to miR-3646, miR-4732, and miR-K12-5, further selectivity studies for similar miRNA families are necessary.Additionally, the need for purification of the target miRNA and the sensor's ability to work with unmodified human samples should be addressed to enable the development of a wearable device.
Another example of FETs adaptable for wearables is the work of Deng et al. [85] in which they recently developed a solution-gated graphene transistor (SGGT) (Figure 5C) to detect miRNA-21 (a biomarker for the early diagnosis of prostate cancer) in clinical samples with minimal sample preparation.The presence of the target miRNA was determined by analyzing the position shift of the Dirac point voltage upon hybridization with immobilized DNA probes on the gate electrode.The sensor achieved a limit of detection of 10 −20 m with a 5 min response time.Selectivity of the sensor to a control miRNA with only one base-pair mismatch to miRNA-21 was also studied, and the signal difference (ΔV Dirac ) between the target and control was achieved at a concentration of 10 −15 m.However, this work required miRNA extraction and dilution steps prior to sensing the target which can be an obstacle for using FET sensors in wearable devices.
One of the main challenges of using FET sensors in wearable devices is the non-specific binding to other chemicals in a biological sample.This limits the sensor sensitivity and necessitates sample preparation. [107,108]To overcome this challenge, Wang et al. [100] have developed an aptamer graphene-Nafion biosensor (Figure 5B) for the detection of cytokines.Using a Nafion film layer on top of the graphene layer would avoid the direct contact of the sensor and chemicals in biofluid but allows the anchored aptamer biorecognition element to interact with target cytokines.This Nafion coating strategy enables their sensor to work in undiluted human sweat, and they reached a LOD of 740 fM for Interferon-gamma (IFN-) detection within 6 min.Such a strategy could be employed with nucleic acid detection to screen unwanted molecules from interacting with the sensitive sensor surface.
In addition to viral nucleic acids, another strategy is to target surface proteins.Nucleic acid targets have the advantage of being readily amplified prior to sensing which improves sensor Representative examples of sensing devices utilizing non-amplification-based detection strategies for nucleic acid detection suitable for wearable devices.A) Flexible FET sensor with conjugated DNA probes for detection of a complementary miRNA target (Reproduced with permission. [47]opyright 2020, American Chemical Society).B) A wearable aptamer-FET sensing system for detection of cytokines in undiluted human sweat (Reproduced with permission. [100]Copyright 2020, Wiley).C) Schematic overview of an ssDNA-functionalized SGGTs biosensor for detecting miRNAs in patient serum samples (Reproduced under the terms of the CC BY 4.0 license. [85]Copyright 2022, The Authors, published by Wiley VCH).D) Schematic of a graphene-based FET sensor with immobilized SARS-CoV-2 spike protein antibody for SARS-CoV-2 detection (Reproduced with permission. [101]opyright 2020, American Chemical Society).
sensitivity and are inherently specific, being able to detect single base-pair mismatch between the target and control.However, nucleic acid-based methods are generally time-consuming and need multiple purification and dilution steps since sample complexity (such as blood, saliva, and urine) can for the enzymes used in acid detection.Surface marker proteins can serve as highly specific target-binding strategies, such as antibody-antigen affinity assays and aptamers as receptors. [109]eo et al. [110] developed a label-free graphene FET (Figure 5D) which used immobilized SARS-CoV-2 spike antibody to detect the SARS-CoV-2 spike protein.Upon conjugation of the target analyte, the dynamic response of the sensor is changed which leads to an increase in the peak current of the sensor.Limit of detections of 1 fg mL − in PBS, and 100 fg mL in universal transport medium (UTM), the conventional buffer for SARS-COV-2 samples, were achieved from cultured virus samples.Also, they showed that this sensor can detect SARS-CoV-2 in clinical sam-ples with a LOD of 242 copies mL −1 .One potential future strategy could be a hybrid nucleic acid and protein approach.

Conclusion
In summary, achieving airborne viral RNA or DNA monitoring with wearables requires the integration of technologies for airborne viral particle collection, nucleic acid extraction, sensing, and reporting into a single wearable device.Recent studies have laid the groundwork for future developments in this space by demonstrating viral detection using facemask-based sensors.
However, since these detect SARS-CoV-2 in exhaled breath, their use is restricted to diagnosing infection in already infected individuals, rather than reporting on environmental exposure to viral particles.While improvements in point-of-care diagnostics are desirable, monitoring viral exposure is also a worthwhile objective since it can inform individuals of possible infection so that they can seek testing and take actions to minimize further viral spread.Improving the ability of individuals to respond to viral exposure will enhance our collective ability to combat pandemics in the future.Realizing this goal will require further improvements in the wearable aerosol collection and nucleic sensing technologies to reliably detect airborne viral particles at low environmental concentrations.
Future research on aerosol sample collection should focus on integrating current wearable collection devices with viral nucleic acid detection platforms.Miniaturization of active samplers and reductions in their collection times should also be the focus of future research.While active air samplers have been shown to achieve reliable detection with sampling times of 30 min (Table 1), reducing the sampling time to below 10 min can help enable near real-time detection of airborne viruses and their associated RNA or DNA. [75]This could allow wearers' unparalleled ability to tailor personal protective measures to their daily activities and reduce infection risk.While only active sampling strategies have currently been applied in integrated wearables for detecting airborne viral RNA, passive sampling strategies are still promising.In the future, research into passive air sampling should look at reducing the required collection time to <1 day.This would dramatically increase the practicality of passive samplers for monitoring personal exposure to viruses by informing individuals of exposure before they become infectious.They can then use this information to obtain testing and take actions that minimize the risk of infecting others.To achieve this, alternative passive collection methods like electrostatic precipitation [60] should be explored further.
Future research on wearable sensing technologies should prioritize improving the sensor's ability to utilize samples directly from patients, eliminating the need for any off-device processing.Nucleic acid-based strategies have shown high sensitivity and precision for viral detection, but their multiple-step protocols for sample preparation, amplification, buffer exchange, and signal production make it difficult to implant them in wearable devices.Recent advances in freeze-dried paper-based materials have shown good potential to overcome this problem by having multiple zones, each designed to perform different sample preparation and sensing operations.However, the high detection time of nucleic acid-based devices in comparison to other biomarkers limits their practical use in real-time monitoring of airborne viral RNA or DNA.Additionally, it is crucial to address the cost and service life of the sensing devices.Most of the reviewed sensors were designed for single use, highlighting the need for further studies on sensor reusability.Improvements in this area will enhance the practical use of wearable sensing devices for airborne viral monitoring by allowing them to respond to multiple viral exposure events.Toward this, researchers have explored regenerative cycles in biosensors. [100]However, it is important to note that single use sensors offer enhanced user convenience by eliminating the need for maintenance, avoiding drift caused by fouling, and reducing the potential for user-introduced errors.Future research on nucleic acid-based detection devices should focus on device design to enable increased sensor lifetime and automation of all unit operations needed for sensing, thereby leading to faster and more efficient viral detection.

Figure 1 .
Figure 1.Overview of the unit operations required for airborne viral nucleic acid detection: sample collection, analyte sensing, and reporting.Airborne viral particles first need to be collected from the air and lysed to extract the viral nucleic acids during the sample collection stage.The viral nucleic acids are then detected through a biorecognition process during the analyte sensing stage, typically through hybridization to a complementary probe oligonucleotide.Binding of the target analyte is then transduced into a user-interpretable signal during the reporting stage.Colorimetric and fluorescent readouts are commonly used for this purpose.Created with BioRender.com.

Figure 2 .
Figure 2. Representative examples of active (A-C) and passive (D-F) air samplers.A) Picture of IOM Inhalable Sampler.Reproduced with permission.[56]Copyright 2023, SKC Inc.. B) Portable cyclone sampler (scale added; adapted under the terms of the CC BY 4.0 license.[57]Copyright 2021, The Authors).C) Sioutas Personal Cascade Impactor Sampler with Leland Legacy pump being worn by a Cambodian poultry worker (Reproduced under the terms of the CC BY 4.0 license.[58]Copyright 2022, The Authors, published by Wiley).D) PDMS-based wearable clip sampler for monitoring personal exposure to SARS-CoV-2 (Reproduced with permission.[52]Copyright 2022, The Authors).E) Electrostatic dust collector (Reproduced with permission.[59]Copyright 2008, American Society for Microbiology).F) ElectroStatic Sampler for Air (ESSA) an electrostatic precipitation sampler (Reproduced under the terms of the CC BY-SA 4.0 license.[60]Copyright 2019, Taiwan Association for Aerosol Research).

Figure 4 .
Figure 4.Representative amplification-based virus nucleic acid detection strategies suitable for wearables A) Target specific RPA reaction in a wearable bandage-like device to detect Zika virus genomic DNA.Fluorescent dyes were used to generate a signal and were detected using a portable UV-torch (Reproduced with permission.[89]Copyright 2019, Elsevier B.V.).B) Wearable wristband device for the detection of HIV-1 DNA using a PDMS microfluidic system; RPA was employed to produce a fluorescent signal upon cleavage of dye from the primer (Reproduced with permission.[87]Copyright 2019, Elsevier B.V.).C) Microneedle wearable device for collection of cell-free DNA (cfDNA) from interstitial fluid (ISF) samples; RPA was used to amplify the target nucleotide and a ruthenium (Ru) complex was used to generate electrochemical signals from the presence of target nucleotide (Reproduced with permission.[90]Copyright 2020, Wiley).D) A lateral flow detection device coupled with RPA amplification for the detection of Plasmodium DNA (Reproduced under the terms of the CC BY 4.0 license.[84]Copyright 2015, The Authors, published by Springer Nature).E) The SARS-CoV-2 mail-insensor workflow begins by using RPA amplicons of viral RNA to turn on the Toehold Switch, which starts a cell-free reaction that produces a reporter protease.The degradation of a gelatin film by the protease is monitored using a wireless resonant sensor to transfer resonance frequency shift data.(Reproduced with permission.[83]Copyright 2022, American Chemical Society).

Figure 5 .
Figure 5. Representative examples of sensing devices utilizing non-amplification-based detection strategies for nucleic acid detection suitable for wearable devices.A) Flexible FET sensor with conjugated DNA probes for detection of a complementary miRNA target (Reproduced with permission.[47]Copyright 2020, American Chemical Society).B) A wearable aptamer-FET sensing system for detection of cytokines in undiluted human sweat (Reproduced with permission.[100]Copyright 2020, Wiley).C) Schematic overview of an ssDNA-functionalized SGGTs biosensor for detecting miRNAs in patient serum samples (Reproduced under the terms of the CC BY 4.0 license.[85]Copyright 2022, The Authors, published by Wiley VCH).D) Schematic of a graphene-based FET sensor with immobilized SARS-CoV-2 spike protein antibody for SARS-CoV-2 detection (Reproduced with permission.[101]Copyright 2020, American Chemical Society).

Table 1 .
Summary of recent active sampling devices applied to the detection of airborne viruses.

Table 2 .
Summary of viral and viral proxy sensing mechanisms suitable for wearable devices.("-" refers to non-amplification).