Electronic tissue sensing and stimulation system and method

Abstract

A system, method, and wearable device configured to both actuate a muscle and to sense the muscle's response to the actuation to determine treatment parameters for a subsequent actuation. The actuation and sensing components are in communication with a machine learning component to evaluate the data output from the sensing and actuation and to determine whether and how to adjust treatment parameters for subsequent treatment sessions.

Description

279430.357_NPTITLEELECTRONIC TISSUE SENSING AND STIMULATION SYSTEM AND METHODCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims priority to U.S. Provisional Application Serial No. 63 / 712,741, filed on October 28, 2024, which application is incorporated by reference herein in its entirety.STATEMENT REGARDING FEDERAL SPONSORSHIP

[0002] This invention was made with the support of the United States government under the Department of the Interior for DARPA D20AC00002. The U.S. government has certain rights in the invention.BACKGROUND OF THE INVENTIONField of the Invention

[0003] The present invention relates to electrical, mechanical, and acoustic tissue sensing and stimulation to counteract muscle degeneration.Background of the Invention

[0004] The present invention was inspired by treatment needs relating to volumetric muscle loss (“VML”), a traumatic or surgical injury that results in the irreversible loss of skeletal muscle tissue beyond the body’s natural capacity for regeneration, leading to chronic weakness, fibrosis, and functional impairment. Muscular degeneration, and corresponding treatment needs, also exist with other injuries and conditions as explained herein.

[0005] Muscular degeneration, a broad term for VML, muscle atrophy, muscle wasting, and neuromuscular degeneration (collectively herein “muscular degeneration”), describes the progressive loss of skeletal muscle mass, strength, and function. These conditions can result from a wide range of causes and have significant implications for mobility, independence, and overall health on individuals suffering from them.279430.357_NP

[0006] Some examples of muscular degeneration include the following: (i) disuse atrophy, which can be caused by physical inactivity or immobilization and is often reversible with exercise and physical therapy; (ii) neurogenic atrophy, which results from nerve damage or neurological diseases like Amyotrophic Lateral Sclerosis (“ALS”), Spinal Muscular Atrophy (“SMA”), and Multiple Sclerosis (“MS”); pathological atrophy (e.g., Cachexia), which is characterized by weight loss, muscle wasting, loss of appetite, and weakness; (iii) muscular dystrophies (e.g., Duchenne, Becker, Limb-Girdle) that cause progressive muscle degeneration and weakness; (iv) sarcopenia, which is age-related loss of muscle mass; and (v) post-surgical function muscle loss after a wide-variety of surgeries, such as total knee arthroplasty (“TKA”); anterior cruciate ligament (“ACL”) reconstruction; meniscus repair or arthroscopy; total hip arthroplasty (“THA”); rotator cuff repair; spinal fusion or laminectomy; fracture fixation (e.g., femur, tibia, humerus); amputation; shoulder arthroplasty (TSA or RS A); Achilles Tendon repair; anterior cervical discectomy and fusion (“ACDF”); lumbar discectomy; shoulder labral repair (Bankart or SLAP); total ankle arthroplasty or fusion; and carpal tunnel or peripheral nerve decompression. Whether because the muscle itself is unable to be exercised or moved normally for a period of time due to injury or illness to the specific muscle(s) or because the specific muscle(s) cannot be moved or worked because of an injury or illness to an adjacent muscle, the lack of movement and use of any muscle in the body eventually leads to muscular degeneration. Clinically, muscle wasting conditions manifest the following symptoms: progressive muscle weakness and shrinking, fatigue, low endurance, difficulty with mobility, facial or limb weakness, trouble swallowing or speaking (in neurogenic causes), weight loss, postural changes and muscle tone reduction.

[0007] In general, the treatment for muscular degeneration depends on the underlying cause but treatments may include any or a combination of the following: physical therapy, exercise, resistance training to rebuild muscle mass, mobility aids, rehabilitation, nutritional279430.357_NP support (high-protein, high-calorie diets), supplements (e.g., vitamin D, B12, creatine), pharmacologic interventions e.g. anti-inflammatory agents, anabolic steroids, or experimental drugs), disease-specific treatments (e.g., riluzole for ALS, corticosteroids for Duchenne MD), and management of the underlying disease or illness. The general practice is to begin treatment for muscular degeneration after the muscle loss becomes visible to the clinician’s eye or several weeks after the injury or illness has been diagnosed, which is when physical therapy normally would begin. This results in a delay between when muscular degeneration may have begun at a cellular level and when it is noticed and treated according to standard medical protocol.

[0008] Physical therapists use a broad range of rehabilitation techniques to combat muscular degeneration conditions, aiming to preserve or restore muscle mass, strength, and mobility. Traditional exercise-based interventions, such as progressive resistance training, eccentric and isometric exercises, and task-oriented movement retraining remain the most effective methods for reversing disuse or age-related atrophy when voluntary muscle activation is possible. These strategies stimulate protein synthesis, improve neural coordination, and help restore functional independence. However, in many clinical populations, including those affected by immobilization, neurological injury, or chronic disease, patients often cannot generate sufficient voluntary contractions to achieve meaningful muscle loading.

[0009] To address this limitation, neuromuscular electrical stimulation (“NMES”) (a type of muscle stimulation or actuation) has become a widely adopted adjunct therapy. NMES delivers controlled electrical pulses to motor nerves and / or muscle fibers to evoke contractions artificially, mimicking the effects of active exercise. When properly applied, NMES can preserve muscle mass during immobilization, enhance recovery after orthopedic surgery, and assist patients with neurological conditions such as stroke, spinal cord injury, or279430.357_NP multiple sclerosis. Over time, repeated NMES sessions can promote hypertrophy, increase local blood flow, and help maintain neuromuscular connectivity even in cases of partial denervation. Despite these benefits, NMES remains underutilized in both clinical and homebased rehabilitation because of several key limitations in current systems discussed below.

[0010] First, most commercial NMES devices lack real-time feedback or progress monitoring, making it difficult for physical therapists or patients to assess the quality or effectiveness of stimulation sessions. Without feedback on muscle activation, fatigue, or contraction strength, clinicians must rely largely on subjective reporting or visual observation to adjust therapy parameters. This absence of quantitative data limits the ability to personalize therapy, track functional gains, or identify when stimulation parameters should be modified for optimal outcomes. In contrast to modem fitness or rehabilitation technologies that provide continuous data streams, NMES systems typically offer no integrated sensing or performance analytics, leaving a major gap between stimulation delivery and physiological response tracking.

[0011] Second, conventional NMES platforms are often unintuitive and cumbersome for patients to operate independently. Correct electrode placement, amplitude calibration, and session programming require expertise that many patients lack, especially those with limited dexterity, cognitive impairment, or pain sensitivity. Small misplacements in electrode position can drastically alter which muscles are activated or whether a contraction occurs at all, reducing consistency and effectiveness. As a result, NMES treatments are often confined to supervised clinic visits rather than daily use at home, limiting long-term adherence and reducing therapeutic benefit. This lack of user-friendly design undermines one of NMES’ s potential strengths, its ability to provide continuous, low-cost, decentralized therapy outside clinical settings.279430.357_NP

[0012] Additionally, NMES-induced contractions tend to rapidly fatigue muscles because stimulation synchronously activates all motor units rather than following the natural, graded recruitment pattern of voluntary muscle control. This leads to short session durations, limited force generation, and discomfort during prolonged use. Patients frequently report sensations of burning or tingling beneath electrodes, further discouraging consistent application. Electrode contact variability, differences in skin impedance, and poor signal delivery across sessions compound these problems, resulting in inconsistent stimulation efficacy between uses or individuals.

[0013] Together, these challenges highlight the gap between the theoretical potential and the practical implementation of NMES. While the technology can effectively evoke muscle activity, current systems lack the closed-loop feedback, adaptive control, and intuitive interfaces necessary for reliable self-administration and long-term engagement. The methods and systems of the present invention solve these problems by incorporating integrated sensing and feedback (such as EMG, impedance, or motion-based metrics) to provide patients and therapists with quantitative data and real-time guidance on placement, stimulation quality, and progress. User-centered designs that include automated calibration, adaptive algorithms that minimize fatigue, and app-based or device-based visualization tools could transform NMES from a technician-dependent modality into an accessible, data-driven therapy for at- home rehabilitation.

[0014] Emerging evidence also suggests that ultrasound-based therapeutic modalities can be used synergistically with NMES to improve tissue recovery and muscle remodeling. Low-intensity pulsed or focused ultrasound has been shown to modulate inflammation, enhance angiogenesis, and reduce fibrotic scar formation in injured or immobilized tissue. When integrated with NMES platforms, ultrasound can help soften or remodel fibrotic regions, improving tissue compliance and facilitating more effective electrical conduction279430.357_NP and muscle recruitment. This combination approach may simultaneously reduce fibrosis while promoting neuromuscular regeneration, offering a multimodal pathway toward restoring both structure and function within chronically damaged muscle.

[0015] There are currently many sensing technologies available, but none of them are incorporated into a user-friendly system of actuation and sensing as per various embodiments of the present invention. Surface electromyography (“sEMG”) is one of the most widely used sensing tools for monitoring muscle activity. It works by detecting voltage fluctuations on the skin surface that occur when motor units fire during contraction, offering insight into activation intensity, fatigue, and coordination. Although sEMG is noninvasive and useful for rehabilitation and NMES feedback, it suffers from several limitations, including signal contamination from nearby muscles (crosstalk), motion artifacts, inconsistent electrode-skin contact, and difficulty isolating deep or small muscles.

[0016] Mechanomyography (“MMG”) captures the mechanical vibrations of contracting muscles using accelerometers, piezoelectric sensors, or microphones placed on the skin. It provides a mechanical perspective on muscle activation and can complement electromyography (“EMG”), especially when electrical signals are distorted by stimulation artifacts. However, MMG signals are weak, highly susceptible to motion noise, and lack standardized processing, which limits clinical adoption.

[0017] Electrical Impedance Myography (“EIM”) measures how a small alternating current travels through muscle tissue to assess its composition and health. Because impedance changes with muscle density, hydration, and fibrosis, EIM can reveal early signs of atrophy or recovery and could be integrated into electrodes for real-time NMES feedback. Still, it is affected by skin and fat thickness, provides only average regional data, and requires complex interpretation before it can become a routine monitoring tool.279430.357_NP

[0018] Ultrasound imaging and elastography offer structural and mechanical information by using sound waves to visualize muscle architecture and measure tissue stiffness. These modalities are particularly useful for detecting fibrosis, monitoring muscle regeneration, and quantifying changes in elasticity during recovery. Despite their high diagnostic value, they are operator-dependent, sensitive to probe positioning, and currently too bulky or expensive for routine integration with NMES devices.

[0019] Force, torque, and pressure sensors quantify the mechanical output of muscle contractions and are often used in rehabilitation equipment and robotic systems. They provide objective measures of strength recovery and can potentially feed back into NMES systems to adjust stimulation based on generated force. Yet, these sensors cannot isolate which muscle produces the force, require careful alignment and calibration, and are often too cumbersome for wearable or at-home applications.

[0020] Current rehabilitation and neuromuscular therapies suffer from a critical gap between stimulation and sensing, which limits both the precision and accessibility of muscle recovery interventions. Existing NMES systems, while effective at evoking muscle contractions, are primarily used in clinics under the supervision of physical therapists and largely operate in an open-loop configuration. This means stimulation is delivered without real-time feedback on how the muscle or tissue is responding. Without sensing or adaptive control, clinicians must rely on visual observation or patient feedback to gauge progress, which often leads to suboptimal dosing. As a result, treatments are frequently underdosed (ineffective) or overdosed (painful or fatiguing), and patients are unable to continue effective therapy outside the clinic setting.

[0021] Sensing technologies such as surface EMG, electrical impedance myography, and ultrasound imaging provide valuable insights into muscle activation and tissue health but are also typically limited to use by trained professionals within clinical environments. EMG279430.357_NP can measure activation but is susceptible to noise and electrode placement errors, while EIM and ultrasound can reveal fibrosis and tissue remodeling but require bulky equipment and expert interpretation. These systems are not designed for home use or seamless integration with stimulation platforms. Consequently, physical therapists have limited real-time data to guide NMES settings, and patients have no objective feedback to monitor progress between visits. This disconnection between actuation (stimulation) and perception (sensing) remains one of the major barriers preventing NMES from evolving into a personalized, adaptive, and data-driven therapy that can extend beyond the clinic.

[0022] Various embodiments of the present invention directly address these gaps by combining muscle stimulation (often electrical) and multimodal sensing into a unified, closed-loop platform that can be safely deployed anywhere, both in the clinic and at home. The system continuously monitors physiological markers, such as sEMG activity, impedance changes, and muscle motion, and uses these signals to automatically adjust stimulation parameters in real time. This enables a feedback loop, where stimulation intensity and timing are continuously optimized based on muscle response rather than preset, static parameters. The integrated sensors and control algorithms provide both patients and clinicians with quantitative metrics of activation strength, fatigue, and tissue remodeling, displayed through an intuitive software interface or mobile application.

[0023] By merging sensing and stimulation into a single adaptive system, this platform transforms NMES from a clinic-bound, technician-operated tool into an intelligent, self-regulating rehabilitation ecosystem. It enables therapy to be personalized, data- informed, and accessible anywhere, empowering patients to engage in effective recovery outside the supervision of a physical therapist. The closed-loop control ensures stimulation remains both effective and comfortable while minimizing fatigue and enhancing long-term outcomes. Simultaneously, the real-time sensing data offer remote monitoring capabilities,279430.357_NP allowing therapists to track progress and adjust treatment plans without requiring in-person sessions.

[0024] Muscular degeneration places a substantial and growing burden on the U.S. population and healthcare system. Studies estimate that roughly 18% of adults aged 60 and older exhibit reduced muscle strength, with 10-16% meeting clinical criteria for sarcopenia, the age-related loss of muscle mass and function. Among nursing home residents and hospitalized individuals, prevalence rates can exceed 50%, underscoring how muscle wasting disproportionately affects the most vulnerable. From the patient’s perspective, this degeneration translates into difficulty performing daily tasks such as standing, walking, or lifting objects, and contributes to falls, fractures, and progressive loss of independence. In one national survey, over half (55%) of adults with weak muscle strength reported difficulty rising from a chair, compared to just 13% with normal strength, highlighting the tangible impact on quality of life.

[0025] Economically, the consequences are profound. United States hospitalizations involving cachexia, a severe muscle wasting syndrome, are associated with hospital stays that are twice as long and over $4,600 more expensive per admission than those without cachexia. Similarly, sarcopenia has been estimated to contribute billions of dollars annually in excess healthcare expenditures through extended hospitalizations, increased long-term care needs, and rehabilitation costs. Beyond direct expenses, the societal toll includes caregiver burden, lost productivity, and reduced quality of life for millions of affected individuals. As the population continues to age, the prevalence and economic strain of muscular degeneration are expected to rise sharply, emphasizing the urgent need for preventative and restorative intervention. These statistics demonstrate a need to treat muscular degeneration more effectively, efficiently, and as early as possible to improve patient response to treatments. The systems, methods, and devices of the present invention address this need.279430.357_NPBRIEF SUMMARY OF THE INVENTION

[0026] To facilitate understanding of the invention, the drawings and description illustrate preferred embodiments thereof, from which the invention, various embodiments of its structures, construction and method of operation, and many advantages, may be understood and appreciated. The drawings hereby are incorporated by reference.

[0027] One embodiment of the present invention is a system for treating muscular degeneration comprising: (i) an actuation subsystem configured to stimulate at least one muscle during a treatment and to generate actuation output data; (ii) a sensing subsystem configured to communicate with the actuation subsystem and to evaluate the actuation output data to identify at least one physiological response by the stimulated muscle and to generate sensing output data based upon the evaluation; and (iii) at least one machine learning subsystem in communication with the actuation subsystem and the sensing subsystem, the at least one machine learning subsystem configured to process the sensing output data to generate at least one treatment parameter.

[0028] For one embodiment of the present invention, the previously-described system is configured so that the machine learning subsystem archives the sensing output data from a first treatment for comparison to sensing output data from a subsequent treatment. In one embodiment of the present invention, the at least one treatment parameter is communicated to the actuation subsystem for use during a subsequent treatment. In one embodiment, the actuation subsystem comprises at least one stimulation method selected from the group consisting of electrical, mechanical, bioelectrical, sound, and heat. In an additional embodiment, the sensing subsystem is configured to incorporate at least one sensing modality selected from the group consisting of electrophysiological, mechanical, optical, thermal, and pressure and wherein the at least one modality can quantify a neuromuscular state or a279430.357_NP therapeutic response. In another embodiment, the actuation subsystem and the sensing subsystem are configured to fit on a wearable device. One embodiment also comprises an array of sensors configured to function as part of the actuation subsystem and the sensing subsystem.

[0029] One embodiment of the present invention is a wearable device for treating muscular degeneration comprising the following: (i) an actuation subsystem configured to actuate at least one muscle and configured to generate actuation output data; (ii) a sensing subsystem configured to receive the actuation output data and configured to identify at least one physiological response by the actuated muscle based upon the actuation output data and to generate sensing output data; and (iii) at least one machine learning subsystem in communication with the actuation subsystem and the sensing subsystem, the at least one machine learning subsystem configured to process the sensing output data to generate at least one treatment parameter.

[0030] For another embodiment of a device of the presentation invention, the actuation subsystem, the sensing subsystem, and the at least one machine learning subsystem are incorporated into a wearable cuff configured to wrap around a body part.

[0031] One embodiment of a wearable device of the present invention is a cuff comprising the following: (i) a skin interface layer that contacts the body part; (ii) a sensing and electrode layer configured with at least one electrode that is integrated into a flexible substrate; (iii) an electronics pod configured to run the machine learning subsystem and to provide communication and energy to the actuation subsystem, the sensing subsystem, and the machine learning subsystem; (iv) a conductive pathway layer comprising conductive fibers or printed metallic traces forming flexible circuits connecting the electrodes to the electronics pod; (v) an insulating and structural layer that isolates the conductive fibers or metallic traces from an environment around the cuff; and (v) an outer shell.279430.357_NP

[0032] For one embodiment of a device, an array of electrodes is configured to contact the body part to be actuated. For another embodiment, the electrodes are configured for actuation and sensing. One embodiment of a device includes a display on the outer shell configured to display instructions to a user. Another embodiment of a device has the machine learning subsystem configured as a continuous, closed loop to process real-time data from the sensing subsystem to provide the ate least one treatment parameter. For one embodiment, the sensing subsystem is electrically and mechanically coupled to the actuation subsystem to ensure temporal alignment of stimulation and sensing.

[0033] One embodiment of the present invention is a method of treating muscular degeneration comprising the following steps: (i) evaluating a muscle’s physiology with a sensing subsystem to produce sensing output data; (ii) processing sensing output data by a machine learning subsystem to produce at least one treatment parameter; (iii) actuating the muscle with an actuation subsystem according to the at least one treatment parameter; and (iv) repeating the evaluating, processing steps to produce at least one updated treatment parameter.

[0034] Another embodiment of a method comprising verifying the proper placement of at least one electrode on a body part prior to evaluating a muscle’s physiology. For one embodiment, the machine learning subsystem comprises the following steps: (i) extracting features from and preprocessing the sensing output data to filter out noise and artifacts from the sensing output data to generate filtered output data; and (ii) processing the filtered output data for model prediction and adaptive control to produce the at least on treatment parameter for use by the actuation subsystem. For one embodiment, these steps are repeated until the sensing subsystem detects no continued muscular degeneration from the muscle’s physiology.

[0035] One embodiment of the present invention is a device for treating muscular degeneration comprising an actuation subsystem configured to stimulate at least one muscle during a treatment and to generate actuation output data and a sensing subsystem configured279430.357_NP to communicate with the actuation subsystem and configured to (i) receive the actuation output data, and (ii) evaluate the actuation output data to determine at least one physiological attribute of the muscle. Also, for this embodiment, the actuation subsystem and the sensing subsystem are configured as a portable system.BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0036] To facilitate understanding of the invention, the drawings and description illustrate preferred embodiments thereof, from which the invention, various embodiments of its structures, construction and method of operation, and many advantages, may be understood and appreciated. The drawings hereby are incorporated by reference.

[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate one or more implementations described herein and, together with the description, explain these implementations. The drawings are not intended to be drawn to scale, and certain features and certain views of the figures may be shown exaggerated, to scale or in schematic in the interest of clarity and conciseness. Not every component may be labeled in every drawing. Like reference numerals in the figures may represent and refer to the same or similar element or function.

[0038] Figure 1 illustrates one embodiment of a schematic of a system, device, and method of the present invention;

[0039] Figure 2 illustrates one embodiment of a method of the present invention;

[0040] Figures 3A and 3B show examples of conventional actuation and sensing modalities;

[0041] Figure 4 illustrates the standard timeline for recovery after VML;

[0042] Figures 5A illustrates prior rat VML research in which NMES was preformed inadequately and 5B illustrates rat VML research supporting the present invention in which279430.357_NP bioelectronic diagnostics informed preferable electrical stimulation parameters leading to enhanced functional recovery;

[0043] Figures 6A through 6F illustrate the ability of one embodiment of the system or method to monitor safe charge injection;

[0044] Figure 7 illustrates the use of EIM with healthy and atrophied muscles;

[0045] Figures 8A through 8F illustrate bioelectronic diagnostics of EIM with one embodiment of the present invention;

[0046] Figures 9A and 9B show raw data extracted from EIM scans of healthy rat’s tibialis anterior according to one embodiment of the present invention;

[0047] Figures 10A through 10D show raw data extracted from EIM scans of rat’s tibialis anterior one week post-VML with and without electrical stimulation according to one embodiment of the present invention;

[0048] Figures 11 A through 1 ID illustrate the phase angle correlation with myo fiber percentage and torque production following neuromuscular injury;

[0049] Figures 12A and 12B illustrate the phases of neuromuscular regeneration following injury;

[0050] Figures 13A and 13B show a representative cross-section of a heathy rat tibialis anterior (“TA”);

[0051] Figures 14A through 14D show a representative cross-section of an inflamed mid atrophy rat tibialis anterior muscle without bio-signal informed electrical stimulation;

[0052] Figures 15A through 15D show a representative cross-section of a less- inflamed mid atrophy rat tibialis anterior muscle with bio-signal informed electrical stimulation;

[0053] Figures 16A through 16F illustrate the use of EMG activity to guide stimulation timing and delivery according to various embodiments of the present invention;279430.357_NP

[0054] Figures 17A through 17F illustrate how the strength of the rats was quantified using the force gauge;

[0055] Figures 18A through 18C show that electrical stimulation increases functional torque with the present invention;

[0056] Figures 19A through 19E illustrate that stimulation with embodiments of the present invention enhance damaged muscle (TA) not causes excessive hypertrophy to compensatory muscles (EDL);

[0057] Figures 20 A through 20F illustrate charge injection capacity in vivo results after 3 -days of electronic stimulation;

[0058] Figures 21 A through 2 IK show histological analyses of muscle tissue across time before and after volumetric muscle loss injury;

[0059] Figures 22 A through 22C show one embodiment of a compression cuff of the present invention;

[0060] Figure 23 shows one embodiment of a cuff of the present invention configured to be wrapped around a person’s leg;

[0061] Figures 24 shows one embodiment of a machine learning component of the present invention; and

[0062] Figure 25 shows two embodiments of wearable devices according to the present invention embodied as a shirt and a pair of pants.DETAILED DESCRIPTION OF THE INVENTION

[0063] The following describes example embodiments in which the present invention may be practiced. This invention, however, may be embodied in many ways, and the description provided herein should not be construed as limiting. Among other things, the following invention may be embodied as systems, methods, or devices. The following detailed279430.357_NP descriptions should not be taken in a limiting sense. The accompanying drawings are hereby incorporated by reference.

[0064] The phrases “in some embodiments,” “in one embodiment,” “in various embodiments,” “according to various embodiments,” “in the embodiments shown,” “in other embodiments,” and the like generally mean the particular feature, structure, or characteristic following the phrase is included in at least one embodiment of the present invention, and may be included in more than one embodiment of the present invention. In addition, such phrases do not necessarily refer to the same or different embodiments.

[0065] If the specification states a component, element, part, or feature “may,” “can,” “could,” or “might” be included or have a characteristic, that particular component or feature is not required to be included or have the characteristic.

[0066] In this document, the terms “a” or “an” are used, as is common in patent documents, to include one or more than one. In this document, the term “or” is used to refer to a nonexclusive “or ” such that “A or B” includes “A but not B,” “B but not A,” and “A and B,” unless otherwise indicated. Furthermore, all publications, patents, and patent documents referred to in this document are incorporated by reference herein, as though individually incorporated by reference. In the event of inconsistent usages between this document and those so incorporated by reference, the usage in the incorporated reference(s) should be considered supplementary to that of this document; for irreconcilable inconsistencies, the usage in this document controls. Before explaining at least one embodiment of the disclosure in detail, it is to be understood that the disclosure is not limited in its application to the details of construction, experiments, exemplary data, and / or the arrangement of the components set forth in the following description or illustrated in the drawings unless otherwise noted. The disclosure is capable of other embodiments or of being practiced or carried out in various ways. Also, it is279430.357_NP to be understood that the phraseology and terminology employed herein is for purposes of description and should not be regarded as limiting.

[0067] As used in the description herein, the terms “comprises,” “comprising,” “includes,” “including,” “has,” “having,” or any other variations thereof, are intended to cover a nonexclusive inclusion. For example, unless otherwise noted, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to only those elements but may also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0068] Further, unless expressly stated to the contrary, “or” refers to an inclusive and not to an exclusive “or”. For example, a condition A or B is satisfied by one of the following: A is true (or present) and B is false (or not present), A is false (or not present) and B is true (or present), and both A and B are true (or present).

[0069] In addition, use of the “a” or “an” are employed to describe elements and components of the embodiments herein. This is done merely for convenience and to give a general sense of the inventive concept. This description should be read to include one or more, and the singular also includes the plural unless it is obvious that it is meant otherwise. Further, use of the term “plurality” is meant to convey “more than one” unless expressly stated to the contrary.

[0070] As used herein, qualifiers like “substantially,” “about,” “approximately,” and combinations and variations thereof, are intended to include not only the exact amount or value that they qualify, but also some slight deviations therefrom, which may be due to computing tolerances, computing error, manufacturing tolerances, measurement error, wear and tear, stresses exerted on various parts, and combinations thereof, for example.

[0071] As used herein, any reference to “one embodiment,” “an embodiment,” “some embodiments,” “one example,” “for example,” “various embodiments,” or “an example” means that a particular element, feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment and may be used in conjunction with other279430.357_NP embodiments. The appearance of the phrase “in some embodiments” or “one example” in various places in the specification is not necessarily all referring to the same embodiment, for example.

[0072] The use of ordinal number terminology (i.e., “first”, “second”, “third”, “fourth”, etc.) is solely for the purpose of differentiating between two or more items and, unless explicitly stated otherwise, is not meant to imply any sequence or order of importance to one item over another.

[0073] The use of the term “at least one” or “one or more” will be understood to include one as well as any quantity more than one. In addition, the use of the phrase “at least one of X, Y, and Z” will be understood to include X alone, Y alone, and Z alone, as well as any combination of X, Y, and Z.

[0074] Where a range of numerical values is recited or established herein, the range includes the endpoints thereof and all the individual integers and fractions within the range, and also includes each of the narrower ranges therein formed by all the various possible combinations of those endpoints and internal integers and fractions to form subgroups of the larger group of values within the stated range to the same extent as if each of those narrower ranges was explicitly recited. Where a range of numerical values is stated herein as being greater than a stated value, the range is nevertheless finite and is bounded on its upper end by a value that is operable within the context of the invention as described herein. Where a range of numerical values is stated herein as being less than a stated value, the range is nevertheless bounded on its lower end by a non-zero value. It is not intended that the scope of the invention be limited to the specific values recited when defining a range. All ranges are inclusive and combinable.

[0075] The following description has set forth aspects of computer system or computer- implemented devices and / or processes via the use of block diagrams, flowcharts, and / or examples, which may contain one or more functions and / or operations. As used herein, the term or graphic of a “block” in the block diagrams and flowcharts refers to a step of a computer- implemented process executed by a computer system, which may be implemented as a machine279430.357_NP learning system or an assembly of machine learning systems. Each block can be implemented as either a machine learning system or as a nonmachine learning system, according to the function described in association with each particular block. Furthermore, each block can refer to one of multiple steps of a process embodied by computer-implemented instructions or programs executed by a computer system (which may include, in whole or in part, a machine learning system) or an individual computer system (which may include, e.g., a machine learning system) executing the described step, which is in turn connected with other computer systems (which may include, e.g., additional machine learning systems) for executing the overarching process described in connection with each figure or figures.

[0076] Circuitry, as needed herein to connect components (as will be known to one skilled in the art), may be analog and / or digital components, or one or more suitably programmed processors (e.g., microprocessors) and associated hardware and software, or hardwired logic. The term “processor” as used herein means a single processor or multiple processors working independently or together to collectively perform a task or a functional unit that interprets and executes instruction data. Also, “components” may perform one or more functions. The term “processing component,” refers to a central processing unit that can include hardware, such as a processor (e.g., microprocessor), an application specific integrated circuit (“ASIC”), a field programmable gate array (“FPGA”), a combination of hardware and software, software, and / or the like. A processing component comprises the hardware and software configured to perform or execute the models, methods, and process of the present invention including performing systematic operations upon data or information exemplified by functions such as data or information transferring, merging, sorting, and computing (e.g., arithmetic operations or logical operations).

[0077] Software may include one or more computer readable instruction that when executed by one or more component, e.g., a processor, causes the component to perform a specified function. It279430.357_NP should be understood that the algorithms described herein may be stored on one or more non-transitory computer-readable medium. Exemplary non-transitory computer-readable media (all examples of memory) may include a non-volatile memory, a random access memory (“RAM”), a read only memory (“ROM”), a CD-ROM, a hard drive, a solid-state drive, a flash drive, a memory card, a DVD- ROM, a Blu-ray Disk, a laser disk, a magnetic disk, an optical drive, combinations thereof, and / or the like. Such non-transitory computer-readable media may be electrically based, optically based, magnetically based, resistive based, and / or the like.

[0078] As used herein, the terms “network-based,” “cloud-based,” and any variations thereof, are intended to include the provision of configurable computational resources on demand via interfacing with a computer and / or computer network, with software and / or data at least partially located on a computer and / or computer network.

[0079] The various embodiments of the present invention may include one or more input device (hereinafter “input device”), one or more output device (hereinafter “output device”), one or more processors or processing component (used interchangeably), one or more communication device (hereinafter “communication device”) capable of interfacing with a network, and one or more memory (hereinafter “memory”) storing processor-executable code and / or application(s) (hereinafter “application ”).

[0080] An input device, the output device, the processing component, the communication device, and the memory may be connected via a path such as a data bus that permits communication among the components of the classification device. Additionally, one device can serve as both an input and an output device, or the input device and output device can be separate devices.

[0081] The input device may be capable of receiving information input from a user and / or the processing component and transmitting such information to other components and / or a network. The input device may include, but is not limited to, implementation as one or more sensors, a279430.357_NP keyboard, a touchscreen, a mouse, a trackball, a microphone, a camera, a fingerprint reader, an infrared port, an optical port, a cell phone, a smart phone, a PDA, a remote control, a fax machine, a wearable communication device, a network interface, combinations thereof, and / or the like, for example.

[0082] The output device may be capable of outputting information in a form perceivable by the processing component. Implementations of the output device may include, but are not limited to, an actuation device, a computer monitor, a screen, a touchscreen, a speaker, a website, a television set, a smart phone, a PDA, a cell phone, a fax machine, a printer, a laptop computer, a haptic feedback generator, an olfactory generator, combinations thereof, and the like, for example. It is to be understood that in some exemplary embodiments, the input device and the output device may be implemented as a single device, such as, for example, a touchscreen of a computer, a tablet, or a smartphone. It is to be further understood that as used herein the term user e.g., the user) is not limited to a human being, and may comprise a computer, a server, a website, a processor, a network interface, a user terminal, a virtual computer, combinations thereof, and / or the like, for example. The output device may display a user interface. The output device may feed into another computer system or processor.

[0083] The processing component or processor may be implemented as a single processor or multiple processors working together, or independently, to execute the application as described herein. It is to be understood, that in certain embodiments using more than one processing component, the processing components may be located remotely from one another, located in the same location, or comprising a unitary multi-core processor, or a combination thereof. The processing component may be capable of reading and / or executing processor-executable code and / or capable of creating, manipulating, retrieving, altering, and / or storing data structures into the memory such as in a database. The processing component may be capable of communicating with the memory via the path (e.g., the data bus). The processing component may be capable of279430.357_NP communicating with the input device and / or the output device communicably coupled, or otherwise connected, to the classification device of the classification system.

[0084] The processing component, processor, or computer processing unit may be further capable of interfacing and / or communicating with a server system via the network using the communication device. For example, the processing component may be capable of communicating via the network by exchanging signals (e.g., analog, digital, optical, and / or the like) via one or more port (e.g., physical ports or virtual ports) using a network protocol to provide updated information to the application or the user interface. In one embodiment, the server system is another embodiment of the classification device, however, the server system maybe constructed, for example, as one or more server having a plurality of CPUs, GPUs, NPUs, TPUs, and / or the like, or a combination thereof. The server system may thus have a processing power available to both execute, or run, an Al model, as well as train, fine-tune, pre-train, instruction-tune, and / or align the Al model. The server system may be specially designed to handle large-scale datasets efficiently.

[0085] In one implementation, the processing component may be operable to receive the electrical signals from an artificial intelligence (“Al”) processor (a type of processor.) The Al processor may be constructed in accordance with the processing component, for example, and, in some embodiments, may be incorporated into the processing component. In some embodiments, the Al processor may be separate from the processing component but may work together with the processing component to execute the application and / or access the memory. In one embodiment, the Al processor may operate at the request of, or be instructed to execute code by, the processing component.

[0086] Exemplary implementations of the processing component or processor also may include, but are not limited to, a digital signal processor (“DSP”), a central processing unit (“CPU”), a graphical processing unit (“GPU”), a neural processing unit (“NPU”), a tensor processing unit (“TPU”), a field programmable gate array (“FPGA”), a microprocessor, a multi-core processor, an279430.357_NP application specific integrated circuit (“ASIC”), combinations thereof, and / or the like, for example. The processing component may include one or more processing component, having the same or different implementations, working together, or independently, and located locally, or remotely, e.g., accessible via the network such as located in the server system, and may include a multi-core, multiprocessor component. As such, the application may be considered a cloud-based application, enabling access to powerful computing resources of the server system and simplifying user experience via the user interface. This implementation as a cloud-based application also drastically reduces processing time, as CUDA and tensor cores (e.g., the Al processors) allow the processing component to perform matrix multiplication at much faster rates.

[0087] In one implementation, the memory may be one or more non-transitory processor- readable medium. The memory may store processor-executable instructions, such as the application, that, when executed by the processing component, causes the processing component of the classification device to perform an action such as communicate with or control one or more component of the classification device and the classification system and / or to perform one or more process such as the classification system. The memory may be one or more memory working together, or independently, to store processor-executable code and may be located locally or remotely, e.g., accessible via the network.

[0088] The memory may be physical memory or implemented as a “cloud” non-transitory processor-readable medium (i.e., the one or more memory may be partially or completely based on or accessed using the network). The memory may store processor-executable code and / or information comprising the database and the application. In some embodiments, the application may be stored as a compiled application file, such as an executable file, for example, or in a structure (or unstructured) format, such as, e.g., in a non-compiled file. The application may be stored in a computer-readable format, and may, in some embodiments, further be stored in a human-readable format.279430.357_NP

[0089] In some implementations, a database may be a time-series database, a relational database, a vector database, or a non-relational database. Examples of such databases include DB2®, Microsoft® Access, Microsoft® SQL Server, Oracle®, mySQL, PostgreSQL, MongoDB, Apache Cassandra, InfluxDB, Prometheus, Redis, Elasticsearch, TimescaleDB, Chroma, Pinecone, Weaviate, and / or the like. It should be understood that these examples have been provided for the purposes of illustration only and should not be construed as limiting the presently disclosed inventive concepts. The database 30 may be centralized or distributed across multiple systems.

[0090] In one embodiment, the database may be a centralized database with a distributed backup database, a distributed database with a centralized backup database, a distributed database with a distributed backup database, or a centralized database with a centralized backup database. In one embodiment, the database abides by, or exceeds, the 3-2-1 backup best practices. In one embodiment, each backup database is maintained as a real-time backup database, e.g., the backup database may be a mirror of the database.

[0091] In some implementations, the various embodiments of the present invention may include, but is not limited to, implementations as a personal computer, a cellular telephone, a smart phone, a network-capable television set, a tablet, a laptop computer, a desktop computer, a network- capable handheld device, a server, a digital video recorder, a wearable network-capable device, a virtual reality / augmented reality device, and / or the like.

[0092] In one implementation, the network may permit bi-directional communication of information and / or data between the server system and / or the classification device of the classification system. The network may interface with the classification device and / or the server system in a variety of ways. For example, in some embodiments, the network may interface by optical and / or electronic interfaces, and / or may use a plurality of network topographies and / or protocols including, but not limited to, Ethernet, TCP / IP, circuit switched path, combinations thereof, and / or the like, as described above.279430.357_NP

[0093] In some embodiments, the network may be the Internet and / or other network. For example, if the network is the Internet, the classification device may interact with the server system via the user interface implemented on the output device and / or the input device, such as a series of web pages or private internal web pages of a company or corporation, which may be written in hypertext markup language (HTML / PHP) and may utilize one or more suitable framework (such as JavaScript, Python, Flask, Django, and / or the like), for example. It should be noted that the user interface of the classification device may be another type of interface including, but not limited to, a Windows®-based application, a tablet-based application, a mobile web interface, an application running on a mobile device, a virtual-reality interface, an augmented-reality interface, and / or the like.

[0094] The network may be almost any type of network. For example, in some embodiments, the network may be a version of an Internet network (e.g., exist in a TCP / IP-based network). In one embodiment, the network is the Internet. It should be noted, however, that the network may be almost any type of wireless network and may be implemented as the World Wide Web (or Internet), a local area network (“LAN”), a wide area network (“WAN”), a low power wide area network “LPWAN”, a LoRa network (e.g., “LoRaWAN”), a metropolitan network, a wireless network, wireless networking technology a “WiFi network”, a cellular network, a Bluetooth network, a Global System for Mobile Communications (“GSM”) network, a code division multiple access (“CDMA”) network, a 3G network, a 4G network, a long term evolution (“LTE”) network, a 5G network, a satellite network, a radio network, an optical network, a shortwave wireless network, a long-wave wireless network, combinations thereof, and / or the like. It is conceivable that in the near future, embodiments of the present disclosure may use more advanced networking topologies.

[0095] While the disclosure has been described in detail and refers to specific embodiments, it will be apparent to one skilled in the art that various changes and modifications can be made without departing from the spirit and scope of the embodiments. Thus, it is279430.357_NP intended that the present disclosure covers the modifications and variations of this disclosure, provided they come within the scope of the appended claims and their equivalents.

[0096] It is to be understood that the invention may assume alternative variations and step sequences, unless specified to the contrary. It also is to be understood that the specific devices and processes illustrated in the attached drawings and described in this specification are simply exemplary embodiments of the invention. Hence, specific dimensions and other physical characteristics related to the embodiments disclosed are not to be limiting.

[0097] It should be understood that the invention may assume alternative variations and step sequences unless specified to the contrary. The specific devices and processes illustrated in the attached drawings and described in this specification also should be understood as exemplary embodiments of the invention. Hence, specific dimensions and other physical characteristics related to the embodiments disclosed are not to be limiting.

[0098] As illustrated in Figure 1, the present invention provides variations of a novel biomedical system 1000, with associated method 2000 and device 3000 embodiments, that leverages machine learning to diagnose and treat muscular degeneration through sensing and therapeutics (actuation / stimulation) in a user-friendly and, usually, portable design. Various embodiment of the platform system 1000, device 3000, and method 2000 are designed to enhance functional regeneration of muscle following, or to treat, muscular degeneration. The underlying technologies and concepts can also be applied to various other muscle-wasting conditions.

[0099] Various embodiments of the present invention comprise the following:• a system 1000 for treating muscular degeneration comprising an actuation subsystem 100 and a sensing subsystem 200;279430.357_NP• a system 1000 for treating muscular degeneration comprising an actuation subsystem 100 and a sensing subsystem 200 connected to one another via a machine learning subsystem 300;• a method 2000 to treat muscular degeneration by processing sensing output data 210, generated from a sensing subsystem 200, through a second machine learning subsystem 400 to generate treatment parameters 120 that dictate a treatment 130 administered by an actuation subsystem 100 and obtaining actuation output data 110 from the actuation subsystem 100 and processing it 110 through a first machine learning subsystem 300 to generate sensing input data 210; and• a portable, wearable device 3000 for treating muscular degeneration comprising an actuation subsystem 100, a sensing subsystem 200, and at least one machine learning subsystem 300 in communication with both the actuation 100 and sensing subsystems 200.

[0100] As used herein, “actuation” is used as commonly understood in the field to refer to the actuation or stimulation of a muscle (generally to trigger muscle contraction). As used herein, “sensing” is used as commonly understood in the field as the ability to sense or measure an attribute of a muscle, such as, the actuation of a muscle, the physiology of the muscle tissue, or the neurophysiology of the muscle, often in response to muscle actuation. These measured muscle attributes are embodied in actuation output data 110, which becomes sensing input data 210 to be evaluated, measured, manipulated, etc. by the sensing subsystem 200. An actuation subsystem 100 includes all of the parts and / or steps necessary to achieve muscle actuation, as exemplified by the various non-liming examples described herein. A sensing subsystem 200 includes all of the parts and / or steps necessary to sense electrical activity within muscle cells, as exemplified by the various non-limiting examples described herein. Various embodiments of the present invention include one or more machine learning subsystems 300, 400 that279430.357_NP incorporate at least one central processing component or unit 700 and connected memory 710 configured to execute the steps or goals detailed herein. For ease of discussion herein, a single machine learning subsystem 300 is discussed as being in communication with both the actuation subsystem 100 and the sensing subsystem 200 (see Figure 1); however, this machine learning subsystem 300 (also referred to herein as a first machine learning subsystem 300) can be configured to handle all of the machine-learning needs of the systems 1000, methods 2000, or devices 3000 of the present invention or there can be a second machine learning subsystem 400 or a plurality of machine learning subsystems 300. Non-limiting examples of some of these technologies are illustrated in Figure 1 and discussed herein.

[0101] One embodiment of the present invention is an adaptive, closed-loop neuromuscular rehabilitation system 1000 that integrates multimodal actuation (an actuation subsystem 100) and sensing (a sensing subsystem 200) within a single intelligent treatment system 1000 (as shown in Figure 1). The system 1000 of this embodiment combines electrical and / or acoustic (ultrasound) stimulation / actuation technologies 100 with electrophysiological, mechanical and / or optical sensing technologies 200 to continuously modulate and monitor neuromuscular activity and / or neuromuscular physiology in real time. Unlike conventional neuromuscular electrical stimulation systems that operate without feedback and require clinical oversight, this embodiment of the invention delivers autonomous, data-driven therapy through dynamic adjustment of stimulation parameters 120 based on continuous physiological feedback processed by a machine learning component 300.

[0102] For one embodiment, embedded sensors 500, such as electromyography, electrical impedance myography, MEMS accelerometers for mechanomyography / inertial measurement units (“IMUs”), ultrasound and temperature quantify muscle activation, tissue health, and metabolic state, enabling precise control of energy delivery. A central processing279430.357_NP component 700 fuses the data streams from these sensors 500 using adaptive algorithms or machine learning 300 to optimize therapeutic outcomes while minimizing fatigue and discomfort. For various embodiments of the present invention, the sensors 500 can provide muscle actuation, muscle sensing, either, or both. The architecture of the system 1000 is modular, scalable, and wireless, supporting both wearable and implantable configurations for in-clinic or at-home use. By uniting actuation 100 and sensing 200 in one adaptive system 1000, this platform transforms neuromuscular therapy from a static, technician-dependent process into a personalized, autonomous, and continuously optimized rehabilitation system 1000 capable of accelerating functional recovery and reducing long-term disability.

[0103] Various embodiments of the system 1000, method 2000, and device 3000 continuously compare multimodal sensing data (sensing output data 220) from each subsequent therapy session (or treatment 130) to prior sessions (or treatments 130), identifying trends in muscle activation, impedance, stiffness, and perfusion. As shown in Figure 2, if progressive changes are sensed 2100, consistent with muscle atrophy or reduced contractile output are detected, the machine learning engine 300 classifies the state and updates predictive models 2200, 2300. Based on these insights, the system 1000, method 2000, and device 3000 (via at least one machine learning subsystem 300) recommend adjustments to stimulation amplitude, frequency, or timing (treatment parameters 120) of the next / subsequent session (steps 2800 to 2900 of Figure 2) to optimize recovery and prevent overuse. This session-to-session adaptation enables a predictive, data-driven rehabilitation loop (shown in Figure 2) that evolves with the patient’s physiology (or neuromuscular physiology.)

[0104] Figure 2 illustrates one embodiment of a method 2000 of the present invention that can be incorporated into various systems 1000 and devices 3000 of the present invention. This embodiment of a method 2000 comprises the following steps:279430.357_NP• Step 0 2050: check the electrode-skin impedance (i.e., verify proper placement and contact quality of the electrodes 500 on the skin 1);• Step 1 2100: employ a sensing component (or sensing subsystem 200) to process sensing input data 210 (which generally is actuation output data 110) and produce sensing output data 220;• Step 2 2200: feature extraction and preprocessing for things like but not limited to digital signal processing (“DSP”) filtering and normalization (DSP executed by a digital signal processor is one method to filter out noise and artifacts from raw biosignals (actuation output data 110)) before ML learning stages;• Step 3 2300: machine learning / control subsystem 300 processing 2300 for model prediction and adaptive control to produce instructions (treatment parameters 120) for an actuation component or subsystem 100;• Step 4 2400: employ an actuation subsystem 100 as instructed by the output (treatment parameters 120) of the machine learning subsystem 300 - this can include actions including but not limited to charge injection safety checks 2500;• Step 5 2600: sensing the muscle / tissue response 110 to the actuation (contractions, perfusion, stiffness change, as non-limiting examples), which is done by the sensing subsystem 2100;• Step 6 2700: Gathering and processing actuation session output data 110 to use as baseline measurements, measurements of prior sessions, and / or training data for the machine learning subsystem 300;• Steps 7 & 8A & 8B 2800, 2810, 2820: Sensing for muscular degeneration 2800 and if muscular degeneration is detected 2810, the machine learning subsystem 300 recommends actuation adjustments, such as adjustments to treatment parameters 120, such as amplitude, frequency, duty. If atrophy is not detection279430.357_NP2820, then the machine learning subsystem 300 recommends a maintenance of current parameters 120 (or in some cases a trailing off or decreasing of actuation parameters 120); and• Step 9 2900: the machine learning subsystem 300 takes the information from Steps 7, 8A, and / or 8B to compute the parameters 120 of the next treatment session 130, which can be fed back into the method 2000 at Step 3.

[0105] Figures 3A and 3B provide some background on exogenous electrical stimulation and sensing. Figures 3A and B show that electrical stimulation and the bio- electronic sensing modalities have been explored as separate platforms. However, to date, they have not been combined into one system as per the present invention. As shown in Figure 3 A, exogenous electric fields have been used to modulate the physiology of a variety of tissues and cells both in vivo and in vitro, including skeletal muscle, nervous tissue, and immune cells. Additionally, electrodes can sense various bioelectronic indicators, such as safe charge injection, muscle function through EMG recordings, and overall tissue composition using bioelectronic impedance spectroscopy. Figure 3B illustrates exogenous electrical stimulation and sensing. As shown in Figure 3B, exogenous electric fields have been used to modulate a variety of tissues both in vivo and in vitro, including skeletal muscle, nervous tissue, and immune cells. Additionally, electrodes can sense various bioelectronic indicators, such as safe charge injection, muscle function through EMG recordings, and overall tissue composition using bioelectronic impedance spectroscopy.

[0106] Currently, muscular degeneration is often treated with physical therapy, autologous muscle transfer, or orthotic braces, none of which regenerate lost muscle tissue. Additionally, physical therapy frequently is not started until several weeks after the injury has occurred. Figure 4 illustrates the standard timeline for muscle treatment and regeneration after injury or the onset of illness. As shown, physical therapy does not start until weeks after the279430.357_NP injury or illness and often after secondary muscular degeneration and fibrosis set is, which means that muscular degeneration already has begun before a physical therapist sees the patient to take baseline muscle measurements and to start testing for muscular degeneration. This standard procedures allows for continued degradation of the muscle before muscular degeneration is tested or watched for. The system 1000 and device 3000 broadly illustrated in Figure 1 and the method 2000 of Figure 2 enable the sensing of muscular degeneration much earlier than traditional treatment protocols and enable better informed treatment due to the facts that the patient can employ many embodiments of the present invention outside of a medical center for more frequent and consistent treatment and because the need for treatment is sensed earlier than standard protocols and because sensing and actuation are embodied in a single system and in communication with one another.

[0107] Systems 1000, devices 3000, and methods 2000 of the present invention integrate both stimulation / actuation 100 and sensing 200 to influence and monitor tissue integrity, physiology, and / or neurophysiology and to better inform treatment protocols 120. Part of the platform’s novelty is the implementation of these stimulation 100 and sensing 200 modalities together and integrating these features through machine learning 300. The systems 1000, devices 3000, and methods 2000 of the present invention create advanced therapeutic and diagnostic devices to identify and treat muscular degeneration, leveraging state-of-the-art biomedical engineering, material science, and machine learning.

[0108] The various embodiments of the present invention system 1000, devices 3000, and methods 2000 combine existing actuation components 100 with existing sensing components 200. Non-limiting examples of the existing technologies that can be incorporated in various embodiments of the system follow.

[0109] Various embodiments of the present invention systems 1000, devices 3000, and methods 2000 integrate multimodal actuation 100 and sensing 200 technologies within a single279430.357_NP closed-loop architecture (see Figures 1 and 2) designed to restore, modulate, and monitor neuromuscular function and physiology. Unlike existing stimulation systems 100 that operate as open-loop devices without physiological feedback, the system 1000, devices 3000, and methods 2000 of the present invention combine electrical, mechanical, acoustic, and optical actuation modalities (all actuation modalities 100) with real-time electrophysiological, mechanical, and optical sensing modalities 200 creating a continuously adaptive therapeutic ecosystem deployable in both clinical and home environments.

[0110] Actuation (Stimulation) Subsystem 100

[0111] The actuation subsystem 100 supports a range of energy-delivery methods (electrical, mechanical, bioelectrical or heat) to achieve precise, patient-specific modulation of muscle and nerve activity. Core actuation technologies 100 include, but are not limited to, neuromuscular electrical stimulation and functional electrical stimulation (“FES”) for evoked contractions, low-intensity pulsed ultrasound (“LIPUS”) and focused ultrasound (“FUS”) for mechanical and bioelectrical modulation. Each modality can be configurable by various treatment parameters 120, including but not limited to, delivery mode (surface, percutaneous, or implantable), waveform characteristics, and therapeutic purpose ranging from contractile activation and trophic support to fibrosis reduction and tissue remodeling.

[0112] Unlike conventional NMES systems that require manual tuning and clinical supervision, the system’s actuation subsystem 100 can incorporate closed-loop adaptive control. Energy output, pulse pattern, and stimulation timing are automatically modulated based on real-time sensor feedback, enabling safe and personalized therapy that minimizes fatigue while maximizing regeneration efficiency.

[0113] Sensing Subsystem 200

[0114] The sensing subsystem 200 integrates electrophysiological, mechanical, optical, thermal, and / or pressure-based modalities to quantify neuromuscular state and / or physiology279430.357_NP and therapeutic response (sensing output data 220). Electrical impedance myography and electromyography can provide information on muscle activation, conductivity, and tissue composition, while mechanomyography and inertial measurement units can capture contraction mechanics, vibration profiles, and limb movement patterns (sensing output data 220.) Ultrasound imaging is used to assess tissue stiffness, architecture, and perfusion, and pressure sensors monitor local contact forces or joint loading during stimulation and rehabilitation (sensing output data 220.) In parallel, optionally, temperature sensors track localized thermal changes associated with perfusion, inflammation, or therapeutic ultrasound exposure, offering an additional layer of physiological feedback for adaptive control (output sensing data 220.)

[0115] These sensors 500 are spatially co-located or integrated within the electrode arrays 510 (see Figure 24B) to generate multimodal feedback signals 110, allowing for both real-time correction of stimulation parameters 120 and long-term tracking of tissue recovery. This integration transforms isolated diagnostic tools into an intelligent feedback network that informs both patient-guided use and clinician oversight.

[0116] Additional Non-Limiting Examples of Actuation 100 and Sensing 200 Technologies / Platforms / Components / Subsystems.

[0117] The closed-loop neuromuscular stimulation and sensing systems 1000 and devices 3000 can be built using a combination of commercially available actuation 100 and sensing 200 components that together create a robust, scalable platform for both research and translational applications. For neuromuscular electrical stimulation 100, several off-the-shelf drivers and systems can generate the necessary biphasic pulses with current and voltage control. Modular platforms such as the Chattanooga 77717 Intelect NMES Digita (https: / / sourceortho.com / product&,'chattanooga-intelect-nmes-digital- unit?currency::::USD&variant:::49418820747561 &stkn::::6b0fea9e3741 &utm (last accessed279430.357_NPOctober 22, 2025) and the Chattanooga Primera TENS / NMES Complete(https: / / www.usmedrehab.con products / primera-tens-nmes-October 22, 2025) allow direct integration with wearable electrodes 500, while compact integrated drivers such as Texas Instruments’ DRV8662 (ti.com / product / DRV8662, last accessed on October 18, 2025) and Analog Devices’ ADuCM350 (analog.com / aducm350 last accessed on October 18, 2025) provide high-efficiency voltage and current regulation for embedded applications. For more advanced control or safety validation, STMicroelectronics’ STIM32 (st.com last accessed on October 18, 2025) reference design can be adapted to textilebased electrode arrays 510. These example technologies provide the foundation for controllable, safe NMES actuation 100 directly from a wearable hub 3000.

[0118] For ultrasound actuation 100, research-grade platforms are available but are not yet fully miniaturized for wearable integration. To connect an ultrasound actuation subsystem 100 to a sensing and actuation system 1000 or wearable device 3000 of the present invention, the embodiment of the system 1000 or device 3000 is configured with a standard port or plug to allow the system 1000 or device 3000 to connect to a standard ultrasound platform such as the following. The Verasonics® Vantage Research Ultrasound System (verasonics.com last accessed on October 18, 2025) and Sonic Concepts H-148 and H-115 transducers (sonicconcepts.com last accessed on October 18, 2025) provide full programmability for laboratory studies. Meanwhile, PiezoDrive’s PD200 and PD32 amplifier modules (piezodrive.com last accessed on October 18, 2025) and Boulder Imaging’s LIPUS system (boulderimaging.com last accessed on October 18, 2025) are examples of compact driver electronics suitable for low-intensity pulsed ultrasound therapy research. These can serve as reference implementations for integrating acoustic actuation 100 into a wearable device 3000.279430.357_NP

[0119] Mechanical and vibrational actuation components 100 are widely available and easily integrated into wearable devices 3000 for complementary stimulation or tactile feedback. Accelerometers 3050 for mechanomyography can be incorporated into the various embodiments of the present invention as can pressure and temperature sensors 3060. Precision Microdrives (precisionmicrodrives.com last accessed on October 18, 2025) offers a range of linear resonant actuators (“LRAs”) and eccentric rotating mass (“ERM”) motors that can provide localized vibration therapy or mechanostimulation. TDK PiezoHapt (product.tdk.com last accessed on October 18, 2025) and HaptX tactors (haptx.com last accessed on October 18, 2025) deliver high-bandwidth haptic actuation suitable for low-frequency mechanical stimulation or sensory feedback. Soft pneumatic networks, such as PneuAct’s air-driven actuators (pneuact.com last accessed on October 18, 2025) can also be integrated for intermittent compression or tissue mobilization applications.

[0120] For the sensing subsystem or component 200, a broad range of high-fidelity, wearable-compatible sensors 500 are commercially available. Electromyography sensing can be achieved using modules such as Delsys Trigno (delsys.com last accessed on October 18, 2025) or open-source options like Myoware 2.0 (https: / / w^fw.sparkfun.com / catalogsearch / result / ?q=:myoware+2.0 last accessed on October 18, 2025) and Olimex EMG Shields (olimex.com last accessed on October 18, 2025). These technologies are based on analog front ends like Texas Instruments’ ADS 1299 (ti.com / product / ADS1299 last accessed on October 18, 2025) and ADS1292R (ti.com / product / ADS1292R last accessed on October 18, 2025), which offer high-resolution, low-noise acquisition suitable for integration with flexible electrodes. Electrical impedance myography can be implemented using impedance analyzers such as Analog Devices’ AD5940 (analog.com / ad5940 last accessed on October 18, 2025) or ADuCM355, both capable of multi-279430.357_NP frequency impedance spectroscopy from 1 Hz to 200 kHz. These chips can be combined with textile electrodes to assess muscle composition, hydration, and fibrosis in real time.

[0121] To measure mechanomyography and motion, low-power MEMS accelerometers such as the Analog Devices ADXL1001 (analog. com / adxllOOl last accessed on October 18, 2025), TDK ICM-42688 (invensense.tdk.com last accessed on October 18, 2025), and Bosch BNO055 (bosch-sensortec.com last accessed on October 18, 2025) can be laminated into the fabric at muscle bellies. These sensors 500 detect minute vibrations and motion patterns associated with muscle contraction 110, complementing electrical sensing. Force, pressure, and strain sensing can be implemented using thin-film and textile-integrated components such as Tekscan FlexiForce A201 (tekscan.com last accessed on October 18, 2025) and Interlink FSR 400 (interlinkelectronics.com last accessed on October 18, 2025), as well as BeBop Sensors SmartFabric (bebopsensors.com last accessed on October 18, 2025) and StretchSense capacitive strain sensors (stretchsense.com last accessed on October 18, 2025) for distributed pressure and stretch mapping. Temperature monitoring can be integrated with Melexis MLX90614 infrared thermometers (melexis.com last accessed on October 18, 2025) or TDK NTC thermistors (product.tdk.com last accessed on October 18, 2025), both proven in wearable designs.

[0122] For ultrasound sensing and imaging, portable diagnostic-grade modules such as Clarius HD3 (clarius.com last accessed on October 18, 2025) and Butterfly iQ+ (butterflynetwork.com last accessed on October 18, 2025) provide handheld imaging for muscle assessment, though these are not yet miniaturized for wearable integration and would, therefore, require a port connection as discussed previously.

[0123] Optical and perfusion sensing for oxygenation or blood flow can be achieved using Maxim Integrated MAX30101 (maximintegrated.com last accessed on October 18, 2025) or ams AS7038RB (ams.com last accessed on October 18, 2025), both of which are low-279430.357_NP power reflectance photoplethysmography sensors compatible with flexible PCBs and fabric- embedded modules. High-precision inertial measurement units such as STMicroelectronics LSM6DSO32 (st.com last accessed, on October 18, 2025) or TDK ICM-20948 are readily available to capture movement, orientation, and posture data for functional tracking.

[0124] Control and communication are handled through commercially available microcontrollers and wireless modules. Real-time control and safety monitoring can be managed by STMicroelectronics STM32H7 (st.com last accessed on October 18, 2025), Nordic nRF52840 (nordicsemi.com last accessed on October 18, 2025), or Espressif ESP32- S3 (espressif.com last accessed on October 18, 2025) for low-latency BLE and Wi-Fi connectivity.

[0125] Safety and isolation are ensured by incorporating Analog Devices ADuM6000 (analog.com) or Texas Instruments ISO124 (ti.com / product / ISO124 last accessed on October 18, 2025) isolation amplifiers and current-limiting watchdog circuits to enforce charge injection limits and user protection for some embodiments.

[0126] While many actuation and sensing components are already commercially available, miniaturized ultrasound actuation and sensing modules are not yet optimized for integration into wearable textiles. To ensure compatibility with this modality, the proposed system can be embodiments in a wearable device 3000 that includes at least one dedicated auxiliary port for connection to external or in-clinic ultrasound actuation systems. This allows clinicians to perform high-resolution ultrasound-assisted therapies using the same platform architecture until fully wearable ultrasound modules become available.

[0127] The above-detailed available technologies are intended to serve as a non- exhaustive list of possible sensing and actuation technologies that can be incorporated into sensing 200 and / or actuation 100 subsystems and incorporated into various systems 1000 or device 3000 of the present invention.279430.357_NP

[0128] Machine Learning (“ML”) Subsystem 300 and 400

[0129] Various embodiments of the present invention incorporate at least one machine learning subsystem 300. For some embodiments, there is one machine learning subsystem (a first machine learning subsystem 300) that processes the output data 110 and 220 from both the actuation 100 and sensing 200 subsystems. For other embodiments, there may be two or more different machine learning components (a first machine learning subsystem 300 and a second machine learning subsystem 400) each of which processes either the output data 110 from the actuation subsystem 100 or the output data 220 from the sensing subsystem 200. The numerical designation of a “first” and a “second” machine learning subsystem merely means two subsystems instead of one. The machine learning subsystem(s) 300 can be configured to take the actuation output data 110, process the actuation output data 110 into sensing input data 210 (for configurations that require that actuation output data 110 be processed in some form prior to being entered into the sensing subsystem 200 as sensing input data 210, and feed that sensing input data 210 into the integrated sensing subsystem 200, wherein the sensing input data 210 is processed to determine whether or not and how to adjust the treatment parameters 120 of the next actuation session 130 in response to what is learned from the data (see Figure 2).

[0130] One embodiment of the data flow within a closed-loop neuromuscular stimulation system 1000, method 2000, or device 300 operates through a four-part cycle encompassing an actuation subsystem 100, a machine learning subsystem 3000, a sensing subsystem 200, and a second ML inference stage 400. Each subsystem performs distinct functions that collectively enable adaptive, data-driven modulation of stimulation parameters 120 and continuous monitoring of neuromuscular recovery.

[0131] Within this embodiment, the actuation subsystem 100 begins the cycle by executing stimulation commands (treatment parameters 120) determined by the ML controller279430.357_NP700. The actuation subsystem 100 receives treatment parameters 120 such as pulse amplitude, width, frequency, duty cycle, electrode mapping, and ultrasound intensity, all constrained by user-specific safety limits and therapeutic goals. Within this actuation subsystem 100, signal generation hardware produces controlled biphasic or kilohertz-modulated electrical pulses, while power conditioning circuitry ensures charge balance and patient safety. These stimulation signals are delivered to the target tissue through electrodes or acoustic transducers 502, and internal instrumentation such as voltage, current, and impedance sensors 504 monitors the accuracy and stability of delivery. For ease of discussion herein, actuation mechanisms or devices 502 and sensing devices 504 are collectively referred to as “sensors 500” (and interchangeably referred to as “electrodes 500”). In some embodiments, both actuation and sensing can be achieved through a single sensor 500; however, other embodiments may employ different sensors 500 for actuation 502 and sensing 504. The actuation outputs 110 from this stage can include the actual stimulation waveform, a log of parameters used, basic physiological response data, and / or safety validation signals. This information is then relayed to the ML subsystem 300 to confirm execution fidelity and inform subsequent control or parameter 120 updates.

[0132] For this embodiment, the machine learning control subsystem 300 processes actuation telemetry 110 along with prior session data 2700 to determine the next optimal stimulation command set 2900 (treatment parameters 120). Within this layer, a control policy engine can be implemented through reinforcement learning or model predictive control balances therapeutic effectiveness against fatigue and comfort constraints. Contextual awareness allows the system 1000, method 2000, or device 3000 to adapt stimulation patterns according to therapy phase, such as warm-up or recovery, while an embedded safety layer can enforce charge, temperature, and current limits before actuation. The ML model 3000 also can predict expected physiological responses 220, such as contraction magnitude or fatigue onset,279430.357_NP based on historical dose-response relationships. The resulting outputs are updated stimulation parameters 120, confidence metrics, and safety flags that drive the next actuation cycle (treatment 130.)

[0133] The sensing subsystem 200 then captures the body’s physiological and mechanical response to stimulation 110, 2600. Multiple sensor modalities 200, including electromyography, electrical impedance myography, mechanomyography, inertial measurement units, pressure and force sensors, and optional temperature or ultrasound probes, collect data reflecting activation strength, tissue composition, stiffness, and perfusion. These analog biosignals undergo digital signal processing (“DSP”), filtering to remove motion artifacts and electrical noise, followed by feature extraction steps 2200 such as calculating EMG amplitude, impedance phase angle, MMG envelope energy, or IMU-derived motion smoothness. The system 1000, method 2000, or device 3000 normalizes these features against baseline or prior-session data to quantify improvement or deterioration in muscle function 2200 The outputs 220 from this subsystem include structured multimodal feature vectors, signal quality indicators, and event flags such as detected contraction or fatigue onset, which are passed to the ML inference layer 2300.

[0134] Finally, the ML inference and adaptation subsystem 300 (step 2300 in Figure 2) processes the extracted features, actuation logs, and historical data to interpret the current neuromuscular state and predict future needs. This subsystem 300 performs state estimation to assess muscle activation, fatigue level, stiffness, and perfusion; anomaly detection to identify electrode misplacement or atrophy progression; and adaptive learning to refine personalized stimulation parameters 120. Through model updates, it generates predictions of optimal stimulation intensity, frequency, and duration for the next cycle or future sessions (step 2900 of Figure 2.) It also produces progress metrics such as fatigue index or muscle recovery trend279430.357_NP and transmits summarized results to a clinician dashboard or mobile application for remote monitoring.

[0135] There are a wide variety of existing machine learning platforms 300 that can be trained on muscular degeneration data and can be incorporated into the systems 1000, methods 2000, and wearable devices 3000 of the present invention. The following is a discussion of a few non-limiting examples of how to incorporate a machine learning component as a subsystem 300 into an embodiment of the present invention.

[0136] Machine learning inference can be executed on compact embedded platforms such as NVIDIA® Jetson Nano (developer.nvidia.com / embedded last accessed on October 18, 2025), Google Coral Edge TPU (coral. ai last accessed on October 18, 2025), or TensorFlow Lite Micro running on microcontrollers.

[0137] The machine learning component or subsystem 300 of the closed-loop neuromuscular platform serves as the analytical and control hub, integrating sensing data and actuation commands (actuation output data 110, sensing input data 210, sensing output data 220, and treatment parameters 120) to personalize therapy in real time and across sessions 130. Embodiments of the ML subsystem 300 can be built from existing edge-AI components. Practical hardware options include a wide variety of computer processing units 700 and associated memory 710, including but not limited to, compact AI compute modules such as NVIDIA® Jetson® (Nano or Orin), Raspberry Pi® 5 paired with a Google® Coral Edge TPU, Qualcomm® RB5, NXP i.MX RT, or low-power microcontrollers such as the STM32H7 or nRF52 series for TinyML® implementations. These can run lightweight ML runtimes including TensorFlow Lite / Micro, PyTorch Mobile, ONNX Runtime, or CMSIS-DSP for efficient classical signal processing and model inference. Model development can leverage tools like Edge Impulse, TensorFlow® Model Maker, or scikit-learn / XGBoost models compiled for embedded deployment. The sensing stack includes biopotential analog front ends279430.357_NP(ADS129x) for EMG, impedance analyzers such as AD5940 or ADuCM355 for EIM, MEMS IMUs for MMG and motion, pressure and force sensors, thermistors for temperature monitoring, and ultrasound front ends for imaging or elastography. Together, these components enable a research-grade implementation that can later be hardened for clinical translation under regulatory standards such as IEC 60601, ISO 13485, and ISO 14971.

[0138] One embodiment of the system 1000, method 2000, or device 3000 architecture follows an edge-first design with optional cloud support. Data acquisition involves synchronized multimodal inputs: EMG, EIM, MMG, IMU, pressure, temperature, and ultrasound. Artifact removal is handled via stimulus-synchronous blanking windows and adaptive filters to eliminate NMES-induced electrical noise. Real-time feature extraction occurs at 50-250 Hz and includes time- and frequency-domain metrics such as EMG RMS, zero-crossings, and median frequency; EIM phase angle and Cole-Cole parameters; MMG and IMU envelope and latency features; ultrasound stiffness and texture; and pressure or temperature trends. These multimodal features are fused by a predictive model initially a classical machine learning algorithm such as Random Forest® or XGBoost for interpretability, progressing to temporal deep learning models like ID-CNNs or GRUs with attention mechanisms for multi-signal weighting. Personalization is achieved through transfer learning layers that adapt to each patient’s physiological profile. The control policy operates through constraint-aware model predictive control (“MPC”) or safe reinforcement learning, which recommends stimulation parameters bounded by safety thresholds. A dedicated safety layer enforces limits on charge density, pulse amplitude, and temperature, ensuring compliance with Figure 2, Step 2500 of the system flow (charge injection safety). Learning occurs locally on- device for low-latency adaptation, with optional cloud-assisted retraining or federated updates to enhance population-level generalization without sharing raw data.279430.357_NP

[0139] The ML subsystem 3000 trains on multimodal time-series data and actuation command histories. Inputs include all sensed features and the full record of delivered stimuli (amplitude, frequency, pulse width, duty cycle, waveform type, and ultrasound power (treatment parameters)). Targets include immediate outcomes such as torque, EMG / MMG amplitude, EIM phase angle, tissue stiffness, fatigue metrics, and subjective comfort as well as session-level indicators like strength recovery, symmetry, ultrasound elastography changes, and clinician-annotated atrophy or fibrosis. The training strategy combines supervised learning for dose-response modeling, self-supervised pretraining for unlabeled time-series, and continual learning for personalization. Performance is evaluated via RMSE for force prediction, AUROC for atrophy or fatigue classification, and uncertainty metrics to inform confidence-weighted control.

[0140] In operation, data flow is bidirectional. From actuation 100 to ML 300 to sensing 200, the ML model interprets the electrical and acoustic stimulation parameters as “dose variables” and correlates them with physiological responses to leam individualized doseresponse curves. It predicts contraction quality, fatigue risk, comfort level, and regeneration likelihood, dynamically adjusting which sensing modalities to prioritize. For example, if fatigue is detected, the system increases EMG / MMG sampling frequency; if fibrosis risk is indicated, it triggers additional ultrasound or EIM scans. Conversely, from sensing 200 to ML 300 to actuation 100, the system 1000 interprets real-time sensor outputs (EMG, EIM, MMG, temperature, pressure, ultrasound) to estimate physiological state variables — activation level, fatigue index, impedance shifts, and perfusion proxies and uses these to classify risk (atrophy, overuse, overheating, or poor electrode contact). The ML control layer 2300 then optimizes actuation parameters 120 within safety constraints, adjusting amplitude, frequency, ramp time, or electrode selection as needed. If uncertainty is high, it defaults to conservative changes or requests clinician confirmation.279430.357_NP

[0141] The ML component or platform 300 operates in two temporal modes. In realtime loops (100-250 ms), features are extracted, predictions are made, stimulation parameters are updated, and safety checks (Step 2500 of Figure 2) are executed before delivery. In session- to-session adaptation, aggregate metrics such as EIM phase angle or EMG amplitude are compared against baseline and historical trends to detect atrophy or maladaptive remodeling (step 2700 of Figure 2). The system 1000, method 2000, or device 3000 then predicts optimal actuation adjustments and recommends the timing of the next therapy session (step 2900 of Figure 2), accounting for recovery dynamics and fatigue.

[0142] Reliability and transparency are ensured through multiple layers. Artifact suppression and robust feature extraction maintain data integrity; uncertainty-aware control prevents unsafe adjustments; explainability tools (e.g., SHAP values or attention maps) make decisions interpretable to clinicians (“phase angle decreased 7%, predicted fatigue increased; reducing frequency and extending rest interval”). Privacy is preserved through on-device inference and optional federated learning for population updates. The result is a self-adaptive, patient-specific rehabilitation engine that evolves with use, continuously optimizing stimulation parameters while maintaining safety and transparency.

[0143] A functional minimum viable implementation can be achieved using Edge Impulse® with TensorFlow® Lite running on Jetson, Coral, or STM32 hardware. This configuration can stream multimodal signals, leam a per-user dose-response map, automatically adjust stimulation within predefined safety limits, and track longitudinal recovery metrics to detect early signs of atrophy. Over time, the ML subsystem transforms the platform from a static stimulator into an intelligent, predictive rehabilitation system capable of autonomous optimization and clinician oversight.

[0144] One embodiment of a system 1000, method 2000 or wearable device 3000 of the present invention is shown in Figure 24, with details of the machine learning hardware and279430.357_NP software. The actuation subsystem 100 and the sensing subsystem 200 act as input devices and data sources sending data to signal preprocessing component 310 which filters the data, removes artifacts, denoises, etc. Then the data undergoes feature extraction 320 and other processes shown in Figure 24 before being converted to treatment parameters 120 and sent to the output devices 330 in the actuation subsystem 100. The machine learning subsystem 300 includes prediction, personalization and uncertainty estimation 340, which can be used with safety guardrails 360 and other guidelines to help determine the updated treatment parameters 120 that go to the output devices 330. The session data also can be stored 370 and used by a recommendation engine 380 to determine the treatment parameters 120 of a subsequent treatment session 130. The minimum hardware and software necessary to run the machine learning subsystem 300 includes a processor 700, memory 710, input / output interfaces 720, storage 730, and power management equipment 740.

[0145] System Integration and Scalability

[0146] At the system 1000 level, a unified control processor (computer processing unit 700) fuses data across sensing modalities 200, applies machine-learning-based algorithms 300 to identify optimal stimulation parameters 120, and adjusts output through a closed-loop feedback mechanism. Data can be communicated through a wireless interface to a clinician or user dashboard, supporting remote monitoring, adaptive recalibration, and longitudinal data analytics.

[0147] The architecture is inherently modular and scalable, allowing each actuation 100 or sensing 200 modality to be selectively engaged depending on indication, recovery phase, or deployment setting. In a home environment, surface electrodes with embedded sensors 500 can deliver self-calibrating stimulation guided by simplified user interfaces. In clinical or implantable configurations, high-density electrode arrays 510 and embedded processors provide precision-level control for complex neuromuscular injuries.279430.357_NP

[0148] This platform fills a critical void in current rehabilitation technologies by bridging the gap between stimulation and sensing, transforming muscle therapy from a static, technician-dependent intervention into a dynamic, data-driven, and autonomous system. By coupling multimodal energy delivery with multimodal feedback, it enables personalized, quantitative, and continuously optimized rehabilitation that can adapt to changing physiological conditions over time. The system’s extensible framework supports future integration of emerging biosensing materials, Al-driven control algorithms, and wearable or implantable interfaces, positioning it as a next-generation therapeutic ecosystem for neuromuscular recovery, remote rehabilitation, and regenerative medicine.

[0149] Session-to-Session Adaptive Intelligence (step 2700 of Figure 2)

[0150] Beyond real-time closed-loop modulation, the system incorporates a longitudinal comparative learning framework that evaluates neuromuscular metrics across sessions. After each therapy session, the sensing subsystem 200 collects multimodal data 110, 210 — including electrophysiological (EMG, EIM), mechanical (MMG, force, pressure), optical (ultrasound, stiffness), and thermal (temperature, perfusion) parameters — which are archived and compared to baseline and prior sessions. For some embodiments, the actuation output data 110 is fed directly into the sensing subsystem 200 as sensing input data 210 (becoming one-in-the same), while in other embodiments, the actuation output data 110 is processed by the machine learning subsystem 300 before becoming sensing input data 210.

[0151] The machine learning engine 300 analyzes trends in these metrics to detect early indicators of muscle atrophy, fatigue, or maladaptive remodeling. Key indicators include a progressive reduction in phase angle (EIM), contraction amplitude (EMG / MMG), or mechanical output (force, stiffness). Deviations beyond statistically learned thresholds trigger the classification of a potential atrophic trajectory.279430.357_NP

[0152] In response, the control algorithm predicts and recommends optimized actuation adjustments 120 — such as modified stimulation amplitude, frequency, or duty cycle — to restore optimal muscle activation while preventing overstimulation. The system also infers the ideal timing 120 of subsequent therapy sessions 130, accounting for recovery dynamics, muscle responsiveness, and fatigue accumulation.

[0153] Together, this cross-session analysis transforms the platform into a predictive, adaptive rehabilitation ecosystem, capable not only of reacting to immediate physiological signals but also of learning and evolving with the user’s long-term recovery profile.

[0154] Wearable Devices 3000

[0155] Various embodiments of the present invention systems 1000 and methods 2000 can be integrated into a wearable device 3000. The size, structure, material, and design of a wearable device 3000 is determined by at least the following factors: (i) the body part or muscle to be treated with the device; (ii) the need to accommodate the hardware necessary for the three components / subsystems of the present invention; (iii) the number and dimensions of the electrodes (or other actuation technology) incorporated into the wearable device (iv) any regulatory or patient safety guidelines. Further factors influencing the design of the wearable device 3000 include biomechanical fit, signal integrity, user comfort, manufacturability, and environmental durability. The biomechanical properties of the wearable device 3000 must accommodate movement without disrupting electrode-skin contact or altering signal pathways. For example, fabric elasticity, seam placement, and tension distribution should allow freedom of motion while maintaining consistent pressure at electrode sites. In some embodiments, dynamic zones or elastic knit structures may be incorporated around joints or muscle bellies to prevent motion artifacts during flexion or extension.

[0156] Signal quality and electrical performance also guide design. The spatial arrangement of electrodes / sensors 500, trace routing, and shielding must minimize crosstalk,279430.357_NP noise pickup, and impedance variability caused by body motion or sweat accumulation. The geometry of conductive pathways may be optimized using computational modeling to ensure uniform current distribution across the targeted tissue. Similarly, thermal and moisture management properties are considered to prevent overheating or irritation during prolonged use.

[0157] User comfort and ergonomics are additional determinants. Since the device 3000 may be worn for extended therapy sessions or throughout daily activities, materials are selected for breathability, flexibility, and skin compatibility. Soft conductive fabrics, medicalgrade silicones, and hydrogels may be used to improve adhesion and reduce pressure points. Weight distribution and placement of embedded electronics, such as processing modules and power supplies, are balanced to avoid localized discomfort or mechanical strain. In some embodiments, modular or detachable pods may be employed to permit washing or swapping components.

[0158] As non-limiting examples, a wearable device 3000 of the present invention could be as small as a small bandage to treat, for example, muscles in a finger or toe. Another embodiment could be a wearable cuff 3000 sized to wrap around an arm or a leg (Figures 22A through 23). Another embodiment could be a piece of close-to-the-body clothing, like a snug shirt or pants that ensure that the electrodes 500 maintain physical contact with the skin over a larger area, like the back.

[0159] Figures 22A through 22C provide one non-limiting example of a sensor array 510 that can be embodied as a compressive cuff 3000, which can be placed around various parts of a patient’s body, including arms and legs. Figure 23 shows an embodiment of a cuff 3000 customized to connect to the user’s body via a belt 3160 and two smaller securing mechanisms 3170 that encircle the user’s leg to hold the array 510 against the user’s skin 1 on the body part to be treated. The cuffs 3000 can be of various sizes and can be adjustable to279430.357_NP accommodate arms and legs of different circumferences. The following descriptions of a specific embodiment of a cuff wearable device 3000 can be implemented via the cuff design of Figures 22A or Figure 23 or any analogous cuff design known to those in the art.

[0160] Figure 22A shows one embodiment of a compressive cuff wearable device 3000 having a functional display 3010 and an optional battery button 3020 on the exterior 3005 of the cuff. Figure 22B shows the interior of the cuff 3000 of Figure 22A with an array 510 of electrodes 500 (again, electrodes 500 and sensors 500 are used interchangeably in this description and can be configured for sensing, actuation, either or both depending upon the embodiment).

[0161] Skin Interface Layer 3040: Figure 22B and 22C show the wearable cuff device 3000 open showing the interior of the cuff. The innermost layer 3040 of this one embodiment of a wearable device 3000 shown in Figures 22B and 22C directly contacts the skin 1. This layer consists of breathable, hypoallergenic, and moisture-wicking compression fabric (for example, a nylon-spandex blend) that maintains stable contact pressure between the electrodes 500 and the skin surface 1. Integrated within this layer are soft conductive electrode pads 500 composed of hydrogel, conductive textiles, or PEDOT:PSS-coated fibers that provide both stimulation and sensing functionality.

[0162] Sensing and Electrode Layer 3120: Above the skin interface are multiple sensor arrays 510 (see Figs. 22B and 22C) integrated into flexible, stretchable substrates. These include tetrapolar electrodes 500 for electromyography and electrical impedance myography, MEMS-based accelerometers 3050 for mechanomyography, thin-film pressure and temperature sensors 3060, and optional piezoelectric ultrasound patches for imaging or therapeutic ultrasound delivery. Each sensing element 500 is positioned over an anatomically relevant muscle or muscle group, ensuring localized signal acquisition and targeted stimulation. As shown in Figure 22B, each electrode 500 in the array 510 can be used as sensory or279430.357_NP stimulation based on clinicians’ desire. For stimulation, electrodes 500 in array 510 can be “shorted” to produce re-configurable cathode / anode. During stimulation, the system 1000 (as embodied in the wearable 3000) can display charge injection rate. In one embodiment, the system 1000 can use a three-electrode cells to calculate voltamulation. If potential is close to the water-splitting window, a warning can show and the system 1000 will stop stimulation. For one embodiment of sensing, a four electrode impedance acquired from 500,000 to 5000 Hz assay called electrical impedance myography can be performed using multiple orientations of electrodes 500. For the embodiment shown in Figure 22B, the fourth row of electrodes 500 is critical because four independent linear electrodes 500 are required for the EIM sensing capability.

[0163] Conductive Pathway Layer 3110: Interwoven conductive fibers or printed metallic traces 3110 (silver-coated nylon, graphene- or copper-based inks, or liquid metal microchannels) form flexible circuits that connect electrodes 500 and sensors 500 to the garment’s central electronic hub 3130. These traces 3110 follow serpentine routing patterns that allow the fabric to stretch and bend with the body without mechanical or electrical failure.

[0164] Insulating and Structural Layer 3070: A thin elastomeric or polyurethane barrier electrically isolates the conductive traces 3110 from the environment while providing mechanical reinforcement. Breathable microperforations preserve ventilation and comfort during extended wear.

[0165] Electronics Pod Connection 3080: The cross-section in Figure 23C shows electronic pod connections 3080 to connect to an electronic pod 3130 (like that illustrated in Figure 23) positioned at a non-load-bearing region such as the waistband, shoulder, or upper arm. The pod 3130 connects to the garment via magnetic snap connectors or low-profile conductive terminals, enabling easy removal for washing or maintenance. Inside the pod 3130 are an artificial intelligence processor (or processing unit 700 and memory 710) to run the ML279430.357_NP component 300, battery, wireless communication interface, and safety circuitry, among other necessary components. The processor 700 executes machine learning algorithms that fuse multimodal sensor data, perform artifact rejection, predict neuromuscular states (such as fatigue or atrophy risk), and dynamically adjust stimulation amplitude, frequency, and duty cycle.

[0166] Outer Fabric Shell 3100: The external layer 3100 of the garment is a soft, durable textile that protects the internal electronics while maintaining flexibility. It may include waterproof or abrasion-resistant coatings for long-term wearability and integrated ventilation zones for temperature regulation.

[0167] In some embodiments of a wearable device 3000 having the various aboveidentified elements (and as illustrated in Figures 22A through 23, comprises an actuation subsystem 100, sensing subsystem 200, and machine learning (ML) subsystem 300 that are operatively connected to form a continuous, closed-loop network that enables real-time data exchange and adaptive control. The manner of interconnection between subsystems may be wired, wireless, or a hybrid configuration, depending on the application, form factor, and regulatory constraints of the wearable platform.

[0168] In some embodiments, the actuation subsystem 100 is connected to the processor 700 (or ML subsystem 300) through shielded conductive traces, flexible printed circuit interconnects, or low-profile wiring harnesses integrated into or laminated within the textile structure (see Figure 22C.) These conductive pathways transmit both stimulation control signals (such as pulse width, current amplitude, or waveform shape) and feedback telemetry (such as electrode voltage, current, and impedance measurements) from the actuation driver circuitry to the processor 700. In a modular embodiment, detachable magnetic or snaptype connectors allow the actuation elements (e.g., electrode arrays or ultrasound transducers) to be removed or repositioned while maintaining reliable electrical contact with the main279430.357_NP electronics pod. The configurations of actuation drivers, their circuitry, and connections are known to one in the field and will vary depending on which of the numerous off-the-shelf actuation subsystems 100 are employed in individual embodiments of the present invention; however, Figures 22A through 23 illustrate some configurations of actuation subsystems 100 according to the present invention.

[0169] The processor 700 and ML subsystem 300 are typically implemented on a central electronics module 300 that houses the embedded microcontroller or system-on-chip (“SoC”), signal conditioning circuits, and communication interface. This module 300 is connected to the sensing subsystem 200 by either wired analog front-end connections or distributed sensor bus networks, such as FC, SPI, UART, or CAN protocols. These buses allow simultaneous acquisition of multiple sensor streams — including EMG, EIM, IMU, temperature, and pressure signals — while maintaining synchronization with stimulation events. In some embodiments, as is known to those with skill in the art, analog multiplexers or differential signal lines are employed to minimize wiring bulk while preserving signal fidelity across multiple channels.

[0170] In alternative embodiments, and as known to those with skill in the art, one or more subsystems (actuation 100, sensing 200, or machine learning 300) communicate wirelessly, using short-range data links such as Bluetooth Low Energy (BLE), Wi-Fi, or proprietary low- latency radio protocols. For example, and as known to those with skill in the art, a wireless module embedded within the electronics pod may transmit sensed physiological data to an external processing unit (e.g., smartphone, tablet, or cloud-based ML platform) for high-level analysis and visualization, while still performing safety-critical control locally. In such embodiments, , and as known to those with skill in the art, encryption and data integrity protocols (such as AES-128 or SSL / TLS) ensure secure transmission of patient information.279430.357_NP

[0171] The sensing subsystem 200 is also electrically and mechanically coupled to the actuation subsystem 100 to ensure temporal alignment of stimulation and recording. In certain embodiments, shared grounding or reference electrodes are employed to reduce motion artifacts and electrical interference. A synchronization clock or trigger line may run between the actuation driver and sensing amplifiers, allowing precise timing of EMG acquisition or impedance measurement immediately before, during, or after stimulation pulses. This synchronization enables accurate mapping of dose-response relationships and real-time ML model updates.

[0172] Power distribution among the subsystems (actuation 100, sensing 200, or machine learning 300) is achieved through a common regulated power bus routed via conductive textiles, flexible copper traces, or embedded energy modules. The processor 700 controls power sequencing to the sensing and actuation subsystems to optimize energy efficiency and safety. In some embodiments, energy harvesting elements (such as triboelectric or thermoelectric generators) or rechargeable flexible batteries are integrated to supply low- voltage power across the garment via distributed bus lines.

[0173] In cloud- or network-connected configurations, the ML subsystem 300 can extend beyond the local processor 700 to include edge-cloud hybrid architectures. In these cases, data acquired by the sensing subsystem 300 is preprocessed locally and transmitted wirelessly to a remote server for higher-order model training or cross-patient analysis. The remote ML engine returns updated parameters or control policies to the local processor, closing the feedback loop. The connection between local and cloud layers may be established through cellular (LTE / 5G), Wi-Fi, or secure loT network protocols (see Figure 24 for a non-limiting embodiment in which storage and memory can be configured as cloud- or network-connected configurations.)279430.357_NP

[0174] As mentioned previously, the cuff 3000 illustrated in Figure 23 attaches to the user’s body via a belt 3160 and secures around the user’s leg via Velcro® or any securing mechanism 3170 appropriate for the material from which the cuff is made. The cuff 3000 comprises optionally interchangeable gel electrodes 500 and a fabric-embedded connector port 3115 for the sensing component to snap off or onto the cuff. An electronic pod placement 3130 can be configured to attach to the belt 3160.

[0175] Alternative embodiments of wearable devices 3000 include a smart therapeutic garment 3000 comprising a sleeve, legging, vest, or wrap, that integrates sensing 200, stimulation 100, and machine learning subsystems 300 directly into flexible, wearable materials. These embodiments transform the platform from a laboratory or clinic-bound system into a comfortable, portable, and discreet device for everyday use in rehabilitation, athletic recovery, or chronic neuromuscular therapy. The garment 3000 can be constructed from a multilayer textile composite consisting of a base compression fabric, conductive pathways, embedded electrodes, and integrated sensors. In one embodiment, the base layer 3040 is composed of a breathable, stretchable textile such as a nylon-spandex or polyester-elastane blend, providing gentle compression (typically 15-25 mmHg) to maintain stable electrode 500 contact and minimize motion artifacts. Interwoven within the textile are conductive fibers or printed traces made from materials such as silver-plated nylon, graphene-coated polyester, or PEDOT:PSS-infused polymers. These pathways serve as the wiring system that distributes electrical and sensor signals across the garment while remaining soft, flexible, and resilient to bending.

[0176] Embedded at specific anatomical locations are electrode and sensor zones designed to align with key muscle groups and nerve pathways. Each zone includes electrodes 500 for neuromuscular electrical stimulation 100 and electromyography sensing 200, enabling the same region to both deliver and monitor muscle activation. Depending on the application,279430.357_NP electrodes 500 may be composed of soft hydrogel pads, dry conductive textiles, or flexible polymer composites that maintain low impedance and skin compatibility. Additional sensors 500, such as MEMS accelerometers for mechanomyography, piezoresistive or capacitive pressure sensors, and micro-thermistors for temperature monitoring, are laminated into the fabric to capture mechanical and physiological signals. For advanced configurations, ultrasound-based patches fabricated from thin piezoelectric polymers such as PVDF or PZT composites can be incorporated for imaging, stiffness sensing, or acoustic stimulation. These sensing elements are connected to a detachable electronic control module, a lightweight pod that houses the processor, power source, and wireless communication components. The module, typically located at a non-load-bearing region such as the upper arm or waistband, can include an edge Al processor (for example NVIDIA Jetson®, STM32, or Google® Coral TPU) to execute onboard ML algorithms for real-time analysis and control.

[0177] The wearable device 3000 is designed to fit snugly yet comfortably, ensuring reliable coupling between the electrodes 500 and skin 1 without requiring gels or adhesives. Proper fit and electrode placement are verified electronically by measuring baseline electrodeskin impedance before each session. If impedance values exceed preset thresholds, the system 1000 alerts the user to adjust the garment 3000 for improved contact. This feedback ensures consistent electrode 500 positioning and repeatable data acquisition across therapy sessions. Once worn, the garment 3000 automatically establishes communication between the sensing and actuation layers, continuously collecting multimodal physiological data including EMG, EIM, MMG, temperature, and pressure while simultaneously delivering controlled electrical or acoustic stimulation through the embedded actuators. These data are processed by the machine learning subsystem(s) 300 to adapt stimulation parameters 120 in real time, optimizing contraction efficiency and safety while preventing fatigue or overstimulation.279430.357_NP

[0178] Integrating the system 1000 and method 2000 into clothing 3000 provides significant advantages for both patients and clinicians. Figure 25 shows one non-limiting embodiment of a shirt and of pants (each a wearable device 3000) with an embedded array 510 according to the various configurations of the present invention detailed herein. For patients, the design of the wearable device 3000 (especially as a cuff or garment) eliminates the complexity of electrode 500 placement and setup, enabling hands-free, autonomous therapy that can be conducted at home or during daily activity. The comfort and simplicity of the design encourage consistent use, increasing total therapeutic exposure and accelerating recovery. For clinicians, the garment 3000 ensures consistent electrode 500 placement, standardized data collection, and remote monitoring capability through a secure mobile or cloud interface. Physiological data captured during everyday wear can reveal long-term trends in muscle strength, atrophy, and recovery, allowing clinicians to make informed treatment adjustments without requiring in-person visits. Athletes and physical therapy patients benefit from continuous muscle performance monitoring, real-time feedback on fatigue and activation, and adaptive recovery stimulation based on their individual physiology.

[0179] Embedding the system 1000 into clothing 3000 also improves data fidelity and spatial precision, as fixed electrode geometry ensures consistent targeting of the same muscle groups between sessions. The continuous compression fit enhances electrode coupling, which minimizes impedance drift and improves signal-to-noise ratios for both sensing and stimulation. The combination of comfort, precision, and automation makes the system 1000 far more accessible than traditional NMES devices, which rely on manual setup and supervision. Moreover, the garment’s modular design allows for region-specific variants such as a leg sleeve for lower-limb rehabilitation, an arm sleeve for stroke recovery, or a vest for core muscle therapy, making the technology adaptable to diverse medical and athletic use cases.279430.357_NP

[0180] Ultimately, embodying the invention as a wearable garment 3000 merges bioelectronic engineering with textile design to create a clinically effective, user-friendly, and scalable rehabilitation platform. It enables continuous, closed-loop therapy that adjusts to the user’s physiological state in real time while seamlessly fitting into daily life. By combining flexible materials, integrated sensors, distributed actuators, and embedded intelligence, this embodiment transforms the system into a form factor that patients can wear comfortably and use independently, bridging the gap between clinical-grade precision and everyday usability.

[0181] Exemplary Research Establishing Effectiveness of the Systems, Methods, and Devices.

[0182] Figures 5A and 5B depict one novel in vivo model of the present invention that was used to validate the technology and protocol (VML + E-stim I / O bioelectronic platform). Because rodents hav a tendency to chew at wearable devices, some of the research for the present invention entailed an implanted electronic stimulation (e-stim) platform, which perform similarly to the platforms that sit on the surface of the skin and validated the invention’s safe charge injection. The close-up in Figure 5B shows where the electrodes were connected to the rat’s tibialis anterior.

[0183] Figures 6A through 6F depict the ability of the systems, methods, and devices of the present invention to monitor safe charge injection (in-vivo e-stim characterization). As shown in Figures 6A through 6F, the ability of the in vivo platform of the present invention to monitor safe charge injection in real-time was validated by using a three-electrode in vivo array. In particular, by analyzing the evoked voltage transient, the injected non-ohmic potential can be quantified. Comparing the injected non-ohmic potential to the watersplitting threshold, which is conventionally used to ensure safe charge injection, one can detect how much more charge one can safely deliver to the animal / patient. Figures 6A through 6F show that embodiments of the present invention have the ability to monitor279430.357_NP charge injection on the body ensuring the stimulation will not operate if (a) the electrodes are not applied correctly and (b) the stimulation pulse is too high (the safety checks.) This ensures that in electrical stimulation mode the platform will not damage or irritate the user. In Figure 6A, the current was injected for therapy. Figure 6B shows the electrode placement in the TA. Figure 6C shows the recorded volage transient needed to calculate the faradaic polarization during electrical stimulation, plotted in Figure 6F. Knowing the faradic polarization of the electrodes ensures that in electrical stimulation mode the platform will not damage or irritate the user.

[0184] Figure 7 illustrates the use of EIM on healthy muscle and atrophied muscle.As previously explained, EIM is a relatively painless, non-invasive technique for the assessment of muscle conditions. One of the key parameters of EIM, is phase angle / reactance (0. ) - 50kHz assesses lipid-containing myofiber membrane density and distribution. Figure 7 shows the underlying biophysiological differences between healthy and atrophied skeletal muscle, highlighting the distinct biomarker signatures observed in each condition. In short, healthy muscle exhibits strong capacitive coupling and well-organized myofiber structure, whereas atrophied tissue demonstrates reduced capacitance and disrupted fiber integrity. In atrophied muscle, EIM at 50KHz shows decrease phase angle and reactance.

[0185] Figures 8A through 8F illustrate bioelectronic diagnostics using EIM. Figures 8A through 8F depict the use of electrochemical impedance myography to quantify and resolve atrophy neuromuscular which inspired this as a key sensing modality of embodiments of the present invention. Figure 8F shows the configuration of external electrodes used for EIM sensing in one embodiment of the present invention. Figures 9A through 10D illustrate some of the raw data extracted from EIM scans of rat muscles according to the research underlying the present invention.279430.357_NP

[0186] Figures 11 A through 1 ID illustrate the phase angle correlation with myo fiber percentage and torque production following neuromuscular injury. Figure 11 A shows representative phase spectra derived from electrical impedance myography sensing. The 50 kHz frequency used for subsequent analysis is shown. Figure 1 IB demonstrates a reduced magnitude of phase angle shift in the electronic stimulation (“E-stim”) cohort compared to the no-stimulation cohort at day 3 and week 1 post-muscular degeneration injury, relative to pristine (pre-defect) muscle. Retention of the 50 kHz phase angle magnitude closer to that of pristine muscle suggests that the stimulation protocol mitigates muscle atrophy in treated subjects. Figures 11C and 1 ID illustrate the correlation between phase angle magnitude and both myofiber percentage and torque output, respectively. Overall, these figures demonstrates the present invention’s ability to quantify EIM-derived phase angle and shows that this biomarker directly corresponds to skeletal muscle condition. This same correlative methodology can be incorporated into machine learning models to improve diagnostic accuracy and enable personalized optimization of stimulation parameters.

[0187] Figures 11 A through 1 ID show that the most pronounced drop in phase angle occurred immediately after defect creation and persisted through week 1. This period corresponds to the phase of active neuromuscular remodeling, highlighting a critical therapeutic window in which electrical stimulation has the greatest potential to influence muscle recovery and regeneration. This is an important difference between the use of the present invention to inform muscle stimulation treatment and traditional methods.Traditional methods begin countering muscular degeneration days or weeks after the initial injury or illness. Whereas, the present invention can be used to gather information about the muscle cells almost immediately after illness or injury and can be used to determine precisely when atrophy or muscular degeneration starts, moving up personalized treatment to the point in time where the cells begin showing degradation.279430.357_NP

[0188] Figures 12A and 12B illustrate the phases of neuromuscular regeneration following injury along with the various types of cells involved and scans of the resting muscle and the muscle cells various days out from the injury. These figures act as a reference or point of comparison for other figures shown in this application. Figures 12A and 12B illustrate the sequential phases that occur during neuromuscular regeneration, or conversely, during extended periods of disuse when muscle activity is absent. Injured or inactive neuromuscular tissue undergoes a series of biological processes beginning with an inflammatory phase, followed by granulation, tissue remodeling, and myofiber regeneration. During the early inflammatory stage, a reduction in myofiber density and tissue capacitance is typically observed. These changes are reflected in electrical impedance myography measurements, where shifts in phase angle and impedance magnitude correspond to the initiation, progression, and eventual resolution of inflammation and tissue repair.

[0189] Figures 13A and 13B show a representative cross-section of healthy rat tibialis anterior (“TA”) muscle, which is characterized by well-organized, oval-shaped myofibers surrounded by tightly packed connective tissue called sarcolemma. This distinct structural organization produces the characteristic electrical properties that can be detected and quantified using our electrochemical impedance myography sensing tools

[0190] Figures 14A through 14D show representative cross-section of inflamed mid atrophy rat tibialis anterior muscle. These images show tissue undergoing neuromuscular atrophy, characterized by reduced coupling between myo fibers and the loss of distinct white sarcolemma boundaries (physiological changes in the tissue.) These spaces are instead replaced by granulation and connective tissue. Such structural changes alter the muscle’s electrical properties and are detected by our electrochemical impedance myography sensing system. The systems, devices, and methods of the present invention utilize these detectable structural changes to identify muscle atrophy much earlier than currently detected, which279430.357_NP allows for earlier treatment, more personalized treatment, less muscular degeneration prior to beginning treatment, and better patient outcomes as a result.

[0191] Figures 15A through 15D show a correlation of sensing data following injury through week 1. Following the same injury model shown in Figures 15 A through 15D, measurements taken up to week 1 demonstrate a continued reduction consistent with the present invention’s sensing data, further supporting the correlation between electrical impedance signals and tissue remodeling during early recovery, which is not being done with the current medical technologies and protocols.

[0192] Figures 16A through 16E illustrates how electromyography was used to inform stimulation conditions. Raw spontaneous surface EMG signals were recorded from sites above and below the injury in the rat tibialis anterior, corresponding to regions proximal and distal to the muscular degeneration defect. EMG burst analysis showed that under the non-stimulated (“NS”) condition, there was a marked decrease in the number of burst events successfully propagating from the proximal to distal side of the injured tissue by Day 4 post-injury.

[0193] Using this information as a metric to guide stimulation timing, electrical stimulation was administered daily through Day 4. This therapeutic regime effectively eliminated the loss of EMG burst propagation, indicating that our stimulation protocol helps preserve healthy neuromuscular signal transmission across and surrounding the injury site.

[0194] Figures 17A through 17F illustrate the capability of the systems, methods, and devices of the present invention to enhance functional neuromuscular regeneration through electrical stimulation by Force Gauge Analysis and Rat Force Gauge Metric. Figures 17A through 17F illustrate an evalution of the torque of rats after injury with and without therapeutic stimulation according to the systems, methods, and devices of the present invention. The rat’s foot is attached to a torque gauge from the max torque produced by the279430.357_NP anterior comparment during dorsiflexile acuation is measured. Six max torque peaks were averaged to compare representative max torque metrics for each animal / timepoint (see Figures 17A through 17F.) Later figures build on these metrics, showing direct correlation between gained strength and sensed physiological changes (EIM and EMG). Figures 18A through 18C show that electrical stimulation increases functional torque week 2 and beyond with use of the various methods, systems, and devices of the present invention. Functional regeneration was quantified using Force Gauge Analysis on the defect muscle, and the data show a significant increase in functional output starting from week 2 and continuing through week 8, with n=10 replicates in both the electrical stimulation and the control (defect-only) condition.

[0195] Figures 19A through 19E show that stimulation using the systems, methods, and devices of the present invention enhance the muscle (tibialis anterior) that was damaged, which was the target muscle. Importantly, the Extensor Digitorum Longus (“EDL”), which is a neighboring muscle that was not targeted, did not show hypertrophy (increased myo fiber hypertrophy) in the stimulation cohort while the TA did. Figures 19A through 19E establish that the systems, methods, and devices of the present invention can be configured to specifically target the injured muscle, to avoid or reverse atrophy, instead of simply improving the strength of neighboring muscles, which could compensate for the injured muscle. The systems, methods, and devices of the present invention are able to target specific muscles of interest.

[0196] Figures 20 A through 20F shows that the various embodiments of the present invention have the capability of ensuring safe charge injection, and preventing injury to user, in an in vivo platform

[0197] In addition to monitoring safe charge injection, the system and method of the present invention also offers electrochemical impedance myography, a non-invasive, painless279430.357_NP technique to assess muscle condition. This four-electrode technique uses low-amplitude, multi-frequency current to evaluate muscle physiology. At 50 kHz, studies have shown that healthy muscle exhibits higher phase angle and reactance compared to atrophied muscle, as seen in research comparing young and elderly patients and in studies of skeletal muscle following neuromuscular injury. One can use this same four-electrode technique to assess patients' skeletal muscle characteristics.

[0198] The platform’s 50 kHz in vivo data reveal a significant decrease in reactance by week 1 in the control group (without electrical stimulation) compared to the therapeutic electrical stimulation cohorts. Preliminary results at week 1 suggest that exogenous electrical stimulation either accelerates the myogenic regenerative process or mitigates myofiber necrosis. Further histological studies and assays will clarify the underlying mechanisms of these changes.

[0199] Importantly, these differences were detected a full week before functional divergences were observed, indicating that the platform's electrochemical impedance myography can predict regenerative states before functional metrics can be measured. This predictive capability offers muscle-wasting clinicians an unprecedented view of their patients' physiological status.

[0200] Figures 21 A through 2 IK illustrate histological analyses across multiple time points before and after volumetric muscle loss injury. Figure 21 shows representative Masson’s Trichrome-stained sections of the injury defect region for four conditions: No-Stim Day 3 (Figure 21A), E-Stim Day 3 (Figure 21B), No-Stim Week 8 (Figure 21C), and E-Stim Week 8 (Figure 21D.) The boxes highlight the injured tibialis anterior cortex from which subsequent quantitative analyses were performed. Electrical stimulation resulted in a significant increase in myofiber percentage from Day 3 through Week 2 compared to the No- Stim cohort (see Figures 21E through 21G.) Immunohistochemical (IHC) staining for279430.357_NPCD 163, a marker of anti-inflammatory M2 macrophages, revealed that electrical stimulation significantly increased M2 macrophage populations at both Day 3 and Week 2 post-injury (see Figures 21H and 211.) This finding is notable, as M2 macrophages are closely associated with pro-regenerative tissue remodeling and neuromuscular repair.

[0201] Similarly, IHC staining for Pax7, a satellite cell marker, showed a significant increase at Week 1 in the E-Stim condition, indicating enhanced stem cell activation and repopulation within the injured tissue (see Figures 21 J and 2 IK.) Collectively, these data demonstrate that properly timed electrical stimulation promotes increased myo fiber regeneration, satellite cell activation, and anti-inflammatory macrophage recruitment, thereby accelerating functional neuromuscular regeneration following muscular degeneration injury.

[0202] While the disclosure has been described in detail and with reference to specific embodiments thereof, it will be apparent to one skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the embodiments. Thus, it is intended that the present disclosure covers the modifications and variations of this disclosure, as well as other applications of the invention, provided they come within the scope of the appended claims and their equivalents.

Claims

279430.357_NPCLAIMSWhat is claimed is:

1. A system for treating muscular degeneration, comprising: an actuation subsystem configured to stimulate at least one muscle during a treatment and to generate actuation output data; a sensing subsystem configured to communicate with the actuation subsystem and to evaluate the actuation output data to identify at least one physiological response by the stimulated muscle and to generate sensing output data based upon the evaluation; and at least one machine learning subsystem in communication with the actuation subsystem and the sensing subsystem, the at least one machine learning subsystem configured to process the sensing output data to generate at least one treatment parameter.

2. The system of Claim 1, wherein the machine learning subsystem archives the sensing output data from a first treatment for comparison to a sensing output data from a subsequent treatment.

3. The system of Claim 1, wherein the at least one treatment parameter is communicated to the actuation subsystem for use during a subsequent treatment.

4. The system of Claim 1, wherein the actuation subsystem comprises at least one stimulation method selected from the group consisting of electrical, mechanical, bioelectrical, sound, and heat.

5. The system of Claim 1, wherein the sensing subsystem is configured to incorporate at least one sensing modality selected from the group consisting of electrophysiological, mechanical, optical, thermal, and pressure and wherein the at least one modality can quantify a neuromuscular state or a therapeutic response.

6. The system of Claim 1 , wherein the actuation subsystem and the sensing subsystem are configured to fit on a wearable device.279430.357_NP7. The system of Claim 1, also comprising an array of sensors configured to function as part of the actuation subsystem and the sensing subsystem.

8. A wearable device for treating muscular degeneration, comprising: an actuation subsystem configured to actuate at least one muscle and configured to generate actuation output data; a sensing subsystem configured to receive the actuation output data and configured to identify at least one physiological response by the actuated muscle based upon the actuation output data and to generate sensing output data; and at least one machine learning subsystem in communication with the actuation subsystem and the sensing subsystem, the at least one machine learning subsystem configured to process the sensing output data to generate at least one treatment parameter.

9. The device of Claim 8, wherein the actuation subsystem, the sensing subsystem, and the at least one machine learning subsystem are incorporated into a wearable cuff configured to wrap around a body part.

10. The device of Claim 8, wherein the cuff comprises: a skin interface layer that contacts the body part; a sensing and electrode layer configured with at least one electrode that is integrated into a flexible substrate; an electronics pod configured to run the machine learning subsystem and to provide communication and energy to the actuation subsystem, the sensing subsystem, and the machine learning subsystem; a conductive pathway layer comprising conductive fibers or printed metallic traces forming flexible circuits connecting the electrodes to the electronics pod; an insulating and structural layer that isolates the conductive fibers or metallic traces from an environment around the cuff; and279430.357_NP an outer shell.

11. The device of Claim 9 also comprising an array of electrodes configured to contact the body part.

12. The device of Claim 11 , wherein the electrodes are configured for actuation and sensing.

13. The device of Claim 10, further comprising a display on the outer shell configured to display instructions to a user.

14. The device of Claim 8, wherein the machine learning subsystem is configured as a continuous, closed loop to process real-time data from the sensing subsystem to provide the at least one treatment parameter.

15. The device of Claim 8, wherein sensing subsystem is electrically and mechanically coupled to the actuation subsystem to ensure temporal alignment of stimulation and sensing.

16. A method of treating muscular degeneration, comprising: evaluating a muscle’s physiology with a sensing subsystem to produce sensing output data; processing sensing output data by a machine learning subsystem to produce at least one treatment parameter; actuating the muscle with an actuation subsystem according to the at least one treatment parameter; and repeating the evaluating and processing steps to produce at least one updated treatment parameter.

17. The method of Claim 16, also comprising verifying a proper placement of at least one electrode on a body part prior to evaluating a muscle’s physiology.

18. The method of Claim 16, wherein the machine learning subsystem comprises: extracting features from and preprocessing the sensing output data to filter out noise and artifacts from the sensing output data to generate filtered output data; and279430.357_NP processing the filtered output data for model prediction and adaptive control to produce the at least on treatment parameter for use by the actuation subsystem.

19. The method of Claim 16, wherein the steps are repeated until the sensing subsystem detects no continued muscular degeneration from the muscle’s physiology.

20. A device for treating muscular degeneration comprising: an actuation subsystem configured to stimulate at least one muscle during a treatment and to generate actuation output data; and a sensing subsystem configured to communicate with the actuation subsystem and configured to receive the actuation output data, and evaluate the actuation output data to determine at least one physiological attribute of the muscle, wherein the actuation subsystem and the sensing subsystem are configured as a portable system.

Images