Electrical stimulation for therapeutic approach

Electrical stimulation as a therapeutic approach is widely applicable in terms of target tissues or target effects. This method can be an alternative to conventional therapies for patients who are resistant to drugs or are ineligible for surgical operations. In addition, as researchers have actively studied how to adjust the parameters for electrical stimulation in order to improve effectiveness, many patients have already received treatments with electrical stimulation. With respect to devices for electrical stimulation, recent studies are focused on developing reliability for safe and long‐term operations. From the point of view of engineers, a comprehensive understanding of how electrical stimulation modulates the biological system is essential to develop advanced strategies that provide effective therapeutic results. Herein, the fundamental mechanisms for delivering electrical stimulation on biological tissues are reviewed along with the requirements that need to be qualified by the electrodes. Furthermore, the latest advances in electrical stimulation devices are discussed followed by an introduction of representative applications of therapeutic electrical stimulation.

gained interest as a therapeutic approach due to its wide applicability and effectiveness.
In comparison, as a therapy, drug administration is a fundamental and conventional method to treat diseases or to support a recovery from injuries. However, drugs lack selectivity to specific lesions, which can cause unnecessary damage to other tissues. Also, there is a limitation in the use of drugs for patients who have allergic reactions or develop resistance to the drugs. For these reasons, drug administration cannot be applied to every patient for ultimate treatment. Surgeries, another conventional therapeutic method, always include some risk to take a surgical incision and to remove the lesions, which are extremely severe intrusions to the body. Sometimes, surgeries are not appropriate if the lesions are close to tissues, which control vital actions, movements, or senses. Thus, electrical stimulation therapy can overcome such limitations and be an alternative therapy for conventional methods. In particular, electrical stimulation can be applied to specific tissues to trigger target responses, which reduces side effects by enhancing selectivity. Also, the versatility of electrical stimulation facilitates the application of electrical stimulation to a broader set of patients, especially for those who are not applicable to conventional methods.
As a therapeutic approach, electrical stimulation modulates the electrophysiological mechanisms of the target tissues and brings about beneficial results. For example, electrical stimulation devices for neurological diseases, such as Parkinson's disease, or epilepsy, are already approved by the Food and Drug Administration and are commercialized. [13][14][15] These devices directly deliver electrical stimulation to the brain or the nerves connected to the brain and show proven effectiveness for reducing the symptoms, such as tremors, bradykinesia, or seizure. Also, as neuromuscular cells are composed of neurons that develop electrically communicating networks, the application of electrical pulses on these cells can bring various benefits. 16 Electrical stimulation can help the rehabilitation of partially injured nerves or block the unwanted propagation of neuronal signals. 17 In addition, electrical pulses exerted by an external device can replace the injured spontaneous pathway to deliver the commands to the muscles from the brain resulting in the recovery of motor functions. 18 Although various attempts at bioelectronics have been made to enhance compatibility, the interface at the tissue-electrode still suffers from the mechanical mismatch of the material and structural designs of the device. [19][20][21][22] To induce effective electrical stimulation on biological tissues, the devices are either attached to the surface of the tissue non-invasively or inserted into the target region for high-resolution selectivity. [23][24][25][26] As the devices are applied to various tissues ranging from the brain to the heart and the skin, the design of the structure and materials has been explored to enhance the therapeutic effectiveness of the stimulation by developing a robust interface between the device and the tissues. [27][28][29][30][31][32] Furthermore, recent research is focused on the miniaturization of the devices, which can be achieved by the replacement of batteries with a self-power generating compartment or the removal of the interconnection by applying a wireless system.
In this review, as an understanding of the basic system is essential to devise advanced strategies for effective electrical stimulation, we first discuss the mechanisms of how electrical stimulation is delivered to tissues. The electrical and mechanical requirements for the electrodes and the parameters related to the electrical pulse are highlighted in terms of the safety and effectiveness of stimulation. In addition, the configuration of electrodes with different functions for electrical stimulation is described. After the examination of basic mechanisms and requirements for electrical stimulation, various applications are introduced according to target diseases and injuries. Lastly, recent trends of research to improve safety and usability are explored through an explanation of various strategies and corresponding examples.

| Mechanism of electrical stimulation
Electrical stimulation through bioelectronics requires the electrodes to transmit the electric charge from an external source to the target tissues. When the electrode is located near the tissue, the injected charges initiate the action potentials at the neurons of the tissue ( Figure 1A, left). This process can be described as an electric circuit, which includes the electrode with resistance (R ic ), and the parallel connection of the Faradaic resistance of the electrode-electrolyte interface (R e ) and electric double layer (EDL) capacitance (C e ) ( Figure 1A, top right). 33 Specifically, when an input of electrical potential for stimulation (V sti ) is applied to the electrode, the electrical potential (V e ) is transmitted through the electrolyte, and a potential difference is created at the cell membrane of the neuron. Then, the influx of sodium ions (Na + ) through Na + channels causes the depolarization of the neurons by increasing intracellular electric potential ( Figure 1A, bottom right). 34 Therefore, the electrical stimulation triggers action potentials of the neurons and facilitates electrical stimulation to the target tissues. In the following section, we review two mechanisms of charge transfer in the electrode-electrolyte interface to understand how electric charges are transmitted from the electrode to the tissue: Faradaic charge transfer and capacitive charge transfer. [35][36][37][38] Figure 1B describes the mechanism of Faradaic charge transfer caused by the direct movements of electrons across the electrode-electrolyte interface. There are various cations and anions in the electrolyte, and they are involved in charge redistribution by participating in the oxidationreduction reaction. 39 When the negative potential is applied to the electrode, the reduction of species in the electrolyte occurs by gaining electrons. For example, water molecules are reduced to form hydrogen gas (H 2 ) and hydroxyl ions (OH − ). It is an irreversible process, and the products raise the pH at the interface, which may result in fatal damage to the tissue. Otherwise, if the electrode is positively charged, anions could be oxidized potentially causing precipitation at the electrode surface. In the case of platinum (Pt), which is an extensively used material for neural electrodes, corrosion occurs by forming platinum tetrachloride ions ((PtCl 4 ) 2− ). The corrosion of Pt electrodes is also an irreversible process and results in electrode failure. 40 Meanwhile, capacitive charge transfer is based on the charging and discharging of the EDL as a simple capacitor ( Figure 1C). EDL refers to a thin electroneutral zone divided into two layers, which is formed when the charged electrode encounters the physiological medium. [41][42][43] The first layer, called the Stern layer, consists of the ions, which adhere to the surface of the electrode. The second layer is called the diffuse layer because the ions in this area flow freely under the electrically attractive force from ions of the Stern layer. This polarization of EDL brings charge redistributions, which produces the potential difference without any electron transfer. Since the charge transfer via EDL does not involve the direct injection of electrons into the tissue, it is stable both electrically and physiologically, but it has low efficiency for electrical stimulation compared to the Faradaic charge transfer.
The mechanism of delivering electrical stimulation is decided by the material of the electrode. The electrical properties of the materials determine the way in which the charge carriers are transferred. Pt is a material, which is commonly used for Faradaic electrodes. [44][45][46] Even though it also functions as a capacitive electrode, the Faradaic charge transfer is the dominant process for Pt to stimulate the nervous system. Platinum-iridium (Pt/Ir) alloy is also a prevalent material for stimulation electrodes based on Faradaic charge transfer. The rigidity and the cost of Pt/Ir electrodes are higher than that of pure Pt electrodes. However, it is a useful material for electrical stimulation due to high stability in the electrochemical reaction. 47 For the capacitive charge transfer, tantalum/ tantalum oxide (Ta/Ta 2 O 5 ) is one of the materials for stimulation electrodes. 48 It is the most representative stimulation electrode due to the properties of biocompatibility and corrosion resistance. [49][50][51] However, because of the expensiveness of Ta, it is typically used in the form of a porous structure developed by coating the Ta metal powder on the titanium (Ti) or tungsten (W) electrode. The porosity allows a large contact area with the electrolyte of EDL, and it is expected to improve the efficiency of stimulation. [52][53][54] Typically, a capacitive charge transfer is preferable to a Faradaic charge transfer for electrical stimulation therapy in terms of safety despite low efficiency. A suitable electrode should be selected for effective and safe stimulation considering the conditions of the electrical properties or target tissue.

| Requirements for effective electrical stimulation
In the process of applying electrical stimulation to a tissue for disease therapy, several factors need to be considered. Especially for effective electrical stimulation, the appropriate conditions must be decided before stimulation to the target tissue. Here, we describe the requirements for effective electrical stimulation: (i) the electrical properties of the electrode, (ii) the mechanical properties of the electrode, and (iii) the type of the stimulation pulse.
There are three major electrical properties that determine the effect of stimulation: impedance, charge storage capacity (CSC), and charge injection capacity (CIC). 55 First, the impedance at the electrode-electrolyte interface is one of the primary electrical properties that determine the efficiency and effectiveness of stimulation. Since electrical stimulation for treatment is usually applied with alternating current (AC), it is essential to consider the impedance representing the degree of opposition to the AC. The impedance of the electrodeelectrolyte interface (Z) is determined by the R e and the C e of the EDL. Then, V e at the target tissue can be expressed as follows. 33 where σ is the conductivity of the extracellular electrolyte, r is the distance between the target tissue and the electrode, and R ic is the resistance of the electrode. Therefore, the lower the impedance and the smaller distance between the target tissue and the electrode, the more effective electrical stimulation is possible. In addition, since a higher frequency of the signal lowers the impedance, more effective electrical stimulation can be applied under a constant V sti . The CSC of an electrode represents the amount of charge that can be stored in the stimulation electrode during electrical stimulation. 36,56 The CSC can be assessed through cyclic voltammetry (CV) measurement. The CV measurements of the stimulation electrode are generally conducted in a water window, the potential range in which oxidation or reduction of water does not occur. The CV curve represents the change in current while increasing and decreasing the potential of the electrode linearly within a specific range. The area of the CV curve indicates the CSC during the stimulation pulse. 57 As the action potential in the tissue is generated by the cathodic pulse, only the CSC under the negative current flows, CSC c , is the crucial parameter to determine the effectiveness of electrical stimulation ( Figure 1D).
Another index for the efficiency of electrical stimulation is CIC. CIC refers to the amount of charge transferred from the electrode through the stimulation pulse to the tissue. 49,58 In general, the higher the CSC, the higher the CIC value. When the charge is injected through the pulse to the tissue, the voltage drop occurs at the electrode-electrolyte interface according to the amount of charge injected. Beyond the threshold voltage, water splitting occurs and electrical stimulation to the tissue becomes difficult. Therefore, a large CIC of the electrode means that the electrode can deliver more charges with a single pulse to the tissue resulting in more effective stimulation. CIC can be measured through the pulse-clamp method. 59,60 The pulse-clamp method measures the charge flowing through the stimulation electrode when rapidly switching from a galvanostatic mode to a potentiostatic mode ( Figure 1E). Since the irreversible Faradaic reaction causes a slow recoverable charge and an unrecovered charge, the CIC is measured until the threshold at which these charges begin to generate. CSC and CIC can be controlled through the shape, area, and materials of the electrode.
The mechanical properties of the electrode are also an important factor for effective electrical stimulation. 61 Recently, although various soft materials have been developed, most stimulation electrodes are still composed of rigid materials. [62][63][64][65] However, since the modulus of the tissue is much smaller than that of rigid materials, a mechanical mismatch between tissue and electrode occurs at the interface. 66 The mechanical mismatch can have several effects on the electrode-tissue interface ( Figure 1F). Also, it is difficult for a rigid electrode to conformally contact the surface of the tissue especially for two-dimensional (2D) electrodes. This reduces the area where the electrode actually contacts the tissue surface, leading to an increase in the impedance. Also, the distance between the target tissue and the electrode increases. In addition, the modulus of the electrode is also related to tissue damage and inflammatory response. When the rigid electrodes are introduced to the soft tissue, the damage is induced due to the mechanical modulus mismatch between the electrode and the tissue. Furthermore, for the chronic operation of the electrical stimulation, micromotions from the respiration and vascular pulsation bring about continuous damage, which triggers inflammatory reactions. 67 The impedance of electrodes increases due to tissue regeneration and the aggregation of glial cells around the electrodes as they hinder the delivery of electrical pulse to the tissue. 33 In this regard, mechanical modulus of material for electrodes should always be considered for three-dimensional (3D) electrodes that are inserted to stimulate tissue invasively. Therefore, it is necessary to match the modulus of electrodes to the tissue for minimal invasiveness.
As mentioned earlier, electrical stimulation for a therapeutic approach is generally applied in the form of pulses. Depending on the type of pulse applied, the efficiency or stability of stimulation may be varied. The stimulation pulse can be divided according to the shape, charge amount, and phase of the stimulus applied ( Figure 1G). One of the most important parameters is pulse phases. In terms of the number of phases, stimulation pulses can be divided into monophasic stimulation and biphasic stimulation. 36 Monophasic cathodic stimulation is a method in which only cathodic pulses are repeatedly applied. Since negative potential accumulates in the tissue by this method, more effective electrical stimulation is possible. However, the large amount of accumulated negative charges causes damage to the tissue in long-term stimulation. Biphasic stimulation applies an anodic pulse after a cathodic pulse. The anodic pulse prevents the accumulation of negative charge in the tissue. Therefore, biphasic stimulation is less efficient, but it is possible to stimulate the tissue more safely. However, biphasic stimulation may reduce the stability of the stimulation electrode. Anodic pulses can accelerate the erosion of an electrode by generating a positive potential at the tissue-electrode interface. 59 Furthermore, the parameters related to the stimulation pulse, such as the charge ratio of the anodic and cathodic pulse, stimulation time, and pulse shape, can be adjusted for the purpose of the stimulation while minimizing tissue damage, electrode corrosion, etc.
There are still several limitations in stimulation electrodes that electrically stimulate tissues for clinical KIM ET AL. applications. Therefore, if the above-mentioned characteristics of various electrodes and stimulation conditions can be optimized, it will accelerate the application for disease treatment using electrical stimulation by achieving excellent efficiency and stability.

| Electrode configuration
So far, strategies to enhance the characteristics of electrodes for electrical stimulations have been reviewed. However, utilizing only one electrode for stimulation cannot effectively and safely modulate the target tissues. In general, most of the current stimulation strategies apply a stimulation electrode system with three electrodes, which are active, reference, and ground electrodes. First, the active stimulation electrode is the electrode that delivers charges to the tissues ( Figure 2A). 68 As it is directly related to the injecting of charges into the target tissues, studies on active electrodes mainly focus on the modifications on the electrode's electrical properties to lower the impedance or to enhance the CIC. This can be achieved by designing the materials for the electrode or by modifications on the surface of the electrode. [74][75][76][77][78] Second, the reference electrode is a non-current-carrying electrode, which is used to form a voltage transient between the active electrodes. Therefore, when using a pulsed potential bias for electrical stimulation, a reference electrode is required against the active electrode. For example, in a monopolar cathodic stimulation, only the active electrode near the target tissue is set as the cathode. The reference electrode (the anode), which generates the voltage transient, is positioned at quite a distance ( Figure 2B). 69 This creates a wide electric field with a spread of stimulation in all directions. However, in bipolar stimulation, another electrode near the active electrode serves as the anode, minimizing the spread of current and yielding a narrower area of stimulation ( Figure 2C). 70 In this case, this electrode near the cathode can be denoted as the reference electrode. Lastly, the ground or return electrodes enable safe operation by extracting and removing the remaining current from the patient after stimulation. 79 Positioning the ground electrodes in an adequate position enables high-resolution stimulation as well as safe operation of the therapeutic device.
In summary, an ideal stimulation process is conducted as follows. The gradient in potential formed between two electrodes (active and reference electrodes) will generate a current flowing through the electrode to the electrode-tissue interface. After this current stimulates the target tissue, it will flow from the tissue to the return electrode, thereby leaving no current residue inside the body.
Current stimulations for therapeutic applications focus on precise and selective stimulation of the target organs. Therefore, proper placement of the electrodes is a crucial requirement for the desired stimulation. For example, for neural probes for deep brain stimulation (DBS), configurations of active electrodes such as a quadripolar electrode, multipolar electrode, and eight lead electrodes have been developed ( Figure 2D). 71,80 In the case of classic quadripolar electrodes, a stack of four Pt/Ir cylindrical electrodes is spaced either 1 or 0.5 mm apart at the distal end. These four electrodes placed inside the brain generate a continuous spherical electric field around the active electrodes with the reference electrode placed on the implantable pulse generator (IPG) positioned near the chest. 59 As the electirc field spreads outside the target area, it can cause side effects, such as stimulating unwanted tissues. Thereby, progress in electrode configurations from quadripolar to multipolar and eight lead electrodes has been made for effective stimulation. These configurations of electrodes enable bipolar stimulations with an action and a reference electrode placed near together for target-specific stimulation. For a retinal prosthesis, which requires high-resolution pixelized stimulation along the retina for high visual acuity, strategies to place the return electrodes around the stimulation electrode can be applied ( Figure 2E). 72 This surrounding of local return electrodes enables inhibition of the spread of electric field and prevents undesired stimulation to adjacent neurons (i.e., retinal ganglion cells [RGCs]). 81 Also, a big return electrode can be placed near these active and reference electrodes to effectively extract the remaining currents from the delicate retina ( Figure 2F). 73 As some stimulation strategies aim at high-resolution stimulations while others focus on applying a uniform electric field to the target, adequate electrode configurations should be applied to the design of the therapeutic device.

| Neuromodulation on brain for neurological diseases
Neuromodulation, which is the treatment of neurological diseases using electrical or other forms of external energy, is being used to treat and improve the quality of life of people with various disorders. 82 Fortunately, neuromodulation on the brain through electrical stimulation is now an alternative when other therapies to control neurological diseases, such as epilepsy, depressive disorder, and Parkinson's disease are unsuccessful or risky to the patient. 83 For example, in the case of epilepsy, ketogenic diets and administration of antiepileptic drugs are implemented. But still, thirty percent of patients do not respond to these therapies. Patients who fail to control their seizures with two or more antiepileptic medications are considered to be medically refractory, and surgery to remove the seizure onset zone (SOZ) of the brain is suggested for these patients. [84][85][86] However, if the epileptogenic zone is in charge of important functions, such as motor, language, or vision, resection of the SOZ is reluctant. In this case, electrical stimulation is performed. Neuromodulation on the brain with electrical stimulation can be categorized into DBS, responsive neurostimulation (RNS), vagus nerve stimulation (VNS), and cortical stimulation methods. Especially, various neurophysiologists have developed multichannel cortical probes or electrodes with various materials and methods to acquire more meaningful outcomes and to enhance stability and reliability. 87 3.1.1 | Deep brain stimulation DBS is a therapeutic method that applies electrical stimulation to center areas of the brain, such as the motor thalamus, globus pallidus internus, and subthalamic nucleus (STN), to control neural activities. This method is effective in treating neurological disorders as well as motor functions. [88][89][90] In clinical practice, a stimulation wire is inserted into a brain nucleus or a fiber tract within the brain, and an IPG is subcutaneously placed on the chest and is surgically connected to the stimulator. 82 The mechanisms for how electrical stimulation alleviates symptoms of neurological diseases are not fully understood. Therefore, in order to develop an effective stimulation platform by exploring the mechanism of DBS and KIM ET AL. quantifying its effect, experiments are being conducted through animal models with various devices.
Previously, rigid materials, such as silicon (Si) or W, were widely used for DBS electrodes. 91 Since these neural probes maintain their thin and long shape for long operation, they are appropriate to be implanted in the brain for the stimulation. However, the large difference in Young's modulus between these electrodes and the brain tissue causes direct tissue damage, an inflammation reaction, and scar formation, resulting in poor signal quality. 92 Therefore, flexible and soft materials are being used to develop biocompatible devices that minimize physical damage on the tissue. Kim et al. developed a polyimide-based flexible neural probe for localized stimulation and recording in the deep region of the brain ( Figure 3A). 87 Firstly, they performed simulations to determine the area stimulated with and without a ground electrode surrounding the stimulation electrode. As shown in Figure 3B, an electrode design with the ground electrode could stimulate only a limited area similar in the size of the STN, which should be targeted for Parkinson's disease. This flexible probe is inserted deep into the brain with the guidance of a metal stick. In in vivo experiments using rats, neural spike signals from the deep brain with a depth of 7 mm were recorded. Recently, as software technology has been developed, RNS that stimulates specific seizure-producing areas in a closed-loop method has been studied as an emerging approach to treat epilepsy. The RNS device continuously records neural signals with cortical strip leads located at the subdural cortex. When abnormal firings, which are epileptic precursor symptoms, are detected through realtime monitoring, electrical stimulation is applied to the epileptogenic zone to alleviate the seizure. 95

| Vagus nerve stimulation
The vagus nerve is a cranial nerve that plays a crucial role in regulating various physiological processes, including heart rate, digestion, and immune function. As another method for neuromodulation, VNS utilizes stimulation electrodes wrapped around the left vagus nerve in the carotid sheath. VNS has attracted interest as a therapeutic approach since it is relatively safe and simple to surgically implant the device without intracranial manipulation. 86 Although the mechanism has not been clarified, researchers have proposed hypotheses. One of them is that when the vagus nerve is electrically stimulated, electrical pulses are transmitted to the locus coeruleus and raphe nucleus, affecting blood flow and electrical excitability in a wide range of brain regions. 96 Numerous studies have been conducted to determine the therapeutic effects of VNS by examining behavior and immunohistochemical outcomes that correlate with different stimulation intensities. 97 Qi et al. fabricated a customized VNS device with bipolar cuff stimulation electrodes that are made of stainless steel insulated by silicone tubes ( Figure 3C). 93 Chronic stimulation with these electrodes for 2 weeks ameliorated the frequency and duration of seizures and reduced the hippocampal neuronal loss in epileptic rats as shown in Figure 3D. Although VNS is less effective than DBS and cannot completely treat disease, it is a safe approach and can be used in combination with other treatments like DBS or RNS, so it has a high potential for clinical applications. 95 Further research, however, is still required to accurately prove its effectiveness for clinical feasibility.

| Cortical stimulation
Electrodes that accompany invasive insertion into tissue trigger physical damage to the tissue, an inflammatory response, and reduced stability. 98 Therefore, devices have been developed to record electrophysiological signals and to apply electrical stimulation on the surface of the brain without penetrating the tissue. The design of the device is important to form a conformal interface minimizing the difference in mechanical properties between brain tissue and the device. Uguz et al. presented a multielectrode array (MEA) with a single neuronal resolution using poly (3,4-ethylenedioxythiophene) polystyrene sulfonate. 70 This MEA can be implanted subdural and be attached to the pial surface developing a low impedance interface ( Figure 3E). To map the neural responses to the stimulation pulses, calcium imaging was performed to create synaptic activation maps and to determine the stimulation mechanism. As shown in Figure 3F, bipolar electrical pulse stimulation produced a neuropil response in a narrower vertical area compared to unipolar stimulation. A pairwise correlation map of the electrodes before and after stimulation reveals the spatial extent of local field potential variations, which indicates the high spatial selectivity of electrodes ( Figure 3G). In addition, Park et al. proposed a system for diagnosing and treating an epilepsy in real-time on a large scale. 94 The authors developed epidural electronics to record electrocorticography and to deliver electrical pulses to the surface of the brain ( Figure 3H). The microelectrodes were fabricated using graphene to obtain low electrical noise, biocompatibility, and flexibility and were mounted on mouse cortex to modulate neuronal activities. When a graphene electrode array detects epileptic discharges, the electrical pulse are applied to stop the bursting discharges at the epileptic site, reducing epileptiform activities in frequency, spike count, and amplitude, regardless of interictal-like activities ( Figure 3I,J).
Such electronic systems for electrical stimulation can be applicable to neuromodulate neurological disorders that are sometimes uncontrollable with medication. However, the mechanism of how electrical stimulation changes network function and eventually leads to treatment is unclear. Therefore, it is crucial to identify the specific mechanisms of treatments by developing diverse types of electrodes for electrical stimulation to find a specific therapy for each disease.

| Motor recovery
Central nervous system (CNS) injuries usually include brain-related injuries and spinal cord injuries and cause long-term disability. These injuries specifically from external trauma, such as physical accidents and sportrelated accidents, have a devastating impact on voluntary control of the body and other physiological KIM ET AL.
-9 of 36 functions, further resulting in a social and economic burden on patients. [99][100][101][102][103][104] Therefore, understanding the recovery mechanism of motor and autonomic functions that regulate the circuits residing in the CNS system is essential.

| Compensating descending pathways for motor function recovery
To achieve voluntary movements, the descending pathways from the brain and brainstem to the spinal cord should be undamaged. For example, the corticospinal tract is a representative pathway for voluntary movements. According to the corticospinal tract, the cortical commands of the motor cortex are transmitted to the target motor neurons in the gray matter of the spinal cord passing through the medulla pyramids, which enable complex movements. Unfortunately, patients with CNS injury suffer from disruptions of the motor signal transmission and often have permanent disconnections of the motor descending pathway. However, this could be compensated by adoptions of different pathways due to the ability of the neuroplasticity, which occurs naturally after injuries. [104][105][106] Depending on the location and severity of the lesions, functional and anatomical reorganizations of the neuron circuits compensate for damaged pathways with alternative pathways, which are connected between intact neurons as shown in Figure 4A. 104 With local motor cortex lesions, the network plasticity occurs in perilesional areas to replace the motor functions of injured region ( Figure 4A1). When the lesion is positioned on one side of the hemisphere resulting in the partial loss of functions, the contralesional cortex usually replaces the function of the ipsilesional cortex with new connections between cortico-other side cortical pathways ( Figure 4A2). For the complete loss of functions of an injured region, the contralesional cortex directly controls the ipsilesional motor functions ( Figure 4A3). In particular, there are also other descending pathways to achieve body movements primarily for locomotion and postural control, such as the rubrospinal and reticulospinal tract, which originate from brainstem projections to the spinal cord. When the lesion is positioned at brainstem pyramids, the reticulospinal and rubrospinal tract contribute to mediate volitional body movements by forming cortico-reticulo/rubro-spinal pathways ( Figure 4A4). In addition, the majority of the reorganized motor circuits spares propriospinal interneurons, which communicate information across the different spinal cord segments and within the spinal cord to achieve complex movements. In this case, the propriospinal interneurons can contribute to indirect compensation pathways between other segments when the lesion is positioned at a dorsolateral funiculus with partial damage ( Figure 4A5). When the lesion is severe such that hemisection of the spinal cord is required, the corticospinal upper neuron axons from the ipsilateral cortex sprout across the midline within the spinal cord for motor function recovery ( Figure 4A6).
3.2.2 | Electrical stimulation for enhancing neuroplasticity Current treatments for motor function recovery include neurorehabilitation, pharmacological, and electrical neuromodulation. Specifically, electrical neuromodulation promotes the neuroplasticity in the brain, brainstem, and spinal cord, enhancing intact descending circuits primarily associated with motor functions as shown in Figure 4B. 107 For example, electrical stimulation of cortical circuits with non-invasive or invasive electrodes can strengthen the corticospinal tract projections between the brain and spinal cord, enhancing locomotor movements. Bonizzato et al. aimed DBS of the brain region to facilitate volitional walking and other locomotor movements ( Figure 4C). 108 They implanted commercial 16-DBS electrodes into the midbrain locomotor region, especially the pedunculopontine nucleus, which commands basic locomotion. Increasing the stimulation frequency, amplitude, or pulse width improved locomotion speed, such as stepping and walking, which means the enhancement of the supraspinal command growth by cortical circuits stimulation. However, depending on the severity of the CNS injury, all supraspinal pathways could be interrupted. In this case, the spinal cord replaces the execution center. Thereby, electrical stimulation targeting different spinal cord segments with epidural, intraspinal, and transcutaneous electrodes can generate various muscle contractions achieving body movements, such as upper or lower limb functions without supraspinal commands. Recent studies showed that epidural electrical stimulation (EES) indirectly activates motor neurons by proprioceptive afferents within the dorsal root of the cervical and lumbar spinal cord, which restores the activation of paralyzed muscles. Barra et al. conducted EES of cervical spinal cord in monkeys resulting in voluntary arm and hand movements by restoring survived spinal circuits ( Figure 4D). 109 EES at different spinal cord segments generates different muscle recruitments, such as a C8/T1 level for hand and forearm muscles, C7 level for triceps, and C5/C6 for biceps and deltoids. With this muscle selectivity, EES enables the monkey's arm to reach, pull, and grasp. Moreover, EES with high-frequency improves muscle strength and movement quality. Wagner et al.
introduced EES of the lumbosacral spinal cord, which allows patients suffering from incomplete and complete spinal cord injuries to enable voluntary control of walking ( Figure 4E). 110 16-electrode paddle array delivered selective stimulation when intended movements are performed. As a result, within 1 week, patients can control the paralyzed lower muscles during rehabilitation with EES. After a few months, they can voluntarily walk or cycle with better performances without EES due to the reconstructed corticospinal pathways between the brain and spinal cord.
Electrical stimulation has been widely applied to improve motor recovery functions in neuromodulation for current clinical treatments. However, many patients still have some difficulties in performing voluntary movements and complex movements, such as regulating their hands and controlling objects. Therefore, electrical stimulation of the brain, brainstem, and spinal cord combined with neurorehabilitation is essential for recovery of motor functions and the enhancement of performances, which maximizes the potential synergies.

| Nerve regeneration
Peripheral nerves are located outside CNS to bridge the CNS and rest of the body. Peripheral nerve injuries (PNI) can cause chronic pain and disability, and 2.3% of extremity trauma patients suffered from PNI in the United States. Unfortunately, peripheral nerve is regenerated by approximately 1 mm per day, which means that the injury can last several months depending on the severity. 111 In addition, experiments show that motor capacity decreases to 35% of the normal if the regeneration of the axotomy is delayed more than 3 months. 112 Therefore, efficient strategies to accelerate regeneration of peripheral nerves are significant for patients undergoing PNI.
To assist the regeneration, hollow tube-like nerve guidance conduits (NGCs) have been commonly used in preclinical studies to bridge and repair the proximal and distal ends of the injured nerve by a nerve regenerating process shown in Figure 5A. NGCs can give advantages, such as providing a pathway that prevents collateral regeneration or blocking myofibroblasts, which can suppress neurite outgrowth. [117][118][119] Various approaches for NGCs have been introduced to efficiently recover the injured peripheral nerves. 120 Among these various types of NGCs, NGCs using electrical stimulation to accelerate nerve growth are one of the most promising approaches. Although the precise mechanism is still unclear, one of the hypotheses is that electrical stimulation affects calcium activity in the cellular membrane, which plays a significant role in cell growth. [121][122][123] Encouraged by the electrical stimulation, the upregulated brain-derived neurotrophic factor elevates the cyclic adenosine monophosphate pathway by a calcium-dependent pathway, which eventually promotes cellular growth by intracellular steps.
Since high electrical conductivity is required for NGCs to effectively stimulate the axotomy, recent studies added various conductive materials to the scaffold. Zhang et al. presented conduits based on poly(ɛ-caprolactone) (PCL) and carbon nanotubes (CNTs). 113 The conduit is fabricated by electrospinning PCL at an optimized rotational speed of 1000 rpm and adding CNTs to create PCL fibers with multiwalled CNTs. The PCL/CNTs composite fibers exhibit high fiber anisotropy and electrical conductivity, facilitating the regeneration of myelin and axons in vivo under electrical stimulation ( Figure 5B).
Stimulating the injured nerve wirelessly is another significant approach in preclinical studies to prevent additional damage to the surrounding tissues, ensure safety, and secure free body movement from the invasive power source. Fang et al. demonstrated a magnetoelectric four-dimensional (4D) stretchable nerve conduit, a combination of 3D printing technology with a high-frequency magnetic field to remotely induce an electric current. 114 A 3D printed stretchable microchannel conduit includes reduced graphene oxide nanosheets sandwiched with a silica shell of an electromagnetized porous carbon layer, termed as nanocookies ( Figure 5C). Under the magnetic field, the conduit can release neuron growth factor and stimulate the nerve electrically to facilitate the nerve regeneration remotely, while the microchannel allows for an effective alignment of cells ( Figure 5D).
The removal of the conduit can be troublesome since it requires additional surgery for extraction, eventually damaging the surrounding tissues. Therefore, conduits that dissolve after the proliferation are one of the desires in the current field of PNI restoration. Wang et al. reported a self-electrifying scaffold that is fully bioresorbable to aid in peripheral nerve regeneration ( Figure 5E). 115 The conduit consists of poly(l-lactic acid) and poly(trimethylene carbonate) (PLLA-PTMC) as the inner layer and porous PCL as the outer layer. PLLA-PTMC serves as a nerve interface, while porous PCL not only supports the scaffold mechanically but also ensures nutrient permeability. Also, the electrodes on either side of the conduit serves as a galvanic cell with magnesium (Mg) as anode, iron-manganese alloy (FeMn) as cathode, and body fluids as the electrolyte, which can self-electrify and generate electric field. The bioresorbable conduit enables the proliferation of axons with some cells aligned in parallel to the electric field ( Figure 5F). Choi et al. also presented a fully bioresorbable wireless stimulating device that is composed of three main parts: antennas, electrodes, and cuff. 116 The antenna has a dualinductive molybdenum (Mo) coil with interlayers of bioresorbable dynamic covalent polyurethane (b-DCPU) and a diode of monocrystalline Si. The serpentine design of the Mo electrode ensures stretchability, while poly (lactic-co-glycolic acid) (PLGA) is used for the stimulation cuff. After the implantation surgery of the device, the primary coil activates the stimulation outside the body to facilitate the nerve regeneration, and the device eventually dissolves completely after a few days (complete degradation after 50 days at 90°C in PBS, pH 7.4) ( Figure 5G).
Peripheral nerve regeneration by electrical stimulation is a widely known and researched area with various designs and materials. However, an accurate understanding of the mechanism is still unknown, which causes inefficiency in terms of electrical stimulating strategy or the materials and designs of the device. Therefore, the precise mechanism needs to be unveiled for further development of desirable devices and to broaden the clinical use of electrical stimulation in nerve regeneration.

| Pain block
Pain refers to an unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage. 124 Among the various ways to treat pain, opioids are effective drugs that can be administrated to the patients with severe pain. However, several side effects, such as nausea, euphoria, and respiratory depression, can occur, and long-term use of opioids may cause drug tolerance, resulting in death. 125 Moreover, the opioid abuse problem is becoming more severe with an increasing number of deaths and economic costs due to opioid overdose in the USA. 126 Therefore, various other methods, such as ultrasound therapy, light therapy, thermal modulation, and electrical stimulation, are being studied as safe and effective ways to treat pain by blocking pain signals to prevent propagation to the brain. [127][128][129][130] Among these methods, electrical stimulation that directly targets and stimulates the nerve has emerged as one of the effective ways to treat pain. Although the exact principle of how applying electrical stimulation can block the propagation of pain signals is still unknown, several methods are used to deliver electrical current stimuli to the nerves and to block the pain signals from propagating. This technique is called "nerve conduction block" and can be divided into three different methods, according to the shape and the location of the electrodes that are used to deliver the electrical stimulation ( Figure 6A).
Firstly, the transcutaneous electrical nerve stimulation (TENS) method is the most commonly used method due to the convenience of these non-invasive electrodes. 134 Generally, four electrodes are attached near the region where the actual pain occurs, and electrical currents are delivered through the electrodes ( Figure 6B). However, despite the strong advantage of non-invasiveness, the electrical current must pass through a long distance of the skin, muscle, and fat layers to reach the target nerves, which hinders effective stimulation.
To overcome these limitations, invasive electrodes are used for the percutaneous electrical nerve stimulation (PENS) method. Electrodes are inserted into the skin and placed in proximity to the targeted nerves. Although this method can block pain signals from propagating through the nerve more effectively than the TENS method, the invasively inserted electrodes may cause temporary pain when being inserted, and tissue disruption problems may occur due to the rigidity of the electrodes. In addition, inaccurate placement of electrodes may damage the nerves, and therefore, an ultrasound guide is commonly used to help with the accurate placement of the electrodes, which makes the treatment procedure complicated ( Figure 6C). 131 Lastly, cuff-shaped electrodes are implemented by attachment directly to peripheral nerves ( Figure  6D). 132,135 As the electrodes cover the peripheral nerves, the distance which the electrical current passes reduces significantly. This enables the blockage of signal propagation within a few milliseconds, being much shorter than TENS or PENS. However, since the electrodes must be directly attached to the target nerves, the attachment process of the electrodes must undergo a surgical procedure, and the electrodes may cause immune reactions to the surrounding tissues due to the modulus mismatch between the electrodes and the tissues. In addition, the electrodes must be removed after applying electrical stimulation, which causes a complicated second surgical procedure.
Lee et al. developed fully bioresorbable peripheral nerve stimulating electrodes that can be attached to the nerves and used for applying electrical stimulation to induce a nerve conduction block ( Figure 6E). 133 Since all components of the nerve stimulating electrodes were composed of bioresorbable materials, the electrodes were fully dissolved in an in vitro environment, which was set as similar to the physiological environment over a period of months. As a result, no second surgical procedures were needed to remove the electrodes after the treatment of pain, overcoming the disadvantages of the original cuff-shaped electrodes used for delivering electrical stimulation. In addition, in vivo experiments were conducted to evaluate the nerve conduction block capability of the nerve-stimulating electrodes. Two different Pt electrodes were attached to the sciatic nerve, which was used as a stimulation electrode to induce action potentials from the nerves, and a recording electrode to record neural signals from the nerve, respectively. The bioresorbable nerve-stimulating electrodes were placed between the stimulation and recording electrodes to conduct the nerve conduction block ( Figure 6F). The amplitude of compound nerve action potential signals was recorded for 10 days with electrical stimulation on the sciatic nerve. Complete nerve conduction block was induced even 9 days after the implantation of electrodes ( Figure 6G).
Applying electrical stimulation directly to the peripheral nerve is an effective way to control pain as mentioned above. However, the exact mechanism of how 14 of 36 electrical stimulation can block the propagation of pain signals is still unclear. For the development of a more effective design of electrodes, the exact mechanism must be studied. In addition, among several methods of delivering electrical current to peripheral nerves, using cuff-shaped electrodes for inducing nerve conduction block is a promising method despite the critical disadvantage of requiring second surgical procedures to remove the electrodes, and studies that can overcome these disadvantages are being conducted.

| Cardiac pacemaker
The heart generates electrical impulses from cardiomyocytes that can beat themselves, and when the impulses are transmitted to heart muscle cells, the heart muscle repeats contraction and relaxation to supply blood necessary to each organ and tissue. In detail, the cardiac impulse (arrows) occurs in the sinoatrial node (SAN), transmits across the atrial myocardium, and moves toward the atrioventricular node (AVN) ( Figure 7A). 136 Then, from the AVN, electrical impulses travel toward the His bundle and bundle branches. The activation of Purkinje fibers, which are the terminal of the bundle branches, causes the activation of the ventricular myocardium. By repeating these processes, the heart repeats contraction and expansion. However, if the SAN does not produce enough electrical impulses (sinus insufficiency syndrome) or if the electrical impulses do not flow smoothly through the conduction system (atrioventricular block), the beating of the heart becomes slower. These arrhythmias are called bradycardia. In bradycardic arrhythmia, the number of times the blood is pumped decreases, resulting in dizziness, shortness of breath, and in severe cases, loss of consciousness and fainting. If these symptoms occur, it may be necessary to implant an artificial pacemaker to replace or supplement the heart's natural role of pacing.
An artificial pacemaker is a medical device that is implanted in a patient with a slow-beating or non-beating heart and adjusts the heart rate to meet the patient's metabolic needs. 142 It mainly functions to detect the patient's heart rate and to apply appropriate electrical stimulation when bradycardia progresses. The pacemaker consists of a main body and electrodes. The main body is implanted under the skin of the patient's left rib and is connected to the electrodes on the patient's heart. However, due to the invasiveness and poor contact of commercial pacemakers on the heart, numerous studies have found the most suitable pacemaker with living tissue.
Park et al. developed an epicardial mesh based on silver nanowires and a conductive rubber. 137 The epicardial mesh contacted conformally on the beating heart in vivo and detected myocardial activities consistently, covering the entire ventricular chambers ( Figure 7B). To monitor the effectiveness of the electrical stimulation by the epicardial mesh, representative electrocardiogram (ECG) waves were measured in normal (control) and postmyocardial infarction (Post-MI) induced rat models. As shown in Figure 7C, the epicardial mesh pacing (MeshP) caused a rise in the QRS complex in both the control and Post-MI-induced model. Remarkably, the MeshP in a Post-MI-induced model showed a similar QRS duration with the ECG wave from the control without MeshP. In addition, this epicardial mesh could treat ventricular fibrillation with a biphasic electrical shock of 2 J ( Figure 7D), demonstrating the clinical potential for cardiac pacing and defibrillation. Gutruf et al. demonstrated a miniaturized cardiac pacemaker for small animal models ( Figure 7E). 138 Rather than covering the entire heart, cardiac pacing was induced by intensively stimulating the left ventricle. Similarly, Sim et al. attached a rubbery epicardial patch with only two pacing electrodes made of silver nanowires, conductive rubber paste, and polydimethylsiloxane to the right ventricle of a porcine heart ( Figure 7F). 139 When pacing the right ventricle electrically, the heartbeat increased from 64 to 120 bpm as shown in Figure 7G. Therefore, this indicates that bradycardia can be treated with only two electrodes in cardiac pacing.
In addition, there is a treatment of the electrodes' surface to make cardiac stimulation more effective. Sunwoo et al. improved the stimulation performance of the device by incorporating Pt nanoparticles. 140 They proposed stretchable epicardial mesh electronics with silver-gold nanowires, Pt nanoparticles, and rubbery elastomer ( Figure 7H). To lower the impedance and improve the CSC, which is related to the charge injection limit, silver-gold nanowires were mixed with Pt nanoparticles ( Figure 7I,J). As a result, the charge injection limits highly increased by 8.4 times (0.61-5.12 mC cm −2 ). The electrically enhanced nanowire composites are useful for stimulating both the left and the right ventricles of the rat heart. Heartbeats (black arrows) were paced via electrical stimulations (red arrows) with a pulse of 6 Hz, 1.5 V, and 1 ms ( Figure 7K). Likewise, there is a cardiac treatment platform with Pt nanoparticles, enabling diagnosis and simultaneous stimulation of cardiovascular diseases. As the beating of heart can be detected as a pressure changes, pressure sensor facilitates the examination on the condition of the heart. 143-145 Hwang et al. presented an epicardial patch that consisted of a pressure sensor array and pacing electrodes. 141 To prove the effectiveness of bradycardia treatment, they induced the bradycardia (86 bpm) in a rabbit and attached the patch to the left ventricle of the rabbit's epicardium. Then, they stimulated the left ventricle with electric pacing while detecting pressures of the epicardium (red line) and surface ECG (black line) in Figure 7L. After electrical pacing, the rabbit recovered to a normal heart rate of 130 bpm on average.
Based on the many studies described above, the efficacy of electrical stimulation through electrodes in the field of artificial pacemakers is well known. In vivo experiments have shown that by applying electrical stimulation, many devices can provide therapeutic solutions for arrhythmias, such as bradycardia and ventricular fibrillation. In addition, since several materials that interact well with cardiac tissue are being developed, the development of a minimally invasive and miniaturized pacemaker is expected in the future.

| Wound healing
The skin is a large organ that covers the entire body and is a complex barrier protecting the body from mechanical/chemical irritants. 146 Furthermore, the skin maintains body temperature by controlling sweat secretion. pressure sensor array and pacing electrodes. Heart rhythm was restored after electrical pacing (0.6 V mm −1 , 2.2 Hz, 1 ms). Reproduced under terms of the CC-BY-NC license. 141   The skin consists of three layers, the epidermis, the dermis and the hypodermis, and all layers vary in their structure and anatomy. 147 The epidermis is the outermost layer that is subdivided into five layers (stratum corneum, stratum lucidum, stratum granulosum, stratum spinosum, and stratum basale). The epidermal cells consist of keratin, a non-water-soluble protein, which protects the body from the external environment. The dermis is composed of stratum papillare and stratum reticulare. And the dermis contains blood vessels, muscles, sensory neurons, and lymphatic systems. 148 Lastly, the hypodermis is located beneath the dermis and is called subcutaneous fascia. Hypodermis connects the dermis to muscles and includes nerves, sweat glands, and hair follicles.
Skin wounds refer to skin damage where the normal anatomical structure and function of the skin is impaired. Skin wounds are classified into acute wounds, surgical wounds (minor burns, accidental abrasions, etc.), and chronic wounds (ulcers caused by diabetes or venous stasis disease, etc.) ( Figure 8A). 153 Wound healing is a complex process that involves interactions between various immunological systems. 154 Generally, the normal wound healing process is divided into four phases: hemostasis stage, inflammation stage, proliferation stage, and remodeling stage. 155 Hemostasis begins within seconds after the initial insult. The constriction of blood vessels and aggregation of platelets to the injured site prevent further bleeding by clotting. The inflammation stage begins with the influx of phagocytic cells. Phagocytic cells such as macrophages remove damaged or dead cells with bacteria and pathogens from the wound. In the proliferation stage, angiogenesis and collagen synthesis occur. Subsequently, regenerated collagen is realigned along the wound site in the remodeling stage. 156 However, depending on the wound, excessive healing can result in the formation of keloid scars or deficient healing can result in an incomplete deposition of new tissues. 157 Various factors interfere with the wound healing process, resulting in chronic wounds that are not fully repaired. Therefore, it is important to heal wounds appropriately to address these problems.
In this section, we introduce electrical stimulation for wound healing. Intact skin generates a range from 10 to 60 mV of transepithelial potential (TEP) called the skin battery. 158 When the Na + and Cl − channels are in the apical plasma membrane and the Na + /K + -ATPase are in the basal plasma membrane, TEP is generated by this distribution of ion channels in epithelial cells ( Figure 8B). 159 If the epidermis is injured, the TEP of a wound site drops and the site becomes a cathode. Then, a direct current (DC) flows from the surroundings to the wound and then out from the wound ( Figure 8C). 157 As ions are transported to maintain TEP, endogenous wound electric field continues to drive until the wound heals. 160 Electrical stimulation has been proposed as a directional cue that changes the intensity of TEP. There are studies showing that electrical stimulation is effective in both acute and chronic wound healing, and various materials and applications have been developed. [161][162][163][164][165][166][167] Electrical stimulation for wound healing uses milliampere levels of current (or millivolt levels of voltage) through two electrodes located on the wound site or surroundings. 149 Two types of current are conducted, which are DC (e.g. constant DC, balanced monophasic pulse, unbalanced monophasic pulse) and AC (e.g. symmetric AC, asymmetric AC). In the DC stimulation case, the cathodic electrode should be placed on the wound site, while the anodic electrode should be placed on the surrounding normal skin ( Figure 8D). In the AC stimulation case, both electrodes should be located on normal skin around the wound ( Figure 8E). DC stimulation has a higher risk of thermal burns than AC stimulation. Both stimulation types allow deeper current penetration into the skin with a higher frequency of stimulation, but DC stimulation has a higher risk of thermal burns than AC stimulation. 168 Jiang et al. developed the smart bandage comprised of an integrated sensor and stimulator for wound care. 150 All electrical components were integrated into a flexible printed circuit board ( Figure 8F). This smart bandage can be attached to wound surfaces with a hydrogel interface. A 6-mm-diameter wound on the dorsum of mouse was created and was stimulated using AC with the voltage oscillating between 0 and 2 V for 6 h per day ( Figure 8G). The stimulation accelerated wound healing and the impedance of the wound increased, meaning a return to an uninjured state. In addition, Jeong et al. presented a fiber ionic triboelectric nanogenerators (iTENG) patch composed of an organogel-filled microtube ( Figure 8H). 151 This device was composed of a woven iTENG, an ionic wire and an ionic patch. When the iTENG contacted the skin, contact electrification occurred. And electrostatic induction occurred when the iTENG moved away from the dermal surface. This triboelectricity can generate AC current to stimulate the wound. The ionic patch was placed on the 1 � 1 cm wound area of mouse. Electrical stimulation was performed every 2 days to the Gel-TENG group through the iTENG patch. After 14 days, the stimulated group (Gel-TENG) had a significantly decreased wound size ( e5%) compared with other groups ( Figure 8I). Likewise, Liang et al. constructed a ring-type electrode with carbon fiber to deliver electrical stimulation on the wound site. 152 The carbon fiber ring (anode) was attached to the surrounding area of the wound site with adhesive film and a needle-type electrode (cathode) located 18 of 36 vertically at the center of the wound. The optimized DC pulse electric field (70% duty, 1000 Hz cycle) stimulated the wound under vacuum. Exerted electric fields were detected with uniform patterns in the wound edge (198.63 � 17.91 mV mm −1 ). To validate the wound healing performance, the device was attached to a 3-cm-diameter wound surface of a Bama miniature pig model. 100 mV mm −1 electric field was applied to stimulate accelerate wound healing. After 14 days of stimulation, the wound area of the stimulated group (+100 mV mm −1 , black) becomes about 6 times smaller than that of the control group (Control, blue) ( Figure 8J). And a −100 mV mm −1 stimulated group (−100 mV mm −1 , red) showed about two times larger wound area. This result represented that stimulation in the opposite direction decreased with healing progress.
Furthermore, hydrogels have been studied for optimal electrical stimulation in terms of materials. Lei et al. showed an adaptive conductive hydrogel, including tannic acid, human-like collagen into polyvinyl alcohol, and borax hydrogel dynamic crosslinking networks. 169 This hydrogel was not only suitable for use in the cavity of deep wounds but also permitted endogenous and external current conduction. Mao et al. also presented hydrogels based on bacterial cellulose and MXene for accelerating the skin wound healing process under external electrical stimulation. 170 As above, electrical stimulation is an effective way to accelerate wound healing. However, optimal stimulation methods, such as frequency, duration, and potential for each body part, are worth further investigation. In addition, from a device-perspective, wireless communication and persistence of attachment on the wound are also challenges to note. If we continue to deal with these challenges, electrical stimulation will become an irreplaceable therapeutic system for wound healing.

| Wireless power transfer
Most bioelectronic devices for electrical stimulation currently utilize a battery-powered system, which accompanies the lead wire for electrical connection. As wireless power transfer (WPT) technology has developed rapidly, the application of a WPT system to devices for electrical stimulation is also being actively promoted. 171 Wireless, battery-free electrical stimulation devices offer a number of advantages for practical applications. For example, these devices are generally small in size, which enable the placement in limited target areas. [172][173][174] Also, since it is lightweight and used without a lead wire, user's inconvenience can be greatly reduced. WPT technology used in devices for electrical stimulation can be classified into radio frequency (RF), near-field communication (NFC), and ultrasound ( Figure 9A). RF, the traditional method for WPT, operates by radiating electromagnetic waves from the transmitting antenna to the receiving antenna. NFC is included in the category of RF and is operated through inductive coupling at a frequency of 13.56 MHz. WPT technology through ultrasound refers to a method in which a piezoelectric material receiving ultrasound generates power.
These WPT technologies for electrical stimulation have been studied targeting various parts of the body, such as the heart, muscle, brain, spinal cord, and peripheral nervous system. For example, Choi et al. developed a wireless cardiac pacemaker, which is externally controllable and programmable ( Figure 9B). 175 For the wireless electrical stimulation, the receiver coil of the device transforms the received RF power (13.56 MHz) to a DC pulse through the RF diode. Subsequently, a DC pulse is delivered to the electrode pads for electrical stimulation of myocardium. This device showed effective wireless cardiac pacing in various in vivo models, including rat and canine models. Similarly, Ausra et al. introduced a battery-free device with on-board computation, which is capable of heart rate detection and cardiac stimulation ( Figure 9C). 176 For WPT, magnetic resonant coupling occurs at a frequency of 13.56 MHz between two coil antennas. Also, temporary energy storage was used to maintain the energy efficiency of the device. The authors conducted an in vivo demonstration with a mouse model by performing simultaneous wireless electrical recording and stimulation. Choi et al. presented the wireless electronic stimulators, which can facilitate regeneration of PNI ( Figure 9D, left). 116 RF power with frequency of 14.8 MHz was supplied to the receiver coil antenna of a device by magnetic coupling (Figure 9D, right). The obtained RF power was transformed into low frequency (20 Hz), cathodic, and monophasic electrical impulses for nerve stimulation. Similarly, Guo et al. also proposed a wireless device for peripheral nerve stimulation, which is also driven by RF power with the resonance frequency of 18 MHz. 180 Burton et al. developed a battery-free platform for stimulation of the brain ( Figure 9E). 177 The device is operated by magnetic resonant coupling at 13.56 MHz. The transmitting antenna is attached to a cage for powering the implanted device in a freely moving rodent. To prevent an overvoltage during energy transfer, a half-bridge rectifier with a Zener diode was used. The transferred energy is then changed to biphasic pulses for electrical stimulation. Huang et al. proposed a wireless and implantable microneedle device for skeletal muscle regeneration ( Figure 9F). 178 The RF KIM ET AL. power transmission system is composed of a double layered antenna and doped Si diode. The resonance frequency for WPT was 6.6 MHz and the rectified output voltage was controlled. The authors conducted the wireless RF-based electrostimulation to C2C12 cells for regeneration of muscle.
In addition to an antenna-based WPT, there is a method of wireless powering through ultrasound. For example, Zhang et al. introduced an implantable piezoelectric device based on a Sm-doped Pb(Mg 1/3 Nb 2/3 )O 3 -PbTiO 3 (Sm-PMN-PT) single crystal for DBS treatment ( Figure 9G). 179 The piezoelectric materials can convert ultrasound energy into electricity for WPT. As the ultrasound reaches the implanted device, the piezoelectric materials generate electrical energy and subsequently electrical DBS is conducted through the stimulation electrode. The authors conducted behavioral experiments with a rat model and demonstrated the capability of wireless DBS.
As mentioned above, the conventional electrical stimulation devices limit the user's behavior due to the bulky nature of the devices. Wireless and battery-free devices with various WPT technologies can alleviate these issues. [181][182][183][184] In addition, a wireless device has a great advantage for observing animal behavior in preclinical trials because there are no restrictions on animal movement after implantation of the device. These improvements with wireless technologies are expected to greatly contribute to the practical use of future implantable devices.

| Energy harvesting by nanogenerators
As the demand for state-of-the-art personalized healthcare grows, the nature of powering these healthcare electronic devices remains as one of the key challenges that needs to be solved. 185 As an attempt to address this issue, engineering works are done to harness mechanical energy to electrical energy. Especially in the field of electrical stimulation therapy where a steady power supply is much needed, most electronic devices must be integrated with methods of efficient mechanical-toelectrical energy conversion to harness human biomechanical energy. In this prospect, piezoelectric nanogenerators (PENG) and triboelectric nanogenerators (TENG) are energy-harvesting technologies that have attracted significant attention due to their simple engineering structures and seamless integration in modern wearable devices. The therapeutic applications of such nanogenerators include plasma treatment, behavior intervention, and neuromodulation of the muscles. [186][187][188]

| Piezoelectric/triboelectric effect
The basis of PENGs rests on the principle of the intrinsic piezoelectric effect of the piezoelectric materials. The piezoelectric potential, which is produced by the presence of an electric dipole in the materials under strain, is the basis of the PENG's operating principle. A PENG comprises two electrodes that are bound to the piezoelectric material on both sides. In PENG, the electric dipoles in the piezoelectric material align in one direction whenever a mechanical force is applied due to the created potential difference between the electrodes via internal polarization. After the removal of the external force, an uncharged state is recovered in PENG. This mechanical action causes current to flow through an outer circuit from the top to bottom of the electrode. Subsequently, the electrical dipoles reverse that causes a reverse current flow after force is released. The schematic mechanism of PENG is shown in Figure 10A. Similar to PENG, TENG has similar potential to power wearable electronic devices for therapeutic applications. TENG devices can convert mechanical energy to useful electrical energy in a process of known as triboelectrification. During the triboelectrification process, an electrical potential is generated between the surfaces where the charges are separated on each contact surface of TENG. In this sense, the performance of TENG is proportional to the change in charge density of the opposite contact surfaces. 197 The repetition of mechanical forces acting on TENG creates the generation of alternating potential, which can be harnessed to power wearable electronic devices. The working mechanism of TENG is shown in Figure 10B. 189 To illustrate the structure and working principles of TENG, the general idea is to utilize sandwiched planar thin films of two different materials stacked with or without an interlayer. For the generation of power, an external force deforms the sandwiched films, which causes friction and generation of opposite electrostatic charges. This creates an interface dipole layer (i.e., triboelectric potential layer), which does not dissipate and neutralize instantly. By using common electrodes, TENG is connected to an external circuit and the electrostatically +100 mV mm −1 stimulated (+100 mV mm −1 , black), −100 mV mm −1 stimulated (−100 mV mm −1 , red) groups. *p < .05, **p < .01. Reproduced with permission. 152 Copyright 2020, Elsevier. AC, alternating current; DC, direct current; iTENG, ionic triboelectric nanogenerator; TEP, transepithelial potential. induced free charges are collected to produce power for the electronic devices ( Figure 10C). 190

| Development of the design structure and materials
To increase charge generation and boost efficiency in selfpowered wearable devices, previous works reported a large contact area or enhanced a surface structure to boost the generation of free charges. Most common devices have typical 2D, thin, and flexible device designs that generate significant power with the onset of a perpendicular force. As shown in Figure 10D, Fan et al. demonstrated a self-powering device using flexible polymer sheets for a cheap, light-weight, and durable energy harvester. This device design was optimized to obtain maximum output voltage and current signal without the consideration of application to a wearable device platform. Another aspect of development is designing an improved surface interface to boost the generation of dipole charges for greater power generation. For example, Zhu et al. designed a substantially high-power output nanogenerators enabled by a nanoparticle-based surface modification ( Figure 10E). 191 This nanogenerator was able to power hundreds of electronic devices by just a trigger of commonly available motions. By using such modified design, this device was able to power 600 commercial light emitting diode (LED) bulbs in real time. However, to address the dynamic motion of the body, bioelectronics must be flexible or, in some cases, stretchable. Likewise, the critical components of electronic devices, such as nanogenerators, must follow this mechanical property to prevent device failure. He et al. demonstrated a fiber-based TENG for advanced structural design for power generation. 192 This can be operated while being stretched at a high strain value up to 70% ( Figure 10F). This nanogenerator has produced enough power to devices, such as commercial capacitor and digital watches.
To apply this energy-harvesting technology to wearable devices, research has been conducted to harness the dynamic mechanical energy of the human body in a method that is seamless and harmless in nature, while not compromising the operational performance of the device. To achieve this, novel sustainable and shapeadaptable nanogenerators have been developed that are equipped to withstand a great level of external stress while ensuring user safety. Wu et al. demonstrated a skin-like liquid single-electrode nanogenerator, which incorporated NaCl that significantly increases conductivity ( Figure 10G). 193 This nanogenerator not only demonstrated good endurance to mechanical deformation as a wearable platform but also showed good comfort and safety characteristics. This nanogenerator can be applied to facial surfaces due to its high sensitivity, providing a promising prospect of a subtle energyharvesting component for advanced therapeutic devices.

| Trending applications of nanogenerators in healthcare therapeutic devices
The inclusion or dependable transmission of electrical power presents a significant problem for medical bioelectronics. To alleviate this issue, nanogenerators are constantly integrated into such clinical bioelectronic devices. Especially in implantable electronics, the wired transmission of power causes secondary issues, such as tissue inflammation and patients' constraints due to their restricted motion. In light of this problem, Lee et al. demonstrated the use of ultrasound to transmit mechanical energy through the skin layers and internal liquids and showed the effectiveness of an ultra-thin implantable vibrating nanogenerator ( Figure 10H). 194 By ultrasound-mediated methods, they demonstrated a tunable transient performance in an ex vivo experiment using a porcine tissue with performance voltage reaching 4.16 V.
Another key feature of using nanogenerators is the concept of stimulus-responsive driven device activation. As an example, Yao et al. demonstrated a battery-free implantable VNS system that responds impulsively to stomach movement ( Figure 10I). 195 This system comprises of a flexible and biocompatible nanogenerator that is implanted at the surface of the stomach, generating biphasic electric pulses in response to the peristalsis of the stomach. The generated electric pulses stimulate the vagal afferent fibers that promote the reduction of food intake and allow patients to modulate their weight effectively. This work demonstrated the deployment of nanogenerators in therapeutic technology that produces artificial nerve signals from coordinated body movements. Similarly, Ouyang et al. demonstrated an implantable bioelectronic that can generate its own power by using the motion of the heart ( Figure 10J). 196 This completely implantable symbiotic pacemaker based on an implantable nanogenerator is able to electrically actuate the heart of large animals while harvesting and storing energy during the process. From the nanogenerator, the device was able to reach an open circuit voltage of up to 65.2 V. With outstanding output performance, high power density, and good durability, the reported implantable nanogenerators provide a promising platform for wireless therapeutic bioelectronic devices that are independent of an external power source for critical applications, such as treatment and diagnosis of diseases.

| Bioresorbable device
So far, conventional biomedical devices have been attached to internal tissues through open operation and require surgical extraction after accomplishing their target purpose. Additional extraction surgery imposes a further financial burden and risk of secondary infections, injury, and immune-mediated inflammatory reactions. 198 To overcome these existing limitations, bioresorbable materials have been introduced to the field of medical engineering. Bioresorbable devices dissolve into the biofluid over time, leaving behind biologically benign endproducts.
In order to fabricate devices that will entirely dissolve in our bodies, every material used in the device should be bioresorbable. For the substrates and passivation layers, chemically hydrolyzable polymers are utilized, and the rate of degradation depends on their molecular weights and crystallinity. Representative bioresorbable polymers from the nature are collagen, alginate, and chitosan. For synthetic polymers, poly(D,L-lactide) (PDLA), polyurethane, poly(α-hydroxy acids), cross-linked polyester hydrogels, and polyester are bioresorbable. 199 Figure 11A shows the process of a structural formula poly(lactic acid) containing PDLA when degrading in the human body. Lactic acid forms a cyclic lactide monomer through the condensation reaction when it meets moisture in the body. The cyclic lactide monomer can then be polymerized to form a high molecular weight polymer, such as poly(lactic acid) and PDLA. The ester bonds in the PDLA polymer chain undergo hydrolysis, leading to the cleavage of the polymer chain into smaller oligomers. 204 Bioresorbable metals can be used for electrodes or in applications of structural formation such as stents, screws, and scaffolds. 133,205 Representative bioresorbable metals are Mg, zinc (Zn), iron (Fe), W, and Mo. 206,207 Figure 11B schematically shows how metal can be decomposed by reacting with fluid in the body. 200 When a bioresorbable metal is implanted in the body, the surface of metal undergoes oxidation, which leads to the separation of metal into cations and electrons from the surface. This process generates hydrogen gas through hydroxide reactions and results in the formation of a protective metal oxide layer on the surface. Then, the high concentration of chloride in body fluids can weaken 24 of 36 the oxide layer, leading to its degradation and accelerating corrosion.
Likewise, bioresorbable materials have been applied to the existing electrical stimulation therapy for better usability and safety. For instance, Koo et al. developed a bioresorbable stimulator instead of the conventional surgical approach in intraoperative electrical stimulation to restore damaged peripheral nerve tissue. 201 This electrode was fabricated using bioresorbable metals, Mg and Mo, and encapsulated with a bioresorbable polymer, PLGA. The device restored the function of nerves demonstrated by an in vivo test. This platform degraded within 30 days including byproducts in PBS at 37°C, which is a similar environment to the human body ( Figure 11C). Moreover, Choi et al. developed a new type of bioresorbable heart pacemaker with a transient closedloop system with the integrated functions of heart rate control and cardiopulmonary status tracking. 202 An implantable system is required to monitor the heart rate and to induce electrical pulses for pacing after the heartrelated surgery and optimally would be removed after the confirmation of a stationary state. Therefore, bioresorbable materials are advantageous as they disappear gradually. For the bioresorbable material, b-DCPU was used as the encapsulating layer, and Mo was used as the conductor material. Through an in vivo test, they demonstrated the feasibility of a bioresorbable pacemaker attached to the intercostal space of a rat ( Figure 11D). As a bioresorbable property of the pacemaker, the authors observed the time to be fully dissolved, which took about 3 weeks.
In addition to metals or polymers, Si can also be used as a material for bioresorbable devices as it hydrolyzes into silicic acid. 208 The rate of Si degradation depends on the ionic content of the solution, pH, temperature, and doping level. Si forms Si(OH) 4 through the oxidation of silicon oxide (SiO 2 ), or a direct equilibrium Si + 4H 2 O↔Si(OH) 4 + 2H 2 . 209 Yu et al. developed a medical device that can detect electrical signals and stimulate the cortical surface using a bioresorbable Si-based electrode. 203 The interface electrode in contact with the nerve was composed of Si, and for the substrate, the authors utilized a bioresorbable polymer, PLGA ( Figure 11E). When the intrinsic muscles of the vibrissae of the rat were stimulated with this device, the authors observed the elongation of the whiskers of the rat. In an accelerated dissolution test performed using PBS with pH 10 at 37°C, Si and SiO 2 were dissolved within a maximum of 1 month, and PLGA was dissolved within 4-5 weeks. The entire dissolution of this device took about 2 months. By combining a bioresorbable metal and Si as an electrode, and polymer as a substrate layer, the realization of bioresorbable device was achieved. Despite the limitations of duration and wireless operation distance, bioresorbable materials can be an attractive application for medical devices that operate in vivo. As they do not require surgical extraction, it has a strong advantage in the biomedical field.

| Endowed electro-neuromodulation for optical stimulation
As the dimension of current electrodes for electrical stimulation is limited to a few micrometers, it possesses a relatively coarse level of neuromodulation. Also, as the electric field propagates, it is difficult to stimulate one type of neuron with high spatial selectivity ( Figure 12A). Recently, optogenetics has gained attention as a therapeutic method for its selective modulations on various physiological structures. [216][217][218][219] This ability is empowered by a rendering of the stimulated cells, sensitive to the action of light. Current methods of neural stimulation rely predominantly on electrical methods; however, the temporal and spatial precision afforded by neural stimulation by light holds promise as a powerful alternative. [220][221][222] This can be achieved by extracting a light-sensitive protein (opsin) that generates an electrical current in response to light, from a species of algae in nature, and expressing it to a target neuron through a virus vector ( Figure 12B). 223 As this light-sensitive opsin is activated by light with a typical wavelength, selective stimulation can be achieved by expressing these opsins in specific target neurons ( Figure 12C). 224 Optogenetics has been mainly applied as a methodological technique to map the electrophysiological responses and functional connections of complex neural tissues, such as the brain ( Figure 12D). 210 However, as stimulating strategies for therapeutic applications mainly require selective and non-invasive characteristics, this optical stimulation method is well-suited and confers great advantages compared to electrical stimulation methods ( Figure 12E). 211 Recently, the first human case of partial functional recovery after optogenetic therapy in a neurodegenerative disease was reported. Sahel et al. reported an astonishing result of vision restoration in a clinical trial by using a red-shifted channelrhodopsin, ChrimsonR. 212 For the past decade, a retinal prosthesis, which electrically stimulates the retina with microelectrode arrays, has been considered the main therapeutic approach for inherited retinal degenerative diseases (e.g., retinitis pigmentosa). However, this device requires complex surgeries by well-experienced surgeons and is still limited to low visual acuity as selective and precise stimulation with a high spatial resolution is difficult with electrical stimulation electrodes. Also, as the prosthesis is implanted in the eye, complex external device interconnections connected to the electrical stimulator can hinder the patient's movement. Compared to the conventional retinal prosthesis, optogenetic therapy does not need any complicated surgery and can be achieved with a single intravitreal injection of an adeno-associated virus vector encoded with ChrimsonR. Also, as the opsins are selectively expressed mainly on the foveal RGCs, selective stimulation to the target neurons can be made.
Still, there are several limitations to the medical application of optogenetics. First, further modifications in the light-responsive characteristics of opsins are required for medical applications. For example, channelrhodopsin-2, an opsin reactive to blue light, exhibits highlight sensitivity and fast on-off kinetics. However, it is not preferable for vision restoration, as blue light is wellknown to possess phototoxicity to the human cornea and retina. Thereby, strategies to shift the action spectra of the opsins while maintaining their sensitivity and kinetics are required. 225 ChrimsonR, mentioned above, is modified to circumvent the utilization of blue light and follow safety regulations for visual therapeutics. However, they exhibit low-light sensitivity, requiring additional amplifying tools of light. Thereby, the patient with an optogenetic-treated retina requires additional bulky goggles that project light with high intensities to activate the ChrimsonR and stimulate the retina ( Figure 12F). Second, as opsins are only reactive to a monochromatic wavelength, a single-wavelength light can only stimulate one type of neuron, thereby limiting the versatility of the device. To overcome these limitations, Li et al. used a bidirectional optogenetic modulation technique by applying dual-color micro-LED probes implanted into the ventral tegmental area of freely moving mice ( Figure 12G). 213 The dual-color micro-LED probes were vertically integrated with an indium gallium phosphide red LED (630 nm), a SiO 2 /titanium oxide-based dielectric filter, and an indium gallium nitride blue LED (480 nm). After expressing ChrimsonR and stGtACR2, which are responsive to red and blue light respectively, these red and blue LEDs were used to modulate neural activities. Demonstrations of the bidirectional optogenetic excitation and inhibition capability could be made with preference and aversion tests of individual and social behaviors. Lastly, as the penetrating depth of the light in biological tissues is limited to a few millimeters to micrometers, implantation of micro-LEDs or optical fibers near the target tissues is inevitable. 214 Therefore, approaches to implant them in minimally invasive to noninvasive forms have been reported. Kim et al. first reported the strategy to implant the cellular-scale microinorganic LEDs (micro-ILEDs) to biological tissues (i.e., the brain) for the optical stimulation of the optogenetically treated brain. 226 The multifunctional device consists mainly of micro-ILEDs for brain stimulation and microinorganic photodetectors for measuring the light intensity emitted by the micro-ILEDs. This device allows for spectroscopic evaluations, such as absorption, fluorescence, and diffuse scattering. The micro-ILED devices were successfully implanted to the brain and selectively stimulated target regions of the brain. Also, the micro-ILED was wirelessly powered to operate in freely moving mouse functioning over several months. Lee et al. developed thin and flexible micro-LEDs and intracranially implanted them between the skull and brain surface for cortical functional mapping ( Figure 12H). After the selective injection of ChrimsonR to the primary and secondary motor cortices of mouse brains, the pulsed red light of the LEDs was used to stimulate the motor neurons with no damage to the brain. Chen et al. developed a strategy to stimulate the deep brain without any intracranial surgery by combining modified opsins with high light sensitivity and optical stimulation with short pulse durations, to have deeper penetration depth ( Figure 12I). 215 ChRmine, the opsin used in this study, exhibited extremely improved operational light sensitivity with large photocurrents compared with existing fast red-shifted variants. Transcranial photoactivation was made with short pulses of 635 nm light from the fiber positioned above the surface of the intact skull and was able to stimulate the midbrain and brainstem structures with an unprecedented depth of up to 7 mm with millisecond precision.
Although optogenetics is in its early stages and has several limitations to overcome for clinical therapeutics, it certainly has promising advantages that are complementary to the aforementioned electrical stimulation methods. Recent studies have implemented various strategies to modify or enhance optical properties of the light source, such as utilizing nanostructures to the micro-LEDs, or to enable effective optical stimulation by forming a cranial window in mice. 227,228 In addition, further clinical studies will greatly enhance its potential for contributions to therapeutic applications.

| CONCLUSION
Electrical stimulation has been studied as a therapeutic approach for a wide range of applications from neuromodulation to wound healing. Owing to its applicability and versatility, it is a promising treatment method, T A B L E 1 Summary of electrical stimulation for therapy. especially for patients who do not respond to medications or are ineligible to undergo a surgical operation. We summarized the applications of electrical stimulation for therapy with respect to clinical trials (Table 1). In order to develop robust compatibility of electrical stimulation devices for various tissues and to acquire the desired responses from the tissues, investigations in terms of materials and structural designs have been extensively conducted. However, the exact mechanisms for how electrical stimulation brings about specific therapeutic results are unexplained for certain diseases and injuries. This obscurity regarding electrical stimulation restricts our understanding of the definite action of the individual mechanism, which impacts the optimization of stimulating conditions. For example, not only epileptic seizures but also depressive disorders can be alleviated by stimulating vagus nerves, and the specific conditions necessary to target one of the symptoms is unknown. As a result, electrical stimulation therapy is usually implemented as a supplementary method. In this regard, to improve clinical feasibility, continuous effort is needed to clarify how electrical stimulation works to remedy symptoms or to recover injuries.
In addition, in the case of clinical applications, there are still obstructions during or after the treatment with electrical stimulation devices, which requires additional surgical procedures. To date, in order to overcome this limitation, applications of WPT, energy harvesting, or bioresorbable materials on electrical stimulation devices have been widely investigated as reviewed in this paper. Unfortunately, these approaches are restricted only to experimental demonstrations due to their weak performance and low durability. Further research to develop reliable systems to deliver specific stimulations without additional surgical operation can improve the practicality of electrical stimulation therapy for patients. For patients who suffer from a disability in locomotor movements due to the spinal cord injury, electrical stimulation devices attached to the spinal cord facilitate the recovery of motor function. If this system adopts WPT, patients can go through the rehabilitation training without the constraints from a wired system along with the electrical stimulation even in their daily life. The combination of electrical stimulation and physical training promotes motor recovery, and it enables spontaneous locomotion without the aid of the device. Conclusively, a comprehensive understanding of the mechanisms and various applications of electrical stimulation therapy will facilitate the progressive developments of devices for delivering electrical stimulation with the promise of restoring the ordinary life of patients and enhancing the quality of their lives.