Neuromuscular interfaces to engineer the phenotype and recruitment of skeletal muscles

Abstract

Systems and methods for transforming a phenotype of a muscle of a human or animal and for biohybrid organ systems are described herein. An example embodiment is directed toward system for transforming a phenotype of a muscle. The system comprises a sensor configured to detect signals associated with a muscle and a signal generator configured to stimulate the muscle. The system further comprises a processor configured to determine a stimulus pattern based on neuromuscular dynamics computed from the signals detected. The neuromuscular dynamics computed are indicative of a current phenotype of the muscle. The processor is further configured to deliver the stimulus pattern determined to the muscle via the signal generator. The stimulus pattern determined and delivered causes a transformation of the muscle toward a target phenotype. Embodiments described herein may be useful for engineering robust, fatigue-resistant, and computer-controllable tissue muscle constructs that can be used for restoring neuromuscular function.

Description

Docket No. 4889-1034001NEUROMUSCULAR INTERFACES TO ENGINEER THE PHENOTYPE AND RECRUITMENT OF SKELETAL MUSCLESRELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Application No.63 / 765,233, filed on February 28, 2025. The entire teachings of the above application are incorporated herein by reference.BACKGROUND

[0002] Skeletal muscles are biological actuators that drive almost all actuation and movement in mammals, including humans. Due to the nanometer-scale molecular machinery regulated by ion control for contraction of muscle cells, muscles can be remarkably efficient and scalable actuators [1], Moreover, muscle can possess inherent biocompatibility and selfsustainability, have unmatched energetics due to metabolic efficiency, scale from micrometric to hundreds of kilograms scale, and enable, through neural innervation, high-precision control of muscle fibers. Muscle can exhibit functional adaptation to environmental demands through hypertrophic and hyperplastic growth and can be capable of self-repair in case of damage. On top of that, muscle can provide an adaptive compliant media with a form factor that can be tensioned, wrapped, or rolled to interface with various organs inside the body for design of biohybrid organ systems.

[0003] In various neurological conditions, such as stroke, spinal cord injury, or amyotrophic lateral sclerosis, a communication pathway between the central nervous system (CNS) and neuromuscular components can be severed, which may result in deficits such as paralysis and organ failure, for non-limiting examples. Additionally, individuals may experience decreased functionality in muscles due to aging, experience difficulty moving around, and have a high probability of falling and injury, which can ultimately limit quality of life.SUMMARY

[0004] Restoring muscle function can be challenging due to the organization of the neuromuscular system. Motor units can be fundamental units in neuromuscular control, they are composed of a motor neuron with its axon, and the group of muscle fibers that a single motor neuron innervates. Motor units can come in different configurations based on the- 1 - 5040745. vlDocketNo. 4889-1034001functional requirements of muscles. Small motor units can have small diameter axons, transmit action potentials slower, and innervate slow-twitch muscle fibers. Fast motor units can have large diameter axons, transmit action potentials at higher speeds, and innervate fasttwitch muscle fibers. Typically, peripheral nerves such as motor nerves can have a combination of slow and fast motor units. Recruitment of motor units can follow the size principle, which states that small units get activated first, and with increasing motor activation, larger units get activated [2], Neuroprostheses and bioelectronic therapies may aim to replace the missing neuronal input by delivering commands using artificial stimulation to restore or aid in muscle function. Functional electrical stimulation (FES) has been the dominant stimulation modality. FES can enable artificial control of muscle actuation and can be performed by incorporating a synthetic component, such as a muscle or nerve electrode. However, due to the distinct biophysics of motor nerves that arise from their morphology, large motor units can get preferentially recruited, which may result in poor controllability and rapid fatigue of muscles [2], These issues may limit the chronic use in daily life of neuroprostheses and bioelectronic therapies.

[0005] A design space of a neuromuscular interface that aims to solve the muscle control problem could seek to engineer the way muscle fibers are recruited (stimulation interface) or to modify the phenotype of the muscle fibers.

[0006] An example embodiment is directed to a system for transforming a phenotype of a muscle of a human or animal. The system comprises a sensor configured to detect signals associated with a muscle of a human or animal and a signal generator configured to stimulate the muscle. The system further comprises a processor configured to determine a stimulus pattern based on neuromuscular dynamics computed from the signals detected. The neuromuscular dynamics computed are indicative of a current phenotype of the muscle. The processor is further configured to deliver the stimulus pattern determined to the muscle via the signal generator. The stimulus pattern determined and delivered causes a transformation of the muscle toward a target phenotype.

[0007] The sensor can be configured to detect the signals responsive to a stimulus delivered to the muscle. The stimulus can be delivered to the muscle by one or more of a nervous system of the human or animal or by the signal generator. The processor can be further configured to cause the signal generator to deliver the stimulus. The stimulus delivered can comprise a waveform having one or more of (i) a low frequency short duration stimulus, (ii) a high frequency long duration stimulus, or (iii) a stimulus having one or more- 2 - 5040745. vlDocket No. 4889-1034001frequencies. In an iterative manner, the sensor can be configured to detect the signals associated with the muscle and the processor can be configured to (i) determine the stimulus pattern based on the neuromuscular dynamics, the neuromuscular dynamics being computed from signals detected responsive to the stimulus to the muscle, and (ii) deliver the stimulus pattern generated to the muscle via the signal generator until the target phenotype of the muscle is achieved.

[0008] The processor can be further configured to compute the neuromuscular dynamics from the signals detected. The neuromuscular dynamics computed can include one or more of: muscle activation, muscle length, muscle velocity, muscle force, tendon length, tendon force, muscle fatigue, muscle fatigue rate, and peak muscle state over a dynamic frequency range. The processor can be further configured to perform one or more of (i) computing the muscle force by forming a muscle-tendon model based on muscle activation, muscle length, and muscle velocity and (ii) computing the tendon force through a tendon dynamic model based on the tendon length.

[0009] The processor can be further configured to infer the current phenotype of the muscle based on the neuromuscular dynamics computed. The processor can be configured to generate the stimulus pattern based on the current phenotype inferred. A phenotype of the muscle can include a fast twitch phenotype or a slow twitch phenotype. The system can further comprise an auxiliary sensing system. The processor can be configured to infer the current phenotype based on the neuromuscular dynamics computed and data acquired from the auxiliary sensing system. The auxiliary sensing system can comprise one or more of magnetic resonance imaging, computed tomography, metabolic profiling, genotyping, and myosin heavy chain isoform profiling.

[0010] The sensor can comprise one or more of: an implantable electrode, a surface electrode, an implantable magnetic sensor, an ultrasound sensor, or an optical sensor.

[0011] The sensor can be configured to detect the signals associated with the muscle using one or more of electromyography, magnetomicrometry, ultrasound imaging, nearinfrared imaging, laser doppler vibrometry, or optical coherence tomography.

[0012] The signal generator can be configured to stimulate the muscle using one or more of: a transcutaneous electrical stimulation, a fine-wire stimulation, a nerve cuff stimulator, an ultrasound stimulation, or a magnetic stimulation.

[0013] The processor can be configured to determine the stimulus pattern based on the neuromuscular dynamics computed and one or more of (i) a model estimating the current- 3 - 5040745. vlDocketNo. 4889-1034001phenotype of the muscle based on a repository of muscle stimulus data or (ii) a reference library comprising stimulation patterns versus muscle phenotypes.

[0014] The signal generator can be coupled electrically to a nerve innervating the muscle. The signal generator can deliver the stimulus pattern to the muscle through the nerve. The signal generator can be coupled electrically to a nerve at a first site and a second site, the first site being proximal to the brain with respect to the second site. The signal generator can be configured to apply a nerve block at the first site. The nerve block applied by the signal generator can comprise a kilohertz-frequency alternating current and the nerve block applied can block propagation, in a direction of the first site, of a stimulus delivered to the second site by the signal generator.

[0015] The processor can be further configured to actuate the muscle by delivering a stimulus to the muscle via the signal generator, the muscle performing a native function of the muscle responsive to the actuating.

[0016] Another example embodiment is directed to a method of transforming a phenotype of a muscle of a human or animal. The method comprises detecting signals associated with a muscle of a human or animal and computing neuromuscular dynamics of the muscle based on the signals detected. The neuromuscular dynamics are indicative of a current phenotype of the muscle. The method further comprises determining a stimulus pattern based on the neuromuscular dynamics computed and delivering the stimulus pattern determined to the muscle. The stimulus pattern determined and delivered to the muscle causes a transformation of the muscle toward a target phenotype.

[0017] Detecting the signals can be responsive to a stimulus delivered to the muscle by a nervous system of the human or animal or by a signal generator configured to stimulate the muscle. The method can further comprise causing delivery of the stimulus to the nerve. The method can further comprise encoding the stimulus delivered with a temporally varying pattern. The detecting the signals associated with the muscle, the computing the neuromuscular dynamics, the determining the stimulus pattern, and the delivering the stimulus pattern can be repeated in an iterative manner until the target phenotype of the muscle is achieved.

[0018] Detecting the signal can comprise detecting one or more of an electrical signal, a magnetic signal, an ultrasound signal, or an ultrasound signal.

[0019] Computing the neuromuscular dynamics can comprise computing, based on the signal detected, one or more of muscle activation, muscle length, muscle velocity, muscle- 4 - 5040745. vlDocket No. 4889-1034001force, tendon length, tendon force, muscle fatigue, muscle fatigue rate, and peak muscle state over a dynamic frequency range. Computing the neuromuscular dynamics can further comprise one or more of: (i) computing the muscle force by forming a muscle-tendon model based on muscle activation, muscle length, and muscle velocity and (ii) computing the tendon force through a tendon dynamic model based on the tendon length.

[0020] The method can further comprise inferring the current phenotype of the muscle based on the neuromuscular dynamics computed. Determining the stimulus pattern can be based on the current phenotype inferred. Inferring a phenotype of the muscle can include a fast twitch phenotype and a slow twitch phenotype. The method can further comprise augmenting the neuromuscular dynamics using data collected from an auxiliary sensing system. The auxiliary sensing system can comprise one or more of magnetic resonance imaging, computed tomography, metabolic profiling, genotyping, and myosin heavy chain isoform profiling. Inferring the current phenotype can be based on the neuromuscular dynamics computed and augmented.

[0021] Determining the stimulus pattern can be based on the neuromuscular dynamics computed and one or more of (i) a model estimating the current phenotype of the muscle based on a repository of muscle stimulus data the stimulus pattern delivered or (ii) a reference library comprising stimulation patterns versus muscle phenotypes.

[0022] Delivering the stimulus pattern can include delivering a transcutaneous electrical stimulation, fine-wire stimulation, an electrical nerve cuff, ultrasound stimulation, or magnetic stimulation.

[0023] Delivering the stimulus pattern to the muscle can comprise delivering the stimulus pattern to a nerve innervating the muscle. Delivering the stimulus pattern to the nerve can further comprise delivering a nerve block signal to a first site along a nerve and delivering the stimulus pattern at a second site along the nerve. The nerve block can prevent a transmission of the stimulus pattern delivered beyond the first site from the second site.

[0024] The method can further comprise, for the muscle of the target phenotype, actuating the muscle by delivering a stimulus to the muscle.

[0025] The method can further comprise, for the muscle of the target phenotype, monitoring the current phenotype of the muscle by detecting the signals associated with the muscle. The method can further comprise, responsive to a change of the muscle from the target phenotype, determining the stimulus pattern based on the neuromuscular dynamics computed from the signals detected and delivering the stimulus pattern determined.- 5 - 5040745. vlDocketNo. 4889-1034001

[0026] Another example embodiment is directed to a biohybrid organ system. The biohybrid organ system comprises an organ system of a human or animal and a myoneural actuator comprising a transplanted muscle of the human or animal. The transplanted muscle is in coupled arrangement with a portion of the organ system. A contraction of the transplanted muscle responsive to a stimulus mechanically actuates the portion of the organ system to perform a physiological function of the organ system.

[0027] The myoneural actuator can further comprise a signal generator functionally coupled to the transplanted muscle. The signal generator can be configured to deliver the stimulus to the muscle. The signal generator can be functionally coupled to the muscle via a nerve of the human or animal. The signal generator can be configured to cause the delivery of stimulus to the muscle via the nerve. The muscle can be decoupled from a native efferent nerve of the muscle, and the nerve can comprise a transected portion of a sensory nerve of the human or animal. The muscle can be reinnervated using the nerve and the muscle can be coupled to the nerve via a restored neuromuscular junction. The biohybrid organ system can further comprise a sensor configured to acquire to detect signals associated with the transplanted muscle a processor. The processor can be configured to compute neuromuscular dynamics of the transplanted muscles based on the signals detected and to transform a phenotype of the transplanted muscle based on the neuromuscular dynamics computed. The processor can be configured to cause the signal generator to deliver the stimulus of the muscle based on a function of the organ system.

[0028] The transplanted muscle of the myoneural actuator can be configured to be fatigue resistant based on properties of a nerve innervating the transplanted muscle or on a phenotype of the transplanted muscle.

[0029] The transplanted muscle can be configured to receive the stimulus from a nervous system of the human or animal.

[0030] The myoneural actuator can be a first myoneural actuator the biohybrid organ system can comprise at least one additional myoneural actuator in coupled arrangement with the organ system. The myoneural actuator and the at least one additional myoneural activator can be configured to actuate the organ system to perform a physiological function of the organ system.

[0031] The organ system can comprise one or more of: (i) a digestive system the myoneural actuator can encircle a portion of an intestine, (ii) a cardiac system and the myoneural actuator can wrap around a portion of a heart, (iii) a respiratory system and- 6 - 5040745. vlDocketNo. 4889-1034001wherein the myoneural actuator can be mechanically coupled to a diaphragm or an intercostal muscle, (iv) a urinary system and he myoneural actuator can encircle a portion of a bladder, (v) a sphincter system, (vi) a smooth muscle system, or (vii) skin and the myoneural actuator can be mechanically coupled to a skin graft.

[0032] Another example embodiment is directed to a method of actuating an organ system of a human or animal. The method comprises surgically coupling a muscle to an organ system of the human or animal to position the muscle in coupled arrangement with a portion of the organ system. The method further comprises causing the muscle surgically coupled to the organ system to contract by stimulating the muscle. The muscle contracting mechanically actuates the portion of organ system to perform a physiological function of the organ system.

[0033] The method can further comprise transecting a nerve of the human or animal and reinnervating the muscle using the nerve transected. The stimulus to the muscle can be deliverable via the nerve. The method can comprise stimulating the muscle via the nerve. The method can further comprise selecting the nerve based on one or more axonal biophysical properties of the nerve. The nerve can be a peripheral nerve or a sensory nerve. The method can further comprise delivering a nerve block to the nerve transected. The nerve block can be configured to prevent transmission of a stimulus of the nerve in an afferent direction.

[0034] The method can further comprise transecting a native efferent nerve of the muscle extracted and operatively coupled. The transecting the native efferent nerve can disconnect the muscle from an efferent signal from a nervous system of the human or animal to the muscle.

[0035] The method can further comprise delivering one or more stimulus patterns to the muscle. A stimulus pattern of the one or more stimulus patterns can cause a transformation of a phenotype of the muscle toward a target phenotype. The target phenotype can be associated with fatigue resistance.

[0036] Surgically coupling the muscle extracted to the organ system comprise, using the muscle: (i) encircling a portion of an intestine, (ii) wrapping a portion of a heart, (iii) mechanically coupling to a portion of a diaphragm or an intercostal muscle, (iv) encircling a portion of a bladder, (v) mechanically coupling to a sphincter, (vi) mechanically coupling to smooth muscle, or (vii) mechanically coupling to a skin graft.

[0037] The muscle can be a first muscle and the method can further comprise surgically coupling at least one additional muscle. The method can further comprise causing the muscle- 7 - 5040745. vlDocket No. 4889-1034001and the at least one additional muscle to contract in coordination to perform the physiological function of the organ.BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawings will be provided by the Office upon request and payment of the necessary fee.

[0039] The foregoing will be apparent from the following more particular description of example embodiments, as illustrated in the accompanying drawings in which like reference characters refer to the same parts throughout the different views. The drawings are not necessarily to scale, emphasis instead being placed upon illustrating embodiments.

[0040] FIG. 1 illustrates an example embodiment of a system for transforming a muscle of a human or animal.

[0041] FIG. 2A illustrates an example embodiment of an implantable system for transforming a muscle of a human or animal.

[0042] FIG. 2B illustrates an example embodiment of a minimally invasive system for transforming a muscle of a human or animal.

[0043] FIGS. 3A-3F are schematic representations of neuromuscular dynamics of a muscle, the neuromuscular dynamics being computed from signals detected from said muscle, according to example embodiments.

[0044] FIG. 4 is a block diagram of an example muscle model architecture that can be used by an example embodiment of a system for transforming a phenotype of a muscle.

[0045] FIGS. 5A-5D are block diagrams of example system identification procedures for estimating characteristics of a muscle that may be used by example embodiments of a system for transforming a phenotype of a muscle.

[0046] FIG. 6 illustrates schematically an example closed-loop stimulation architecture that can be used by an example embodiment of a system for transforming a phenotype of a muscle.

[0047] FIG. 7 illustrates schematically another example closed-loop stimulation architecture that can be used by an example embodiment of a system for transforming a phenotype of a muscle.- 8 - 5040745. vlDocketNo. 4889-1034001

[0048] FIG. 8 illustrates schematically another example closed-loop stimulation architecture comprising a nerve block that can be used by an example embodiment of a system for transforming a phenotype of a muscle.

[0049] FIG. 9 illustrates schematically another example closed-loop stimulation architecture usable during normal operation of the muscle by an example embodiment of a system for transforming a phenotype of a muscle.

[0050] FIG. 10 illustrates schematically an example closed-loop stimulation architecture that can be used by an example embodiment of a system for transforming a phenotype of a muscle of a myoneural actuator (MNA).

[0051] FIG. 11 illustrates schematically an example closed-loop muscle control architecture usable by an example embodiment of a system for transforming a phenotype of a muscle after phenotype transformation.

[0052] FIG. 12 illustrates schematically an example embodiment of a system for transforming a phenotype of a muscle of a human.

[0053] FIG. 13 illustrates schematically an example embodiment of a system for transforming a phenotype of a muscle of an animal, with insets illustrating embodiments of components of the system.

[0054] FIG. 14 is an x-ray radiograph illustrating the system described hereinabove with respect to FIG. 13, the system implanted in a large-animal model, e.g., a pig, at week 4 postimplantation.

[0055] FIG. 15 illustrates schematically an example embodiment of a system for transforming a phenotype of a muscle of an animal, wherein the system is configured for open-loop characterization and closed-loop control of the muscle.

[0056] FIG. 16A is a plot illustrating example muscle twitch dynamics over time, according to an example embodiment.

[0057] FIG. 16B is a plot illustrating example time-based muscle dynamics following an increasing amplitude sweep stimulation protocol, according to an example embodiment.

[0058] FIG. 16C is a plot illustrating an example neuromechanical recruitment curve following the amplitude sweep stimulation protocol of FIG. 16B.

[0059] FIG. 16D is a plot illustrating example time-based muscle dynamics following an increasing frequency sweep simulation protocol, according to an example embodiment.

[0060] FIG. 16E is a plot illustrating an example neuromechanical firing rate curve following the frequency sweep stimulation protocol of FIG. 16D.- 9 - 5040745. vlDocket No. 4889-1034001

[0061] FIG. 17A illustrates an example closed-loop neuromechanical control architecture that may be used by an example embodiment of a system for transforming a phenotype of a muscle.

[0062] FIG. 17B is a block diagram of an example closed-loop control architecture that can be implemented by the controller of FIG. 17A

[0063] FIG. 17C illustrates example movement of an ankle joint caused using the system of FIG. 17 A.

[0064] FIGS. 18A-18C are plots plot illustrating closed-loop neuromechanical control of an ankle joint following a square reference trajectory of varying amplitudes, according to an example embodiment.

[0065] FIGS. 19A-19D illustrate neuromechanical control of a knee joint, according to example embodiments.

[0066] FIG. 20 illustrates schematically example embodiments of myoneural actuators and applications thereof.

[0067] FIG. 21 illustrates schematically an example embodiment of a myoneural actuator.

[0068] FIG. 22 illustrates schematically another example embodiment of a myoneural actuator implemented in a rodent model.

[0069] FIG. 23 illustrates histological cross-sections of a native motor nerve (left) and a pure sensory nerve (right), according to an example embodiment.

[0070] FIG. 24 is a plot illustrating cumulative probability densities of axon cross-section area for native motor nerves and pure sensory nerves, according to an example embodiment.

[0071] FIG. 25 is a plot illustrating mean axon cross-sectional area and axon crosssection area variability for native motor nerves and pure sensory nerves, according to an example embodiment.

[0072] FIG. 26 illustrates fluorescence imaging of an example embodiment of a myoneural actuator comprising a muscle reinnervated by a sensory nerve.

[0073] FIG. 27 illustrates histological images of muscle cross-sections before and after myoneural manipulation, according to an example embodiment.

[0074] FIG. 28A illustrates histological images of a fluorescent cross-sectional slide of a myoneural actuator, according to an example embodiment.

[0075] FIG. 28B is a plot illustrating percent composition of slow and fast fibers in native muscle and myoneural actuators based on histological images similar to those of FIG. 28 A.- 10 - 5040745. vlDocket No. 4889-1034001

[0076] FIG. 29 is a plot illustrating mass and maximum myoneural actuator force of a muscle of a myoneural actuator without stimulation after surgical intervention, according to an example embodiment.

[0077] FIGS. 30A and 3 OB are plots illustrating mass and maximum force, respectively, of native muscle and myoneural actuators, according to example embodiments.

[0078] FIG. 31 is a plot illustrating individual twitch force responses for a native muscle and a myoneural actuator, according to an example embodiment.

[0079] FIG. 32 is a plot illustrating sequential twitch force responses for native muscles and myoneural actuators, according to an example embodiment.

[0080] FIG. 33 is a plot illustrating twitch force loss per cycle based on the sequential twitch force responses of FIG. 32.

[0081] FIG. 34 is a plot illustrating cumulative probability densities of the twitch forces of native muscles and myoneural actuators of FIG. 32.

[0082] FIG. 35 is a plot illustrating twitch force variability of the twitch forces of native muscles and myoneural actuators of FIG. 32.

[0083] FIGS. 36A and 36B are plots illustrating force response of native muscles and myoneural actuators under continuous stimulation, according to an example embodiment.

[0084] FIG. 37 is a plot illustrating fatigue dynamics of native muscles and MNAs based on the force response under continuous stimulation of FIGS. 36A and 36B.

[0085] FIG. 38 illustrates an example embodiment of a system 3800 comprising a closed-loop force control architecture that can be used to actuate a muscle of a myoneural actuator.

[0086] FIG. 39 is a plot illustrating closed-loop control performance of native muscles (top) and myoneural actuators (bottom) across force target values, according to an example embodiment.

[0087] FIGS. 40 A and 40B are plots illustrating quantification of the closed-loop control performance of the native muscles and the myoneural actuators of FIG. 39 over stimulation cycles.

[0088] FIG. 41 illustrates schematically an example embodiment of a system comprising a myoneural actuator with a nerve block.

[0089] FIG. 42 is a plot illustrating normalized force applied by a myoneural actuator, similar to the MNA schematically illustrated in FIG. 41, with or without a nerve block.

[0090] FIG. 43 illustrates schematically an example embodiment of a myoneural actuator comprising a nerve block and a closed-loop control architecture.- 11 - 5040745. vlDocket No. 4889-1034001

[0091] FIGS. 44 A and 44B are plots illustrating force modulation of the myoneural actuator illustrated schematically in FIG. 43 under nerve block.

[0092] FIG. 45 illustrates schematically an example embodiment of a biohybrid organ system for the small intestine.

[0093] FIGS. 46A-46J illustrate an example embodiment of a method for implementing a myoneural actuator, which may be used as part of a biohybrid organ system.

[0094] FIGS. 47A-47H illustrate example embodiments of methods for characterization and control of a myoneural actuator.

[0095] FIGS. 48A-48L illustrate schematically example embodiments of a method of constructing and validating a biohybrid gastrointestinal organ system.

[0096] FIG. 49 illustrates example embodiments of biohybrid organ systems.

[0097] FIG. 50 illustrates an example embodiment of a methodology for design of a biohybrid organ systems using MNAs.

[0098] FIG. 51 illustrates schematically example embodiments of myoneural actuators during stages of construction, refinement, and application.

[0099] FIG. 52 is a schematic view of a computer network in which embodiments may be implemented.

[0100] FIG. 53 is a block diagram illustrating an example embodiment of a computer node in the computer network of FIG. 52.DETAILED DESCRIPTION

[0101] A description of example embodiments follows.

[0102] As referenced hereinabove, a design space of a neuromuscular interface that aims to solve a muscle control problem could, according to an example embodiment, engineer a way muscle fibers are recruited (stimulation interface) or modify a phenotype of muscle fibers

[0103] Regarding stimulation interfaces, multiple attempts have been made to engineer the synthetic interface, either through muscle or nerve electrode designs [3], Additionally, multiple stimulation protocols have been developed to improve muscle fatigue resistance under FES. These approaches may reduce fatigue only narrowly, as the biophysical interaction between a motor nerve and electrical stimulation remain unchanged. Recently, a genetic engineering approach, where photosensitive ion channels (opsins) are exogenously expressed in motor neurons and axons, has been explored to achieve natural recruitment order- 12 - 5040745. vlDocket No. 4889-1034001under optical stimulation. This approach may completely remove the nerve electrical biophysics as it relies on opsin activation, and it was shown that muscles can be controlled in a natural manner, with natural recruitment profiles, improved controllability and fatigue resistance [2,4] . However, this approach relies on gene therapy, which still requires extensive scientific and clinical investment to be translated to humans.

[0104] Regarding engineering muscle fibers, approaches where glucose metabolism is augmented through chemical dosing have been explored and may provide minimal improvement in fatigue [5], Modifying the muscle fiber phenotype may stand as a more promising alternative. Skeletal muscle is one of the few organs that after full differentiation still maintains plasticity, meaning it can undergo reprogramming (transdifferentiation) to modify muscle fiber type. Seminal work has established example requirements for this transformation, indicating that a phenotype of a muscle is determined by nerve impulses the muscle receives [6], This requirement can enable phenotype transformation in fully differentiated muscles using electrical nerve stimulators, from slow-twitch to fast-twitch muscle and from fast-twitch to slow-twitch muscle. However, several limitations may have precluded clinical success [7], The phenotype change of muscle fibers may comprise a staged process rather than a single triggering event, whereas current strategies may employ an openloop stimulation approach. This usage of open-loop stimulation for a staged process can limit the accuracy of phenotype transformation that can be achieved [7], Another limitation of the open-loop stimulation approach can include that conditioning stimulation protocols may not be optimal as there is no continuous phenotype sensing, which may result in lengthy and uncomfortable protocols. Another limitation can include that when stimulation protocols are stopped, skeletal muscles can return or revert to their original fiber type if they receive signals from their original innervation. This process of revering to an original fiber type can limit the chronic use of skeletal muscles in their modified phenotype state.

[0105] As presented herein, example embodiments of devices, systems, and methods can be directed toward engineering skeletal muscle for enhanced control and applications of said engineered muscles, for example, as a biohybrid organ system. An example embodiment can be aimed at transforming muscle fiber type, which can comprise a variety of systems and methods to perform closed-loop adaptive stimulation to precisely transform muscle phenotype. The embodiment can comprise sensing of muscle-tendon dynamics (which may also be referred to herein as neuromuscular dynamics), which can allow for continuous estimation of a variety of muscle states such as activation, length, velocity, fatigue, among- 13 - 5040745. vlDocket No. 4889-1034001others. These states can be used to estimate a current muscle phenotype. Based on the current muscle phenotype, a closed-loop controller can determine optimal stimulation patterns for effective and rapid phenotypic transformation. Once a muscle is transformed, the controller can be used in a neuroprosthesis for fatigue-resistant control to restore motion, or the controller can be volitionally used for improved muscle function by elderly individuals, for non-limiting examples of applications. This adaptive stimulation approach can allow for precise transformation of muscle fiber types. Due to consistent phenotype feedback (e.g., phenotype sensing or estimation based on the muscle-tendon dynamics sensed), intermediate types (e.g., phenotypes) such as 2A can be achieved, where fibers preserve fatigue-resistant and high force contractile properties. Additionally, this approach can enable use of a transformed muscle right after transformation, as a precise phenotype is known throughout the stimulation protocol. Furthermore, if the muscle starts the process of going back to its original phenotype, such deviations can be sensed, and the stimulation protocol can be resumed.

[0106] Another example embodiment can be aimed at modifying a stimulation interface, which can comprise systems and tissue engineering methods to modify an axonal composition of a nerve innervating a muscle for improved fiber recruitment. Motor denervation can remove volitional control of the skeletal muscle, and reinnervation with a peripheral nerve with desired recruitment biophysics, based on axonal cross-section area, can promote enhanced fiber recruitment under electrical stimulation. Additionally, reinnervation of the muscle with a pure sensory nerve (which may be characterized by a more uniform axon size compared to a motor nerve), can preserve the muscle self-sustainability while enabling artificial computer control via a restored neuromuscular junction between the sensory nerve and muscle fibers. Such an engineered construct may be referenced herein as a myoneural actuator (MNA), which can be used for a variety of applications, such as, for non-limiting example, to create a biohybrid organ system or as an enhanced stimulation interface for the control of muscles.

[0107] Approaches such as the embodiments described hereinabove can be used in combination. For example, after an MNA is reinnervated, stimulation patterns through its sensory nerve can be used to transform the phenotype of the muscle. Such combinatorial approaches can provide several advantages, most notably, that the MNA may have native motor innervation removed, which can prevent the muscle from going back to its native phenotype. Another advantage can include that impulse patterns could be more natural- 14 - 5040745. vlDocket No. 4889-1034001through the enhanced recruitment interface. This enhanced recruitment interface can allow for a fully customizable actuator that can be used for a myriad of applications, such as creating biohybrid organs and neuroprosthetic systems.

[0108] FIG. 1 illustrates an example embodiment of a system 100 for transforming a phenotype of a muscle 102 of a human or animal. The system 100 can comprise a sensor 104 configured to detect signals associated with the muscle and a signal generator 106 configured to stimulate the muscle. In some embodiments, the sensor 104 can comprise magnetomicrometry (MM) sensors and magnetic beads implantable in the muscle 102. The system 100 can further comprise a processor, e.g., a controller 108, configured to determine a stimulus pattern based on neuromuscular dynamics, which can comprise muscle states 110, computed from the signals detected by the sensor 104, wherein the neuromuscular dynamics computed can be indicative of a current phenotype of the muscle. The processor 108 can be further configured to deliver the stimulus pattern to the muscle via the signal generator 106.

[0109] In some embodiments, the signal generator 106 can be functionally coupled to the muscle 102 via nerve 112 innervating the muscle 102. The signal generator can couple to the nerve, for example, using a nerve cuff 114, at a point between a central nervous system (CNS) 116 of the human or animal and the muscle 102. In other embodiments, the signal generator 106 can be configured to stimulate the muscle 102 using other modalities, for example, non-invasive modalities such as surface electrodes, ultrasound stimulation, or magnetic stimulation, as non-limiting examples. The stimulus pattern determined and delivered can cause a transformation of the muscle toward a target phenotype. In some embodiments, the signal generator 106 can be a component configured to generate an electrical signal, which can be encoded with spatially varying patterns.

[0110] Additional example embodiments of a system for transforming a phenotype of a muscle are further described hereinbelow.

[0111] Implantable neuromuscular system

[0112] According to an example embodiment, a system for transforming a phenotype of a muscle can comprise an implantable neuromuscular system. The system can comprise elements for stimulation and sensing.

[0113] FIG. 2A illustrates an example embodiment of an implantable system 200a for transforming a muscle 202a of a human or animal. The system 200a can comprise a sensor configured to detect signals associated with the muscle, which can comprise an electrical- 15 - 5040745. vlDocket No. 4889-1034001sensor (e.g., an implanted electrode) 205a or magnetic sensor (e.g., implanted magnetic beads) 204a. The magnetic sensor 204a can comprise measuring changes in magnetic fields of the magnetic beads to determine dynamics of a muscle such as length, contraction, or velocity, for non-limiting examples. The system 200a can further comprise a signal generator, e.g., an implantable pulse generator (IPG) 206a, configured to stimulate the muscle 202a. The IPG 206a may be coupled to a nerve 212a innervating the muscle 202a, for example, at one or more of a proximal nerve cuff 214 or a distal nerve cuff 215 (wherein proximal and distal are in reference to a central nervous system). The system 200a can further comprise a processor, which, in some embodiments, can be contained within the IPG 206a. The processor can be configured to determine a stimulus pattern based on neuromuscular dynamics computed from the signals detected, wherein the neuromuscular dynamics computed are indicative of a current phenotype of the muscle 202a, and to deliver the stimulus pattern determined to the muscle 202a, wherein the stimulus pattern determined and delivered can cause a transformation of the muscle toward a target phenotype.

[0114] In some embodiments, the signal generator, e.g., the IPG 206a, can be electrically coupled to a nerve 212a innervating the muscle. In some further embodiments, the IPG 206a can be electrically coupled to the nerve 212a at multiple locations, e.g., the first or proximal nerve cuff 214 and the second or distal nerve cuff 215. As further described hereinbelow, the multiple locations can be used to control a stimulus delivered to the nerve 212a, for example, using a nerve block. The sensor can comprise one or more of the implanted magnetic beads 204a, which can comprise beads in the muscle 212a and in a tendon 218a coupled to the muscle, or the implanted electrodes 205a. According to some embodiments, the IPG 206a can control the stimulation patterns to the nerve cuffs 214, 215 and muscle electrode 205a and can record from the nerve cuffs 214, 215 or muscle electrodes 205a. The magnetic beads 204a can be tracked wirelessly from outside the skin with a wearable sensor array 9as further illustrated hereinbelow with respect to FIG. 2B.

[0115] According to some embodiments, the signal generator configured to stimulate the muscle can comprise an implantable pulse generator (IPG), e.g., the IPG 206a. The IPG 206a can deliver programmable stimulation patterns and can be configured receive power and communication signals through a skin boundary, for example, using technologies that include but are not limited to inductance power transfer, resonant inductive coupling, capacitive wireless transfer, magnetodynamic coupling, and ultrasound power transfer. The IPG 206a can connect to a neural tissue interface, such as a nerve or muscle electrode for stimulation.- 16 - 5040745. vlDocket No. 4889-1034001In addition to electrical stimulation, optical stimulation with or without exogenous protein expression can be used as a stimulation modality. In some embodiments, at least one neural tissue interface may be required, and additional interfaces may be implemented for, e.g., a case wherein a proximal electrical nerve block is required.

[0116] According to some embodiments, the sensor can comprise one or more of the following sensors. One or more muscle electrodes can be configured to sense electromyographic (EMG) signals. In some embodiments, the electromyographic signals can be acquired and transmitted by the IPG. One or more magnets can be implanted in muscle and tendon tissue to sense tissue displacement using magnetomicrometry [8,9], The implanted magnets can be tracked via arrays of magnetic field sensors placed on the surface of the skin. These arrays can track the changes in magnetic fields corresponding to tissue movement, providing a measurement of tissue displacement.

[0117] Minimally invasive neuromuscular system

[0118] Another example embodiment of a system for transforming a phenotype of a muscle can comprise a minimally invasive neuromuscular system. The minimally invasive system can comprise non-invasive or minimally invasive stimulation modalities including, but not limited to, transcutaneous electrical stimulation, fine-wire stimulation, ultrasound stimulation, and magnetic stimulation, of either the nerve or muscle.

[0119] FIG. 2B illustrates an example embodiment of a minimally invasive system 200b for transforming a muscle 202b of a human or animal. The system 202b can be similar to the system 202a of FIG. 1 A and can comprise one or more sensors configured to detect signals associated with the muscle 202b. The sensors can comprise, for non-limiting example, a surface electrode 205b, which can be placed on a skin surface in close proximity to a nerve 212b and the muscle 202b, or magnetic sensors comprising a magnetic field sensor 220, which can a wearable magnetic field sensor array, configured to track magnetic beads 204b, which can be implanted through a minimally-invasive procedure in muscle 202b and tendon tissue 218b. The system 200b can further comprise a signal generator 206b (e.g., a pulse generator that sits outside the skin) configured to stimulate the muscle 202b and a processor. In some embodiments, the processor may be contained in the PG 206b. The processor can be configured to determine a stimulus pattern for stimulating the muscle 20b, wherein the stimulus pattern is based on neuromuscular dynamics computed from the signals detected from the muscle 202b, and to deliver the stimulus pattern determined to the muscle 202b via- 17 - 5040745. vlDocket No. 4889-1034001the signal generator 206b. In some embodiments, the pulse generator 206b can comprise channels configured for performing stimulation and recording of the skin electrodes, e.g., the electrode 205b.

[0120] In some embodiments, the sensor can comprise, for non-limiting example, one or multiple electrodes, e.g., the electrode 205b, placed on the skin to sense electromyographic signals or one or multiple magnets 204b can be implanted in muscle and tendon tissue using a minimally invasive procedure using a delivery tool through skin. Similar to the implantable neuromuscular system 200a of FIG. 2 A, the magnets 204b can be tracked from outside the skin using magnetomicrometry. Other sensing techniques that can be used for sensing muscle-tendon dynamics can comprise, for non-limiting example, ultrasound, near infrared imaging, laser doppler vibrometry, and optical coherence tomography.

[0121] In addition to wearable sensing modalities, other modalities, which may include, for non-limiting example, computed tomography, metabolic profiling, genotyping, myosin heavy chain isoform profiling such as western blot, PCR, immunohistochemistry, and myosin ATPase activity, can be used to supplement wearable sensing to infer muscle phenotype. In some embodiments, elements from the implantable and minimally invasive systems can be used interchangeably based on functional requirements to generate multiple embodiments.Neuromuscular sensing

[0122] According to some embodiments, stimulation patterns, which can be delivered either from a central nervous system (CNS) of a human or animal or through a signal generator, e.g., the IPG 206a of FIG. 2A or the PG 206b of FIG. 2B, can generate neuromuscular dynamics that can sensed via sensors as described hereinabove, for example, the electrodes or the magnetic sensors. In some embodiments, EMG sensors, e.g., the electrodes 205a, 205b, can acquire signals that, after signal conditioning, can be input to a muscle dynamics model that can estimate muscle activation (a). In other embodiments, MM can sense changes in magnetic fields from implanted magnets 204a, 204b and can compute a distance between magnets, e.g., a distance between two or more magnets. In muscle tissue, such a distance can be directly correlated with muscle length (1). Muscle length can be differentiated to obtain muscle velocity (Idot). In some embodiments, muscle velocity can also be obtained from coils placed outside of the skin that can sense a changing magnetic field of the implanted magnets for a muscle velocity estimation. Muscle activation, length and velocity can be input into a muscle-tendon model that can compute force generated by- 18 - 5040745. vlDocket No. 4889-1034001the muscle (F). In tendon tissue, the distance between magnets can be directly correlated with tendon length. Tendon length can be directly correlated with tendon force. Tendon length can be input into a tendon dynamic model to compute tendon force. As tendon is linearly arranged with muscle, tendon force may be equal to muscle force (F).

[0123] FIGS. 3A-3F are schematic representations of neuromuscular dynamics of a muscle, the neuromuscular dynamics being computed from signals detected from said muscle, according to example embodiments. While the example embodiments are presented individually for ease of visualization, it should be understood that combinations of one or more of the dynamics described can be used.

[0124] FIG. 3A illustrates schematically an of a system 300a for sensing neuromuscular dynamics (which may be similar to the computing the neuromuscular dynamics described hereinabove) or muscle-tendon dynamics based on signals detected from a muscle.Stimulation 306a, which may be delivered from a CNS or signal generator to a muscle 302a, can generate muscle-tendon contraction dynamics that can be sensed one or more sensors. Electrodes can be configured to sense electrical activity generated by muscles, electromyography (EMG) 305a, from which muscle activation (a) can be determined at a muscle dynamics block 324. Magnetic beads in the muscle can sense tissue displacement (e.g., MM 304a) which is used to determine muscle length (1). Muscle length can be differentiated 325 to obtain muscle velocity (Idot). A muscle-tendon model 326 can be used to determine muscle force (F) from muscle activation, muscle length and velocity inputs. Magnetic beads in tendon tissue can sense displacement (e.g., MM signals of the tendon) which is used to determine tendon length, which can be used to compute tendon force through a tendon dynamic model 327.

[0125] FIG. 3B illustrates schematically an of a system 300b for sensing neuromuscular dynamics or muscle-tendon dynamics of a muscle 302b, the neuromuscular dynamics comprising peak timing. Stimulation patterns 306b can arrive at the muscle 302b, for example, from a nervous system of a human or animal or a signal generator, and neuromuscular dynamics can be sensed, for example, by sensing modalities such as electromyography 304b and magnetomicrometry 305b. In some embodiments, the stimulation patterns can comprise, for non-limiting example, low frequency and short duration stimulation patterns, can be aimed at generating twitch-like responses in muscles. Muscle states such as activation, length, or force can be sensed or computed and analyzed. Time from stimulation to peak force can be estimated and recorded for multiple trials. A plot- 19 - 5040745. vlDocket No. 4889-1034001328a of FIG. 3B illustrates example peak timing of fast and slow twitch muscle phenotypes. According to some embodiments, slow-twitch muscles may show a larger time to achieve peak force (ts), compared to fast-twitch muscles (tf). Accordingly, time to peak force may serve as a measurement to estimate muscle fiber composition.

[0126] FIG. 3C illustrates schematically an of a system 300c for sensing neuromuscular dynamics of a muscle 302c, the neuromuscular dynamics comprising velocity. The system 300c may be similar to the system 302b of FIG. 3B and may be configured to evaluate twitch muscle contractions. The system 300c can comprise EMG 304c and MM 305c of the muscle 302c responsive to a stimulation pattern 306c. Based on the signals detected from the sensing modality, muscle velocity can be recorded (which can be performed over several trials) and processed, and the muscle velocity can be subsequently used to determine current contractile properties of the muscle. A plot 328b illustrates example measurements of velocity of fasttwitch and slow-twitch muscles. According to some embodiments, fast-twitch muscles have a greater velocity compared to slow-twitch muscles.

[0127] FIG. 3D illustrates schematically an example embodiment of a system 300d for sensing neuromuscular dynamics of a muscle 320d, the neuromuscular dynamics comprising fatigue. The system 300d may be similar to the system 302b of FIG. 3B and may be configured to evaluate twitch muscle contractions. The system 300d can comprise EMG 304d and MM 305d of the muscle 302d responsive to a stimulation pattern 306d. According to an example embodiment, the stimulation patterns 306d can comprise high frequency and long duration patterns, which can be aimed at generating a sustained, tetanic muscle contraction. Neuromuscular states (which may also be referred to herein as neuromuscular dynamics) such as muscle force, muscle length, muscle activation can be recorded and processed. A first plot 328c-l illustrates example measurements of force that can be applied by the muscle 302d over time and a second plot 328c-2 illustrates example measurements of fatigue rate of the muscle 302d over time.

[0128] Muscle-tendon dynamics over time can be used to determine current fatigue properties of the muscle. According to some embodiments, fast-twitch muscles may show a pronounced rapid decay of force (e.g., as illustrated by the plot 328c-l) during sustained contraction. In contrast, slow-twitch muscles may show sustained force for longer periods. As such, measurements of muscle fatigue over time can be used to determine fiber composition.

[0129] According to some embodiments, the measurements of muscle fatigue can be used to estimate fatigue rate, which may be helpful for determining with higher precision fatigue- 20 - 5040745. vlDocket No. 4889-1034001dynamics of the muscle. Fatigue rate may be estimated by differentiating time-domain neuromuscular state measurements (e.g., as illustrated in the plot 328c-2). According to some embodiments, fast-twitch muscles may exhibit exponential fatigue dynamics whereas slow-twitch muscles may show logarithmic-like fatigue dynamics. Both fatigue and fatigue rate measurements can be used to estimate muscle fiber type composition.

[0130] FIG. 3E illustrates schematically an of a system 300e for sensing neuromuscular dynamics of a muscle 302e, the neuromuscular dynamics comprising fatigue. The system 300e may be similar to the system 302b of FIG. 3B and may be configured to evaluate twitch muscle contractions. The system 300e can comprise sensors configured for EMG 304e and MM 305e sensing of the muscle 302e responsive to a stimulation pattern 306e. According to an example embodiment, the stimulation pattern 306e can comprise one or more frequencies over a range of frequencies, which may be useful for dynamically stimulating the muscle 302e. Neuromuscular states (which may also be referred to herein as neuromuscular dynamics) such as muscle force, muscle length, muscle activation can be recorded and processed. In some embodiments, a frequency versus neuromuscular state plot, for example, the plot 328d, can inform a frequency at which a neuromuscular state is at a maximum. Said maximum of a given neuromuscular state can be helpful for determining contractile properties of the muscle. According to some embodiments, slow-twitch muscles may be characterized by having peak muscle states at lower frequencies, whereas fast-twitch muscles may be characterized by having peak states at higher frequencies.

[0131] FIG. 3F illustrates schematically an example reference library of neuromuscular dynamics that can be generated using of systems 300f for sensing neuromuscular dynamics of muscles 302f-l, 302f-2. The systems 300f may be similar to the system 302b of FIG. 3B and may be configured to evaluate one or more contractile properties of the muscles 302f-l, 302f-2. The system 300f can comprise sensors configured for, for non-limiting examples, EMG and MM sensing, of the muscles 302f-l, 302f-2 responsive to a stimulation pattern 306f. While two muscles 302f-l, 302f-2 are illustrated, a plurality of muscles can be used to generate the reference library of neuromuscular dynamics.

[0132] According to an example embodiment, stimulation procedures, for example, the stimulation patterns described hereinabove (which can comprise, for non-limiting example, low frequency short duration patterns, high frequency long duration patterns, or patterns having one or more frequencies), can be applied to a variety of muscle instances with distinct fiber compositions, to create a library of muscle phenotypes. A muscle instance can be a- 21 - 5040745. vlDocket No. 4889-1034001muscle state at a particular time point, a particular muscle from the same individual, or a muscle from multiple individuals.

[0133] In some embodiments, measurements of a muscular or neuromuscular state (for non-limiting example, muscle force, muscle length, and muscle activation) acquired responsive to such stimulation patterns can be used to compare different muscles, for example, between muscles of a same individual and across individuals. The plot 328e-l illustrates example neuromuscular dynamics versus stimulation frequency for four muscles, M1-M4). Additionally, muscle states can be tracked longitudinally to estimate muscle phenotype transformation (for example, when a muscle goes from a given phenotype, e.g., type 2X to another phenotype, e.g., 2A), which may be responsive to stimulation signals from the signal generator or the CNS of a human or animal. A phenotype transformation plot, e.g., the plot 328e-2, can be used to determine an exact phenotype state of a muscle, which can be based on neuromuscular dynamics characteristics of muscles, for example, the plot 328e-l.

[0134] According to an example embodiment, the plot 328e-l may describe curves that can be obtained during muscle characterization stimulation protocols, wherein a muscle can be stimulated using different frequencies and a corresponding signal (e.g., MM or EMG as non-limiting examples) can be detected by a sensor. Neuromuscular states can be computed based on the signals detected. With respect to the plot 328e-l, the x-axis may describe different stimulation frequencies and the y-axis may describe a measured variable, for example, the neuromuscular dynamics or states. In some embodiments, depending on a phenotype of a particular muscle, a feature of a plotted curve, e.g., a peak or a tough, can occur at different frequencies. For non-limiting example, peaks may occur at lower frequencies for slow twitch muscles and at higher frequencies for fast twitch muscles. The different curves may represent different muscles and these curves can be obtained at multiple timepoints throughout the phenotype transformation of the muscles, which may enable generation of a plot similar to the plot 328e-2.

[0135] According to some embodiments, the plot 328e-2 may explain a progression a muscle goes when a phenotype of the muscle is being transformed. The plot 328e-2 can be built from the curves from the plot 328e-l, and may comprise multiple curves similar to the plot 328e-l at different time points. In some embodiments, a first part of the plot 328e-2 (e.g., the stimulation portion on the x-axis) can include when the muscle is being actively stimulated to transform the phenotype of the muscle, e.g., going from type 2X to type 2A, and from 2A to type 1. The y-axis may represent phenotype stages. A second part of the plot- 22 - 5040745. vlDocket No. 4889-1034001(e.g., the recovery portion on the x-axis) can include when the muscle is not being stimulated anymore and the muscle goes back to its original phenotype, going from type 1 to type 2A, and from type 2A to 2X. According to some embodiments, the plot 328e-2 may represent time dynamics of phenotype transformation and different stages the muscle goes during this transformation.

[0136] In some embodiments, stimulation patterns may come from a CNS of the human or animal and, based on longitudinal data collection over a range of different activities, the muscle phenotype can be determined from the stimulation patterns of the CNS. Yet another embodiment can combine stimulation patterns from both the CNS and the signal generator to generate muscle phenotype estimates using larger data sets and a variety of dynamics.Closed-loop adaptive stimulation for phenotype transformation

[0137] According to some embodiments, neuromuscular sensing can provide a variety of muscle state measurements (e.g., neuromuscular dynamics) that can be useful for building phenotype estimators. The ability to generate a variety of stimulation patterns from a signal generator, e.g., the IPG or PG described herein with respect to FIGS. 2 A and 2B, respectively, can enable dynamic stimulation of the muscle, which may be useful for system identification procedures. In order to design a closed-loop controller, a system identification procedure can comprise generating a mathematical model of a current phenotype of a muscle for later use in the controller.

[0138] FIG. 4 is a block diagram of an example muscle model architecture 430 that can be used by an example embodiment of a system for transforming a phenotype of a muscle. A proposed muscle model architecture 430 can comprise, based on a stimulus 406 delivered to the muscle, computing neuromuscular dynamics, which can comprise, for non-limiting example, a muscle recruitment block 432, a muscle activation dynamics block 434, forcelength curves 436, force- velocity curves 438, and a segment dynamics block 440. The muscle recruitment block 432, the muscle activation dynamics block 434, the force-length curve 436, and the force-velocity curve 438 can be used to estimate muscle force 441. The segment dynamics block can be used to estimate joint torque 442 of the muscle from muscle force 441.

[0139] According to an example embodiment, such a muscle model architecture 430 can estimate the dynamics generated by a muscle, for non-limiting example, torque 442, based on a given stimulation input 406. Depending on a given application, a model can also include a- 23 - 5040745. vlDocket No. 4889-1034001version for isometric conditions (which may comprise a static nonlinearity (SNL), which can account for the recruitment curve of the muscle, and a linearized dynamic system (LDS), which can account for the muscle activation and muscle-tendon dynamics.

[0140] FIGS. 5A-5D are block diagrams of example system identification procedures for estimating characteristics of a muscle that may be used by example embodiments of a system for transforming a phenotype of a muscle. For example, the system identification procedures can represent a model for estimating dynamic functions of a muscle based on an input (e.g., a stimulus) and an output (e.g., neuromuscular dynamics or state) of the muscle. According to some embodiments, such procedures can be used to estimate or infer a current state or phenotype of a muscle.

[0141] FIG. 5A is a block diagram of an example overall system identification procedure 530a for inferring a current state or phenotype of a muscle. The system identification can comprise an input (e.g., a stimulus or stimulus pattern) 544a, a signal generator 506 (e.g., an IPG, which may be configured to deliver said stimulus to a muscle), the muscle 502, sensors 504, and an output 545a, which may comprise signals associated with the muscle detected by the sensors. The muscle may, responsive to the stimulus, exhibit muscle dynamics, as further described hereinbelow.

[0142] FIG. 5B is a block diagram of example muscle dynamics of a muscle. The muscle dynamics may be exhibited by the muscle 502 of FIG. 5A in response to an input, e.g., a stimulus 544b, and may comprise muscle recruitment 532 as well as muscle activation and muscle-tendon dynamics 534. Said muscle recruitment and said muscle activation and muscle-tendon dynamics may be associated with an output 545b, e.g., a measurable force generated by the muscle. In some embodiments, the block diagram of FIG. 5B may represent an overall muscle model.

[0143] FIG. 5C is a block diagram of example dynamic functions 546, 547 of the muscle based on the muscle dynamics 532, 534 of FIG. 5B. The dynamic functions 546, 547 of the muscle can comprise, responsive to an input 544c (e.g., a stimulus) and producing an output 545c, a static nonlinearity block (SNL, which may comprise a function X(u) representing muscle recruitment) 546 and a linearized dynamic system block (LDS, which may comprise a function H(s) representing muscle activation and muscle tendon dynamics) 547. According to some embodiments, the block diagram of FIG. 5C may be a mathematical representation of the muscle model of FIG. 5B. As described hereinabove, the SNL block 546 and the LDS- 24 - 5040745. vlDocket No. 4889-1034001block 547 can represent the muscle recruitment and the muscle activation and muscle-tendon dynamics of FIG. 5B.

[0144] FIG. 5D is a block diagram of an example method 530d employable by a system identification procedure for inferring a current phenotype of a muscle. In some embodiments, the method can determine dynamic functions, e.g., the dynamic functions X(u) 546 and H(s) 547 of FIG. 5C, of a muscle based on an input (e.g., a stimulation) and an output (e.g., a neuromuscular state or muscle variable, e.g., force for non-limiting example, output) data, which may be referenced herein as u and j', respectively. The method may further comprise terms v, v7, x7, and v2, which may represent intermediate functions estimated at corresponding steps. The method 530d can include estimating 501 an LDS (H(s)) based on the input u and the output y. The method 530d can further include estimating 503 v7, wherein vl is based on an inverse of the LDS, and estimating 505 xl based on vl and the input u. The method 530d can further comprise estimating 507 v2 based on xl and the input u and estimating 509 an updated LDS based on v2 and the output y. The method can comprise repeating 511 steps 503, 505, 507, 509 can be repeated until the model converges. The method 530d can further comprise, upon convergence of the model, identifying 513 current SNL and the LDS characteristics of the muscle.

[0145] The system identification procedure can be performed until the model converges. According to an example embodiment, a convergence criteria can comprise, for non-limiting example, X(u) and H(s) functions that fit experimental data with an R2(e.g., coefficient of determination) of less than 0.90. After the model converges, parameters (e.g., the dynamic functions) from the identified system can be used to estimate the phenotype of the muscle as they provide dynamic information that may be dependent on one or more of fiber contractile properties and muscle recruitment. In some embodiments, the final X(u) and H(s) functions may form the muscle model described herein with reference to FIG. 5C, which may serve as a mathematical representation of the muscle based on the input u and the output y data. In some embodiments, the mathematical model may subsequently be inverted to estimate a stimulation that should be given to a muscle based on a desired muscle variable, for example, as further referenced hereinbelow by reference estimator 630 of FIG. 6.

[0146] In another embodiment of the system identification procedure, information from aforementioned non-wearable sensing modalities (e.g., MRI, PCR, myosin heavy chain isoform profiling, for non-limiting examples) can help validate a dynamic estimation of a system identification procedure.- 25 - 5040745. vlDocket No. 4889-1034001

[0147] According to some embodiments, the system identification procedure can provide a model that estimates a muscle state output based on particular stimulation inputs. This model can be inverted to determine a required stimulation, for example, a stimulation pattern for changing a phenotype of a muscle, based on a given muscle state (e.g., a current phenotype of the muscle or a desired phenotype of the muscle) [4], The model can be used in a closed-loop adaptive control architecture for precisely stimulating the muscle for phenotype transformation.

[0148] FIG. 6 illustrates schematically an example closed-loop stimulation architecture 650 that can be used by an example embodiment of a system for transforming a phenotype of a muscle 602. The stimulation architecture 650 can comprise sensors 604 configured to detect signals, e.g., contraction dynamics, from a muscle 602, which can be used to compute neuromuscular dynamics such as muscle force, muscle length, muscle activation, muscle velocity, or muscle fatigue rate. A reference estimator 630 can comprise an inverse of a model estimated from a system identification procedure, e.g., the models described herein with respect to FIGS. 5A-5D. The reference estimator 630 can, based on the muscle state, determine a stimulation that can drive the muscle 602 to a current phenotype of the muscle 602. A desired phenotype block 652 can indicate a stimulation value that is required to achieve a desired (target) muscle phenotype. A processor, e.g., a phenotype controller 608, can be informed by the reference estimator 630 and the desired phenotype block 652, and can compute stimulation parameters (e.g., based on a difference between outputs from the reference estimator and the desired phenotype block) that can be transmitted to a signal generator 606, e.g., an IPG. The signal generator 606 can be configured to deliver a stimulation pattern to the muscle 602 based on the stimulation parameters. In some embodiments, a processor as described herein with reference to FIG. 1, can comprise the reference estimator 630, the desired phenotype 652, and the phenotype controller 608.

[0149] FIG. 7 illustrates schematically another example closed-loop stimulation architecture 750 that can be used by an example embodiment of a system for transforming a phenotype of a muscle 702. The architecture 750 can be similar to the architecture 650 of FIG. 6 and can further comprise using information from an auxiliary sensing system 754, for example, a non-wearable sensing system (e.g., MRI, PCR, myosin heavy chain isoform profiling) that can provide morphological features of the muscle, to inform a determination of a current phenotype of the muscle 702. Elements of the architecture of FIG. 7 similar to those of the architecture of FIG. 6 are indicated by like numbers but incremented by 100.- 26 - 5040745. vlDocket No. 4889-1034001

[0150] The stimulation architecture can comprise sensors 704 configured to detect signals, e.g., contraction dynamics, from the muscle 702, which can be used to estimate neuromuscular dynamics such as muscle force, muscle length, muscle activation, muscle velocity, or muscle fatigue rate. A reference estimator 730 can comprise an inverse of a model estimated from a system identification procedure, e.g., the model described herein with respect to FIGS. 5A-5D. A phenotype controller 708 can be informed by the reference estimator 730, a desired phenotype block 752, and the auxiliary sensing system 754 (which may comprise a morphology block provided by the auxiliary sensing system described hereinabove), and can compute stimulation parameters (e.g., based on a difference between outputs from the reference estimator and the desired phenotype block) that can be transmitted to a signal generator 706, e.g., an IPG. The signal generator can be configured to deliver a stimulation pattern to the muscle based on the stimulation parameters.

[0151] According to some embodiments, an approach that does not require a model can be used in place of an inverse of a model from the system identification procedure. Such an approach can utilize a stimulation versus muscle state curves, for example, the curves 328e-l, 328e-2 described hereinabove with respect to FIG. 3F, to estimate the stimulation that matches a current muscle phenotype. The desired phenotype block 752 can also use such phenotype curves (e.g., the stimulation versus muscle state) to estimate the stimulation value required based on a target phenotype. According to some embodiments, the phenotype controller 708 can be configured to determine a stimulation pattern based on stimulation patterns associated with the reference estimator 730 and the desired phenotype block 752. The phenotype controller 708 can determine stimulation parameters (e.g., frequency, pulse width, amplitude, or duration for non-limiting examples) associated with such a stimulation pattern that can be used by a signal generator to stimulate the muscle.

[0152] FIG. 8 illustrates schematically another example closed-loop stimulation architecture 850 comprising a nerve block that can be used by an example embodiment of a system 800 for transforming a phenotype of a muscle 802. The architecture can be similar to the architecture 650 of FIG. 6, and similar elements are indicated by like numbers but incremented by 200. The architecture 850 can be configured to deliver a nerve block 856, which can comprise, for non-limiting example, a kilohertz frequency alternating current (KHFAC).

[0153] The architecture 850 comprises sensors 804 configured detect signals associated with a muscle 802 that can be caused by stimulation from a CNS 816 of a human or animal or- 27 - 5040745. vlDocket No. 4889-1034001a signal generator 806, e.g., an IPG. The signal generator 806 can be coupled to a nerve 812 innervating the muscle at a proximal site 814 and a distal site 815, and the nerve block 856 (KHFAC) can be delivered to the proximal site 815 to block neural signals traveling in an antidromic direction towards the CNS 816. Based on the signals detected by the sensors 804 (e.g., based on neuromuscular dynamics such as muscle force, muscle velocity, muscle activation, muscle length, or muscle fatigue rate computed from the signals detected), a reference estimator 830 can be configured to determine a current phenotype of the muscle (e.g., determine a stimulation that can drive a muscle to a current phenotype of the muscle). A phenotype controller 808, which may be informed by the reference estimator 830 and a desired phenotype block 852, can be configured to determine stimulation pattern comprising stimulation parameters to be transmitted to the signal generator 806, which may be configured to stimulate the muscle 802. The signal generator 806 can further be configured to deliver the nerve block 856 for blocking the antidromic stimulation toward the CNS 816.

[0154] According to some embodiments, some stimulation patterns, for example, high-frequency stimulation configured to transform muscle from slow to fast-twitch fibers, could cause undesired sensations to a human or animal. Additionally, CNS signals may revert a muscle phenotype, e.g., from a desired or target phenotype toward an original phenotype. For such situations, a nerve block, which can comprise, as a non-limiting example, a kilohertz high-frequency AC (KHFAC) electrical signal, which can be delivered at the proximal region of the nerve, can block neural signaling from the CNS to the muscle and also prevent the stimulation pattern from the IPG from traveling to the CNS

[0010] , In some embodiments, concurrently to the stimulation pattern being delivered to the muscle for phenotype transformation, the IPG can deliver the high-frequency stimulation (nerve block) at the proximal nerve section. In another embodiment, the KHFAC nerve block may not be applied concurrently, but in situations when the CNS is naturally activating the muscle with signals that differ from those of a desired phenotype. In such embodiments, the nerve block can prevent the muscle from going back to its original phenotype. Such a nerve block can be performed for as long as the muscle needs to be kept in the desired phenotype

[0155] According to some embodiments, after a closed-loop adaptive stimulation architecture is implemented and a desired muscle phenotype has been achieved, a phenotype of a muscle may need to be continuously evaluated to ensure that the muscle is not reverting. Such a reversion of the muscle to an original phenotype may occur when, for example, during normal operation, signals from either a CNS of a human or animal or artificial stimulation- 28 - 5040745. vlDocket No. 4889-1034001differ from the desired phenotype (e.g., a stimulation pattern associated with the desired phenotype. As described hereinabove, neuromuscular dynamics can be continuously monitored by a sensor configured to detect signals associated with the muscle. A phenotype evaluator can be configured to determine if the muscle can be used for normal operation or if further phenotype transformation, for example, by comparing current muscle states with a previously saved desired phenotype.

[0156] FIG. 9 illustrates schematically another example closed-loop stimulation architecture 950 usable during normal operation of the muscle by an example embodiment of a system 900 for transforming a phenotype of a muscle 902. The architecture 950 may be similar to the architecture 850 of FIG. 8 and similar elements may be indicated by like reference numbers but incremented by 100. The architecture 950 can comprise the sensors 904 configured to detect signals from the muscle 902. Based on the signals detected by the sensors 904 (e.g., based on neuromuscular dynamics computed from the signals detected), a phenotype evaluation block 958 can determine if the phenotype of the muscle is appropriate for normal operation 960. If the phenotype is appropriate, the muscle 902 can continue to be used for normal operation 960 (e.g., actuating the muscle to perform a function of said muscle). If the phenotype is not appropriate, for example, the muscle 902 is reverting toward an original phenotype of the muscle 902, the architecture 950 can comprise elements for transforming the phenotype of the muscle 902. The architecture 950 can comprise a reference estimator 930 and a desired phenotype block 956 configured to feed a phenotype controller 908, the phenotype controller 908 configured to determine a stimulus pattern deliverable to the muscle via a signal generator 906, e.g., an IPG.

[0157] According to some embodiments, responsive to a muscle state remaining unchanged or changed only marginally (e.g., with respect to a desired or target phenotype after transformation), no phenotype transformation may be executed and the muscle can be operated normally, e.g., by the signal generator. Significant changes in a phenotype of a muscle may cause muscle states to be passed to the reference estimator. Similar to the architectures 650, 750, 850 described hereinabove with respect to FIGS. 6-8, the phenotype controller 908 can determine stimulation patterns based on, from the reference estimator 930, a stimulation that can drive a muscle to the current phenotype of the muscle (which has deviated from the target phenotype) and on, from the desired phenotype block, a stimulation associated with the target phenotype.- 29 - 5040745. vlDocketNo. 4889-1034001

[0158] According to some embodiments, the architecture 950 comprising such a feedback loop can be implemented at different operational rates (e.g., frequency of evaluating a muscle) depending on an application and desired phenotype. The phenotype evaluation block can also determine if changes to normal operation of a muscle are required, which may depend on functional requirements of the muscle. For example, it has been reported that muscle actuation in cardiomyoplasty could revert the desired phenotype of the muscle

[0011] , In such embodiments, the phenotype evaluator 958 can be used to determine if stimulation protocols need to be resumed for phenotype transformation or if stimulation actuation protocols need to be modified to reduce the phenotype transformation reversal.

[0159] FIG. 10 illustrates schematically an example closed-loop stimulation architecture 1050 that can be used by an example embodiment of a system 1000 for transforming a phenotype of a muscle 1002 of a myoneural actuator (MNA) 1062. The myoneural actuator 1062, as further described hereinbelow, may aim to repurpose biological skeletal muscle as an actuator for computer-based control. According to some embodiments, volitional control of the skeletal muscle by a CNS of a human or animal can be prevented by denervation of a motor nerve, e.g., motor nerve 1013, of the skeletal muscle. Control of the skeletal muscle by a computer-based system can be established by reinnervation using another nerve, for example, a purely sensory nerve, e.g., sensory nerve 1012. Sensory reinnervation can enable muscle sustainability, restored neuromuscular junction, and an enhanced recruitment interface in the myoneural actuator.

[0160] According to some embodiments, the architecture 1050 can be similar to the architecture 650 of FIG. 6 but applied to the myoneural actuator 1062. Said architecture 1050 can comprise sensors 1004 configured to detect signals associated with the muscle 1002, e.g., signals of contractions of the muscle 1002. A reference estimator 1030, which can be configured to infer a current phenotype of the muscle, and a desired phenotype block can feed a phenotype controller 1008. The phenotype controller 1008, which may also receive an input from a desired phenotype block 1056, can be configured to determine a stimulus pattern, comprising a set of stimulus parameters), transmittable to a signal generator 1006 for delivery to the muscle 1002. The stimulus pattern determined and delivered to the muscle 1002 can enable further engineering of the myoneural actuator to transform a phenotype of the muscle 1002 of the myoneural actuator 1062.

[0161] FIG. 11 illustrates schematically an example closed-loop muscle control architecture 1150 usable by an example embodiment of a system 1100 for transforming a- 30 - 5040745. vlDocket No. 4889-1034001phenotype of a muscle 1102 to control the muscle 1102 after phenotype transformation. Both phenotype control and closed-loop control can be achieved through the same hardware stimulation and sensing architecture. After the muscle 1102 has been transformed to a desired phenotype, for example, from a fast to a slow twitch phenotype, the architecture 1150 can comprise sensors 1104 configured to sense muscle or neuromuscular dynamics. The muscle dynamics can be transmitted to a controller 1108 configured to determine a control signal based on a reference signal from a reference block 1156 and a current state (e.g., phenotype) of the muscle based on the muscle dynamics transmitted. An inverted model block 1130 can be used to estimate an additional (e.g., feedforward) control signal based on the reference signal. A stimulus pattern transmitted to a signal generator 1106, e.g., an IPG, can comprise a combination of the control signal determined by the controller and the additional control signal estimated by the inverted model. The signal generator 1106 can be configured to deliver the stimulus pattern to the muscle 1102.Example Embodiment of System for Neuromechanical Control and Phenotype Transformation

[0162] Example embodiments of systems, devices, and methods described herein can be used for neuromechanical control and for transforming a phenotype of a muscle of a human or animal. Specific application of the aforementioned uses of the embodiments can depend on a stimulus pattern delivered by a signal generator to the muscle. While

[0163] FIG. 12 illustrates schematically an example embodiment of a system 1200 for transforming a phenotype of a muscle of a human. The system, which may be similar to the system 200a of FIG. 2A, can comprise implantable technologies configured to interface with the central nervous system (CNS) and peripheral nervous system (PNS). In some embodiments, systems can comprise a CNS sensors (not shown), for non-limiting example, electrocorticography or intracortical implants configured to record brain signals for motor. The system can further comprise a signal generator configured to stimulate a muscle, for nonlimiting example, an epidural spinal cord implant 1206-1 configured to spatiotemporally stimulate spinal circuits to restore motor function or a peripheral implant 1206-2 configured to stimulate peripheral nerves and record neuromuscular signals to restore motor function.

[0164] An inset at the bottom of FIG. 12 illustrates schematically an example bioelectronic platform of the system 1200, which may be used for transforming a phenotype of a muscle. The bioelectronic platform can comprise direct muscle dynamics sensing (e.g.,- 31 - 5040745. vlDocketNo. 4889-1034001using a magnetometer array 1204-1 or nerve cuffs 1204-2 for non-limiting examples, a processor 1206 (e.g., an adaptive controller), and a wireless implant 1206-2 configured for stimulation of muscles, which can be used for closed-loop neuromechanical control.According to some embodiments, the bioelectronic platform can be capable of multi -joint closed-loop control. Furthermore, as noted in the top right comer of FIG. 12, technologies interfacing with a CNS may present greater risks, improved functional coverage, and decreased specificity when compared to technologies interfacing with a PNS. The bioelectronic platform can further comprise a transceiver 1264, which may be useful for communication between components, e.g., the processor 1208, the sensors 1204-1, 1204-2, and the implant 1206-2.

[0165] FIG. 13 illustrates schematically an example embodiment of a system 1300 for transforming a phenotype of a muscle of an animal, with insets illustrating embodiments of components of the system. The system 1300 can further be used for control and actuation of muscles of the animal. The system 1300 can comprise implantable components and can be configured to interface with a peripheral nervous system of the animal, e.g., a pig. According to the example embodiment, the system 1300 can comprise a sensor 1304, which can include a magnetomicrometry sensing subsystem comprising implantable magnetic beads 1366, e.g., mm paralyne-coated neodymium (N48SH) magnetic beads, that may be tracked from outside the skin using an external wearable magnetometer sensing array. The sensing array can comprise, as non-limiting examples, magnetic field sensors 1367, onboard electronics 1638 (e.g., a microcontroller, a communication module, and a battery), and a case 1369. The communication module can be a wireless communication module, e.g., a Wi-Fi module configured to transmit signals detected in association with a muscle to a computer 1308-1. The system 1300 can further comprise a neural stimulation component 1306 with an external processor 1370 and an implantable stimulator 1371 configured for electrical stimulation (for example, using stimulation channels 1372 configured to stimulate nerves through nerve cuffs 1314, as illustrated). The external processor may also be configured for communication between modules, e.g., WiFi-based communication. The neural stimulation component 1306 can perform a function of a signal generator. For example, the external processor 1370 can be configured to generate a stimulation pattern based on information received from a computer 1308-2 and the implantable stimulator 1371 can be configured to deliver the stimulus pattern to a muscle of the animal via the stimulation channels 1372.- 32 - 5040745. vlDocket No. 4889-1034001

[0166] The computer 1308-1 can be configured to compute neuromuscular dynamics, for example, by running a tracking algorithm in real-time to estimate a 3D position of each magnetic bead 1366 and to compute a distance between two given beads. The distance between the beads can change as the muscle contracts, which may effectively capture muscle length dynamics. The computer 1308-1 can be configured to determine the stimulus pattern to be delivered to the muscle of the animal, for example, by implementing a closed-loop control component 1350 integrating feedback and feedforward architectures, as further described hereinabove with respect to FIG. 11. According to an example embodiment, the closed-loop control component can comprise a controller 1374 configured to receive muscle state (muscle length) from the computer 1308-2 and to compare the value with a desired state of a desired state block 1356 to compute an error between a current state and the desired state. The desired state block 1356 can comprise, for non-limiting example, a pre-defined trajectory obtained from real-world gait neuromechanical parameters. The error can be used by a feedback controller 1374 to compute the stimulation pattern. A feedforward model 1330 can compute a stimulation value based on the desired state block 1356. The stimulation pattern sent to the stimulator can be a summation of the feedback and feedforward portion. This closed-loop controller 1350 can run in real-time to adapt stimulation patterns based on direct muscle state. The scale bar of the bottom left inset indicates a length of 1 mm.

[0167] FIG. 14 is an x-ray radiograph illustrating the system described hereinabove with respect to FIG. 13, the system implanted in a large-animal model, e.g., a pig, at week 4 postimplantation. The radiograph shows a wireless implantable stimulator with connectors (e.g., the implantable stimulator 1371), nerve cuffs (e.g., the nerve cuff 1312) on a peroneal nerve for ankle dorsiflexion and on a tibial nerve for knee flexion, and magnetic beads (e.g., the magnetic bead 1366) in ankle dorsiflexor and knee plantarflexor muscles.

[0168] FIG. 15 illustrates schematically an example embodiment of a system 1500 for transforming a phenotype of a muscle of an animal, wherein the system is configured for open-loop characterization and closed-loop control of the muscle. The system 1500 can further be used for control and actuation of the muscle and can comprise sensors such as a knee magnetometer sensing array 1504-1, which can be placed on a knee flexor muscle, and an ankle magnetometer sensing array 1504-2, which can be placed above an ankle dorsiflexor muscle. The sensing arrays 1504-1, 1504-2 can be configured to detect signals associated with the muscle via implanted magnetic beads. The system 1500 can further comprise a processor 1570, a transceiver 1564, and an implanted stimulator 1571. The implanted- 33 - 5040745. vlDocketNo. 4889-1034001stimulator 1571 can be configured to stimulate the muscle based on a signal received from the processer 1508 via the transceiver 1564. The processor 1570, the transceiver 1564, and the implanted stimulator 1571 can constitute a signal generator configured to stimulate the muscle

[0169] Stimulus patterns can be determined by computers 1508-1, 1508-2, which can comprise an implant computer 1508-1, and transmitted to the processor 1570, e.g., over WiFi communication, and then from the transceiver 1564 to the implanted stimulator 1571 to stimulate the muscle via a corresponding peripheral nerve. Signals associated with muscle dynamics, e.g., MM signals, can be recorded by the sensors 1504-1, 1504-2 and transmitted to a sensing computer 1508-2 (e.g., the MM computer), for example, over WiFi, for the MM computer to compute muscle length and record data. The implant computer 1508-1 and the sensing computer 1508-2 can operate cooperatively to compute neuromuscular dynamics associated with the muscle and to determine a stimulus pattern based on the neuromuscular dynamics computed.

[0170] According to some embodiments, the closed-loop neuromechanical control architecture can comprise sending neuromuscular dynamics, e.g., muscle length dynamics, from the sensing computer 1508-2 to the implant computer 1508-1, e.g., over ethernet or over Wi-Fi. The implant computer 1508-1 can be configured to operate a closed-loop controller and determine stimulation patterns based on real-time muscle state (which may be representative of a current muscle phenotype or state) and reference trajectories (which may be representative of a desired or target phenotype or state). According to some embodiments, the open-loop and closed-loop setups can be configured to utilize fully wireless communication, wired communication, or any combination thereof.

[0171] FIG. 16A is a plot 1628a illustrating example muscle twitch dynamics over time, according to an example embodiment. The muscle twitch dynamics may be detected from a muscle using a system similar to the one described in FIG. 15 and may be responsive to 1 Hz stimulation pulses. A trace 1629a illustrates a zoomed-in view of a twitch contraction of the dotted box. The trace 1629a may show a fast neuromechanical profile of a single muscle contraction, which can include a rapid rise and exponential decay that can be precisely resolved at millisecond time scales. According to some embodiments, MM can allow estimation of contraction dynamics such as rise time, peak displacement, and relaxation time for neuromechanical characterization.- 34 - 5040745. vlDocket No. 4889-1034001

[0172] FIG. 16B is a plot 1628b illustrating example time-based muscle dynamics following an increasing amplitude sweep stimulation protocol, according to an example embodiment. The muscle dynamics may be detected from a muscle using the system of FIG.15 responsive to the amplitude sweep, which can comprise sweeping from 2 to 3.5 A at 70 Hz. An insert may illustrate a zoomed-in view of a tetanic contraction as measured by MM.

[0173] FIG. 16C is a plot 1628c illustrating an example neuromechanical recruitment curve following the amplitude sweep stimulation protocol of FIG. 16B. The recruitment curve of the plot 1628c may show modulation of muscle state with increasing stimulation amplitude and may be used to build a feedforward model for a closed-loop controller.

[0174] FIG. 16D is a plot 1628d illustrating example time-based muscle dynamics following an increasing frequency sweep simulation protocol, according to an example embodiment. The sweep stimulation protocol comprises sweeping from 1 to 100 Hz. Insets 1629b-l, 1629b-2, 1629b-3 illustrate the muscle dynamics (length) responsive to stimulations of frequency 1 Hz (1629b-l), 10 Hz (1629b-2), and 70 Hz (1629b-3). The insets for 1 Hz and 10 Hz illustrate 1 peak and 10 peaks, respectively, for frequencies below tetanic contraction frequencies. The inset for 70 Hz stimulation exhibits a sustained muscle length at tetanic contraction frequencies.

[0175] FIG. 16E is a plot 1628e illustrating a neuromechanical firing rate curve following the frequency sweep stimulation protocol of FIG. 16D. The firing rate curve may show modulation of muscle state with increasing stimulation frequency.

[0176] FIG. 17A illustrates an example closed-loop neuromechanical control architecture that may be used by an example embodiment of a system 1700 for transforming a phenotype of a muscle 1702. The system 1700 can also be used for controlling and actuating the muscle 1702. The system 1700 may be similar to the system of FIG. 15 but applied toward a closed-loop free-space control of ankle dorsiflexion. A stimulation block 1706, e.g., a signal generator, can be configured to deliver a stimulation, e.g., a stimulation pattern, to a muscle 1702 (tibialis anterior (TA) muscle) via a nerve cuff 1714 disposed at a site on a peroneal nerve 1712 between a CNS and the muscle 1702. The stimulation can cause a muscle state change measurable by a sensor 1704 (a magnetomicrometry system). A controller 1708 can be configured to update a stimulation based on muscle states 1710, which can include an actual state (which may also be referred to as a current muscle state) and a desired muscle state.- 35 - 5040745. vlDocketNo. 4889-1034001

[0177] FIG. 17B is a block diagram of an example closed-loop control architecture 1750 that can be implemented by the controller 1708 of FIG. 17A. The controller can comprise a reference trajectory block 1756 informed by neuromechanical parameters such as muscle length frequency and amplitude from gait characterization for non-limiting examples (which may be acquired from prior measurements or experiments). An error between a muscle state 1710 measured using sensors (e.g., the MM sensors 1704 of FIG. 17A) and a desired muscle state from the reference trajectory block 1756 can be fed into a controller 1709 (e.g., a proportional-integral (PI) controller), which can be configured to determine a stimulation value. A feedforward model 1730, which can be derived from, for example, a recruitment curve based on open-loop characterization, can compute a stimulation value that can be summed to the stimulation computed by the feedback controller 1709. This stimulation can then be applied to a peripheral nerve, e.g., the peroneal nerve 1712 of FIG. 17A, via a stimulator 1706.

[0178] FIG. 17C illustrates example movement of an ankle joint caused using the system of FIG. 17A. The movement, as indicated by the arrows in the top and bottom panels for OFF and ON states, respectively, can comprise dorsiflexion control caused by, e.g., actuated by, a closed-loop neuromechanical control system.

[0179] FIG. 18A-C are plots 1828a, 1828b, 1828c illustrating closed-loop neuromechanical control of an ankle joint following a square reference trajectory of varying amplitudes, according to an example embodiment. The neuromechanical control may be effected by the system 1700 of FIG. 17A, which may be configured to apply a stimulation pattern to cause a contraction of said amplitudes in the muscle.

[0180] The plot 1828a illustrates closed-loop neuromechanical control of a muscle 1802a that may interface with an ankle joint following a square reference trajectory of amplitude 1 mm of the muscle, as indicated by the reference trajectory trace. A zoomed in panel may illustrate muscle-length control with sub-millimeter accuracy. A shaded region 1629a indicates muscle shortening as distance between the magnets decreases. Lines indicating the reference trajectory and the muscle length are indicated by the legend.

[0181] The plot 1828b illustrates closed-loop neuromechanical control of a muscle 1802b that may interface an ankle joint following a square reference trajectory of amplitude 2 mm of the muscle, as indicated by the reference trajectory. The reference trajectory and the muscle length are denoted by the legend of FIG. 18 A. A panel (right) illustrates a zoomed in- 36 - 5040745. vlDocketNo. 4889-1034001view of the dotted box region 1829b of the plot 1828b and may indicate sub-millimeter accuracy of actuation of the muscle.

[0182] The plot 1828c illustrates closed-loop neuromechanical control of a muscle 1802c that may interface an ankle joint following a square reference trajectory of amplitude 3 mm. The reference trajectory and the muscle length are denoted by the legend of FIG. 18 A. A panel (right) illustrates a zoomed in view of the dotted box region 1829c and may indicate sub-millimeter accuracy of actuation of the muscle.

[0183] FIGS. 19A-19D illustrate neuromechanical control of a knee joint, according to example embodiments. The neuromechanical control can be effected by a system similar to the system 100 of FIG. 1, wherein the nerve cuff 1712 is placed on a sciatic nerve. The system can implement a closed-loop stimulation architecture similar to the closed-loop control architecture 1750 of FIG. 17B.

[0184] FIG. 19A illustrates example movement of a knee joint, e.g., knee flexion. The movement is indicated by the arrows in the top and bottom panels for OFF and ON states, respectively.

[0185] FIG. 19B is a plot 1928a illustrating closed-loop neuromechanical control of a muscle interfacing a knee joint following a sine reference trajectory with amplitude 4 mm. A panel (right) shows a zoomed in view of a region of the plot indicated by a dotted box region 1929 and may show muscle-length control with sub-millimeter accuracy and a slight temporal delay, which may be due to a wireless update stimulation bandwidth.

[0186] FIGS. 19C and 19D are plots 1928b, 1928c, respectively, illustrating neuromechanical control of a muscle interfacing a knee joint comprising increasing and decreasing continuous mechanical amplitudes. The plot 1928b illustrates neuromechanical the muscle responsive to a stair square reference trajectory with amplitudes of 1 to 3 mm and 3 to 1 mm. The plot 1928c. illustrates neuromechanical control of the muscle responsive to a stair sine reference trajectory with amplitudes of 2 to 5 mm and 2 to 5 mm. The legend of the plot 1928a of FIG. 19B similarly applies to the plots 1928b, 1928c.

[0187] According to some embodiments, models that can be implemented by closed loop controller for controlling actuation of a muscle or for transforming a phenotype of the muscle may rely on reference trajectories or desired states. The controllers can track a current trajectory or state against the reference trajectories or states to inform control. According to some embodiments, such reference trajectories can be established by collecting data of muscle neuromechanics during an activity, for non-limiting example, walking. An animal can- 37 - 5040745. vlDocket No. 4889-1034001be instrumented calibrated magnetometer sensing arrays, which can comprise, for example, the magnetometers described herein with reference to FIG. 13, along with a vest to hold the processor, a belt to hold the transducer, and markers for motion tracking. A computer collecting magnetomicrometry data can be synchronized digitally with the motion tracking system and data can be collected as the animal walked at different speeds from 1.0 to 1.6 m / s, generating canonical gait patterns. Data collection can proceed for one or more trials, and can comprise a given time interval, for example, 30 seconds, at each speed. In some embodiments, data collection can be fully untethered, the animals freely moving within the treadmill and on a walkway where the treadmill was placed. Continuous recordings may demonstrate stable measurements of muscle mechanics during freely moving locomotion. According to some embodiments, after each muscle contraction cycle, muscle length may return to a baseline. With respect to an example embodiment of the swine model described herein with reference to FIGS. 13-19D, muscle mechanics from the tibialis anterior at 0.8 m / s and 1.0 m / s may show cyclic muscle contraction during ankle dorsiflexion. Similarly, muscle mechanics from the biceps femoris may show cyclic shortening-lengthening patterns during knee flexion. These results may demonstrate that the sensing component of the platform reliably can capture physiologically meaningful muscle neuromechanics throughout the gait cycle. Parameters such as frequency and amplitude of muscle mechanics can be used to inform reference trajectories for closed-loop control.

[0188] Example embodiments of closed-loop control of a muscle can also be initially tested through open-loop characterization, for example, using the system 1500 of FIG. 15. In some embodiments, for open-loop characterization, stimulation patterns developed through a custom-made GUI on the implant computer can be utilized. The commands can be sent wirelessly to the processor. For ankle contraction, a nerve cuff on a peroneal nerve can be used, whereas for knee contraction the cuff on the sciatic nerve branch was used. Muscle mechanics can be recorded with the MM system, magnetic field data is sent wirelessly to the MM computer. According to some embodiments, closed-loop neuromechanical control, muscle dynamics are sent over ethernet from the MM computer to the implant computer. The implant computer is running the closed-loop controller and updating the stimulation patterns based on real-time muscle state and reference trajectories.

[0189] According to some embodiments, open-loop stimulation protocols to characterize neuromuscular mechanics across the neural activation range, for example the neuromuscular mechanics of FIGS. 16A-E. Stimulation at 1 Hz may reliably generate twitch- 38 - 5040745. vlDocketNo. 4889-1034001neuromechanical responses with precisely resolved mechanics including a rapid rise and a fast decay (FIG. 16A). Twitch characterization can be used to estimate contraction dynamics such as rise time, peak displacement, and relaxation time. Under an amplitude sweep protocol, increasing amplitude stimulation can generate increasing muscle length, with barely noticeable displacements above a stimulation threshold and saturated responses after the maximal recruitment stimulation (FIG. 16B). The resolved measurements from MM may show characteristic tetanic contraction neuromechanics. A resulting curve, which may be referenced as a neuromechanical recruitment curve (FIG. 16C), can be used to estimate the stimulation range at which the muscle responds, as well as to inform a feedforward model for the closed-loop controller. Under a frequency sweep protocol, increasing stimulation frequency can also generate increasing muscle length. The resolution of MM recordings can allow capturing individual displacements after each stimulation pulse (FIG. 16D). These measurements may demonstrate that neuromechanical firing rates can be captured (FIG. 16E) As described hereinabove, data such as twitch, recruitment, and firing rates, which may be indicative of muscle state, can be obtained and used, for example, as reference trajectories or desired states, for closed-loop control of a muscle.

[0190] Materials and Methods

[0191] An example embodiment of a system and data collection protocol is described hereinbelow. It should be understood that the system and data collection described are provided for purposes of exemplification and not intended to be limiting with respect to the system, reference data, or closed-loop operation of the system. The system and protocols described may be in relation to the system and protocols of FIGS. 13-19D. The system can comprise a magnetomicrometry system with minor adaptations for a Yucatan minipig model configured for muscle tracking using two boards at the same time. The system can comprise a custom magnetic field sensing array and a separate control board. The sensing array can comprise 96 three-axis magnetic field sensors arranged, for non-limiting example, in an 8x12 grid with 5.08 mm spacing. Each sensor can be supplied with nonmagnetic capacitors, and seven digital multiplexers, which may enable time-domain multiplexing so that all sensors could be addressed by a single microcontroller. The sensing array can be connected via a custom adapter board to a microcontroller module, which can be wireless and battery powered. The sensor board and control board can weigh 24 g and 12 g, respectively, and can be housed in a custom 3D-printed enclosure with a hook-and-loop-based strap for attachment to the limb. The microcontroller can be configured to sense magnetic field signals from all- 39 - 5040745. vlDocket No. 4889-1034001sensors at 155 Hz and streamed the data wirelessly to a dedicated magnet tracking computer. A real-time tracking algorithm can be run on the tracking computer to estimate a three-dimensional position of each implanted bead and compute a differential distance between the beads.

[0192] According to the example embodiment, with respect to a stimulation subsystem, a fully implantable 32-channel pulse generator can be used for current-controlled peripheral nerve stimulation. In some embodiments, the implantable pulse generator (IPG) can be powered inductively by a skin-mounted external transceiver that also enabled bidirectional RF communication and magnetic self-alignment. The IPG can be placed in a subdermal pocket near the hip and connected to epineural nerve cuff electrodes implanted on the target motor nerve branches. Three circumferential nerve cuffs can be used: one on a sciatic nerve branch innervating the biceps femoris and two redundant cuffs on a peroneal nerve innervating the tibialis anterior. The cuffs can include integrated sutures for mechanical fixation around the nerve, which can be useful for stable chronic positioning. The IPG can comprise a titanium housing, which can also serve as a system ground. Stimulation waveforms can comprise charge-balanced, biphasic pulses delivered through selected cuff contacts. For open-loop characterization, stimulation paradigms (1 Hz twitch stimulation, amplitude sweeps, and frequency sweeps) can be programmed from a custom graphical user interface (GUI) running on an implant computer. In some embodiments, commands can be transmitted wirelessly to the processor and relayed through the external transceiver to the IPG. According to some embodiments, amplitude sweeps can be delivered as 70 Hz trains over the neuromechanical recruitment range (typically 2-3.5 mA), while frequency sweeps can range from 1 to 100 Hz at an amplitude that can generate 50% of maximal muscle response to characterize firing-rate-dependent neuromechanical responses. All stimulation parameters can be selected within hardware limits of the IPG, which can support, according to the example embodiment, current amplitudes from 0.5 pA to 15 mA, pulse widths from 1 ps to DC, and stimulation frequencies up to 30 kHz.

[0193] According to an example embodiment, data collection protocols, e.g., gait data collection, using a system for control or phenotype transformation of a muscle can comprise one or more gait recording sessions on a motorized treadmill under untethered conditions. Sessions may be scheduled periodically throughout an implantation period and each included trial can comprise a range of walking speeds. In some embodiments, each session can comprise of several continuous trials at treadmill speeds between 1.0 and 1.6 m s '. Trials can- 40 - 5040745. vlDocket No. 4889-1034001include, for example, a minimum of 30 s in duration at each speed, and only periods of uninterrupted, steady forward locomotion may be retained for analysis. The calibrated magnetomicrometry system can be synchronized with a motion capture system via a microcontroller, during which magnetometer signals can be streamed to the base station laptop. Triggers, e.g., a digital transistor-transistor logic (TTL) trigger, can be used to ensure temporal alignment of magnetomicrometry and kinematic data. Kinematic recordings can be acquired using a multi-camera (e.g., an eight-camera system) mounted around a treadmill enclosure, and reflective markers can be positioned on the hind limb according to a standard anatomical marker set for swine gait. Speed blocks can be presented in random order, which may be helpful for minimizing adaptation and ordering effects. Magnetomicrometry and motion capture data streams can be logged continuously and stored for offline analysis.

[0194] According to an example embodiment, open-loop characterization of an animal model can be used to characterize neuromuscular mechanics across a full activation range, which may be useful for inferring feedforward control models. Stimulation patterns, which can be performed on an animal under light sedation, can be programmed through a custom graphical user interface (GUI) running on the implant computer and transmitted wirelessly to the implantable pulse generator (IPG). For ankle dorsiflexion, stimulation can be delivered through a nerve cuff mounted on the peroneal nerve, whereas knee flexion stimulation can be delivered through the cuff on the sciatic nerve branch. Muscle mechanics can be recorded simultaneously using the MM system, with magnetic field data streamed wirelessly to the MM computer for real-time muscle length computation.

[0195] According to an example embodiment, three open-loop stimulation paradigms were used (FIGS. 16A-16E). First, twitch characterization can be performed using 1 Hz stimulation pulses to resolve single-contraction neuromechanical dynamics, including rise time, peak displacement, and relaxation time. Second, neuromechanical recruitment curves can be obtained using amplitude sweep protocols consisting of 70 Hz stimulation trains with increasing current amplitudes spanning an activation threshold to saturation (typically 2-3.5 mA). These sweeps can generate graded increases in muscle shortening from barely detectable responses above threshold to saturated tetanic contractions. Third, neuromechanical firing-rate curves can be obtained using frequency sweep protocols in which stimulation frequency was increased from 1 to 100 Hz at fixed amplitude. At low frequencies, individual muscle twitches may be resolved following each pulse, whereas at higher frequencies, sustained tetanic contractions may be observed. This recruitment- 41 - 5040745. vlDocket No. 4889-1034001relationship and firing rate frequency to achieve sustained tetanic contraction can be used to construct an inverse neuromechanical model for the feedforward component of the closed-loop controller.

[0196] According to some embodiments, closed-loop neuromechanical control can comprise real-time muscle-state estimates from the magnetomicrometry system and streamed to an implant computer (e.g., as illustrated in FIG. 15). A combined feedforward-feedback controller can compute stimulation updates at 50 Hz, which can be transmitted wirelessly through an external transceiver to an IPG to modulate nerve-cuff stimulation in real time during ankle and knee control experiments (FIG. 15). In some embodiments, the controller architecture can a proportional-integral (PI) feedback controller, which can be calibrated once and held constant across trials, in parallel with a feedforward model derived from an inverse of a neuromechanical recruitment curve. The desired reference trajectory can be defined from physiologically measured gait neuromechanical parameters, including muscle shortening amplitude and frequency obtained during freely walking locomotion. Closed-loop performance cab be evaluated using both square and sine reference trajectories (FIGS.18 A-19D). Square waveforms can be used to test the ability of the platform to hold constant muscle states, whereas sine waveforms can be used to emulate cyclic gait-like muscle dynamics. The controller can track square reference trajectories with amplitudes of 1 (average error 0.2135 mm), 2 (average error 0.0125 mm), and 3 (average error 0.0152 mm) mm, and sine trajectories with amplitudes up to 4 (average error 0.1517 mm) or 5 mm, while also following increasing and decreasing stair-step amplitudes to assess neuromechanical modulation. Across all conditions, the platform can achieve sustained closed-loop control with sub-millimeter tracking accuracy over several minutes of continuous operation.Myoneural actuator (MNA)MNA design rationale

[0197] According to some embodiments, a myoneural actuator (MNA) can repurpose biological skeletal muscle as an actuator for control by a computer-based system. In some embodiments, design of an MNA can involve denervating a native motor nerve of a base muscle to eliminate voluntary control of the muscle by the CNS of a human or animal. Such denervation can typically lead to a loss of contractility. To preserve muscle contractility, a base muscle of an MNA can be reinnervated, for example, using a purely sensory nerve

[0012] , Example embodiments described hereinbelow may be directed to an MNA implemented in a- 42 - 5040745. vlDocket No. 4889-1034001rodent motel and may utilize a lateral gastrocnemius (LG) muscle as an actuator base and a sural nerve (SN) for reinnervation using a sensory nerve. The embodiments described are not intended to be limiting and it should be apparent to one of skill in the art that an MNA that comprise different combinations of muscles and nerves.

[0198] FIG. 20 illustrates schematically example embodiments of myoneural actuators and applications thereof. According to some embodiments, the capability to control organ actuation can enable modulation of human or animal biological functions.

[0199] For instance, actuation of muscle-tendon stretch via a serially coupled MNA can facilitate modulation of neural afferents responsible for limb perception, thereby providing proprioceptive feedback for bionic or virtual limbs (loops 1-4). Such embodiments may have previously been described. According to some embodiments, MNAs may possess the potential to replicate organ mechanics, such as intestinal contractions, which can be driven by neural or virtual cues. Such replication of biological organ mechanics could assist in restoring impaired organ function while offering physiological feedback, such as appetite regulation (loops i-iv). Example embodiments of biohybrid organ systems, which are further described hereinbelow, can comprise neural cues or virtual triggers 2001, which can cause actuation 2003 of a target organ of the biohybrid organ systems. The actuation 2003 may comprise organ function emulation 2005 using MNA control. The organ function emulation 2007 may further produce physiological feedback, for example, to a CNS of a human or animal.

[0200] FIG. 21 illustrates schematically an example embodiment of a myoneural actuator 2162. The myoneural actuator 2162 can be configured to be fatigue resistant and can comprise a reversible nerve block 2156, which may be helpful for neural isolation by preventing unwanted or unintended signaling along an afferent or efferent direction. As described hereinabove, the myoneural actuator 2162 can comprise a muscle 2102, wherein a native motor nerve (not shown) of the muscle has been transected (i.e., the muscle has been denervated). The muscle can further be reinnervated using a nerve, for example, a pure sensory nerve 2112. The nerve 2112 can be electrically coupled to a signal generator configured to stimulate the nerve at one or more locations 2114, 2115. As such, control of the myoneural actuator 2162 (which can comprise the nerve 2112 and the muscle 2102) may be redirected from the nervous system to artificial computer control. The signal generator can be configured to stimulate the nerve at a first location 2114 to stimulate the nerve 2112, which may cause bidirectional stimulation propagation, and at a second location 2115, which may be proximal to the CNS, to deliver a nerve block 2156 to ensure neural isolation from the- 43 - 5040745. vlDocket No. 4889-1034001nervous system during operation of the myoneural actuator. This myoneural manipulation can enhance a motor recruitment interface, which can result in a fatigue-resistant actuator with improved performance in closed-loop control systems, as further described hereinbelow. Additionally, according to some embodiments, reinnervation of the muscle by a sensory nerve can result in a restoration of neuromuscular junctions (NMJ) 2176 with the sensory nerve 2112, which is also further described hereinbelow.

[0201] FIG. 22 illustrates schematically another example embodiment of a myoneural actuator 2262 implemented in a rodent model 2201. The MNA 2262 can comprise a LG muscle 2202 of the rodent 2201 with a transected native tibial nerve (TN) 2213, which is a motor nerve. The LG muscle 2202 of the MNA 2262 can be reinnervated by a sural nerve 2212 and the SN may form neuromuscular junctions 2274 with the muscle 2202. The myoneural actuator 2262 can further comprise an external actuation control system 2200 (e.g., a system for transforming a phenotype of a muscle or of control and actuation of the muscle, coupled to the SN 2212 and configured to provide external actuation control to the MNA.

[0202] According to some embodiments, motor nerves, such as the tibial nerve (TN), can possess variable axon diameter sizes for innervating different muscle fiber populations [4], This non-uniform axonal composition can lead to preferential recruitment of large-diameter fibers during extraneural electrical stimulation, which can result in unnatural motor unit activation or accelerated muscle fatigue [2,4], In contrast, pure sensory nerves can be characterized by a more uniform axonal composition. To confirm this, histological analyses can be performed for the TN, which innervates the lateral gastrocnemius (LG) muscle, and the sural nerve (SN), which can be used, for non-limiting example, for MNA reinnervation.

[0203] FIG. 23 illustrates histological cross-sections of a native motor nerve (left) and a pure sensory nerve (right), according to an example embodiment. The scalebars indicate distances of 50 pm.

[0204] FIG. 24 is a plot 2428 illustrating cumulative probability densities of axon crosssection area for native motor nerves and pure sensory nerves, according to an example embodiment. The plot 2428 may be generated using histological analysis of motor and sensory nerves, for example, individual neurons of the histological images of the native motor nerve and the pure sensory nerve of FIG. 23.

[0205] FIG. 25 is a plot illustrating mean axon cross-sectional area and axon crosssection area variability for native motor nerves and pure sensory nerves, according to an- 44 - 5040745. vlDocket No. 4889-1034001example embodiment. The plot 2528 may comprise a cohort size if n = 6. For the mean axon cross-sectional area 2529a, a - value of 0.0022 (corresponding to the ** of the plot) was computed using a two-sided Mann-Whitney U test. For the axon cross-sectional variability 2529b, a -value of 0.017 (corresponding to the * of the plot) was computed using an unpaired two-sided / -test.

[0206] As indicated by FIGS. 23-25, the SN, being a pure sensory nerve, may display smaller and more uniform axon cross-sectional areas compared to the TN, a motor nerve. Based on these findings, it may be hypothesized that an MNA reinnervated with a pure sensory nerve could exhibit impartial fiber recruitment during extraneural electrical stimulation, which may allow for fatigue-resistant actuation while maintaining selfsustainability.

[0207] According to some embodiments, nerve-muscle histological analyses can be conducted to examine myoneural interactions resulting from reinnervation, which may be helpful for evaluating the aforementioned hypothesis.

[0208] FIG. 26 illustrates fluorescence imaging of an example embodiment of a myoneural actuator comprising a muscle reinnervated by a sensory nerve. Indicators for neurofilament and synaptophysin may be indicative of the sensory nerve and a-bungarotoxin (a-BTX) may be indicative of nicotinic acetylcholine receptors (nAChRs), which may be evidence of formation of neuromuscular junctions. FIG. 26 may illustrate that reinnervation with a sensory nerve can successfully restore neuromuscular junctions, for example, based on an overlay of the staining of the sensory nerve (e.g., neurofilament and synaptophysin) and the neuromuscular junction (e.g., a-BTX). The scale bar represents a distance of 50 pm.

[0209] According to some embodiments, restoration of NMJs can allow modulation of an MNA through natural neural pathways via nerve stimulation, which may replicate functional capabilities of a motor nerve while employing a pure sensory nerve for artificial computer control.

[0210] FIG. 27 illustrates histological images of muscle cross-sections before and after myoneural manipulation, according to an example embodiment. That is, the histological images appear to confirm a presence of healthy muscle fibers in a muscle of an MNA, with no or minimal observable morphological differences or significant muscle fiber type transformations, after denervation and reinnervation of the muscle using a sensory nerve.

[0211] FIG. 28A illustrates histological images of a fluorescent cross-sectional slide of a myoneural actuator, according to an example embodiment. The fluorescent cross-sectional- 45 - 5040745. vlDocket No. 4889-1034001slide can be processed to form a grayscale and a segmented image. The slide can be stained with anti-myosin antibodies to determine slow and fast muscle fiber percentages. According to some embodiments, each fluorescent spot may represent a muscle fiber. From the segmented image, muscle fibers may be counted to obtain a quantification of a total number of fibers corresponding to a particular anti-myosin antibody. In some embodiments, his quantification may be made for slow type and fast type muscle fibers using 2 different antibodies. In some embodiments, sing a total number of muscle fibers in a slide, which can be quantified using staining, e.g. using 4',6-diamidino-2-phenylindole (DAPI) staining, a percentage of each fiber type can be quantified.

[0212] FIG. 28B is a plot 2828 illustrating percent composition of slow and fast fibers in native muscle and myoneural actuators based on histological images similar to those of FIG.28A. Fiber composition (e.g., percentages of slow (left) and fast (right) twitch fibers) may be similar between native skeletal muscle and MNAs.MNA sustainability

[0213] According to some embodiments, a design of an MNA can eliminate efferent neural signaling from the central nervous system, which may result in muscle inactivity when the MNA is not artificially operated. Prolonged inactivity could potentially lead to reduced muscle mass and contractility, which may require periodic external interventions, such as electrical stimulation, to maintain functionality. According to an example embodiment, longterm sustainability of an MNA without such interventions can be assessed based on mass and maximum isometric force of three separate MNA cohorts measured at 9, 12, and 15 weeks after MNA construction. No external interventions were applied prior to assessment.

[0214] FIG. 29 is a plot 2928 illustrating mass (left) 2929a and maximum myoneural actuator force (right) 2929b of a muscle of a myoneural actuator without stimulation after surgical intervention, according to an example embodiment. Terminal testing was performed based on the cohorts of 9, 12, and 15 weeks described hereinabove. The plot indicates that MNA mass, computed as a percentage of mass of a corresponding animal, may be consistent at 9, 12, and 15 weeks post-surgical intervention for creation of the MNAs, wherein NS denotes not significant. The plot further indicates that maximum force applied by the MNA may also be relatively consistent at the same time intervals. Statistical analyses may comprise one-way analysis of variance (ANOVA) tests.- 46 - 5040745. vlDocket No. 4889-1034001

[0215] FIGS. 30A and 30B are plots 3028a, 3028b illustrating mass (3028a) and maximum force (3028b), respectively, of native muscle and myoneural actuators, according to example embodiments. The plot 3028a may indicate an absolute loss of muscle mass, which may be due to the reinnervation process. According to some embodiments, no further atrophy may be observed once reinnervation was completed. Data of the plot may be derived from a cohort of native (n = 8) and MNA (n = 13) muscles and statistical analysis may comprise a two-sided Mann-Whitney U test (P = 2.5* 10'4).

[0216] The plot 3028b may indicate that no significant differences may be observable in isometric force applied by the MNA native muscles of equivalent mass. Data for the plot may be derived from the cohort of FIG. 30A and statistical analyses may comprise an unpaired two-sided / -test. NS represents not significant. Such findings may suggest a potential for long-term enhanced recruitment of muscle fibers under computer control in a self-sustaining muscle actuator. The plots 3028a, 3028b may share a common legend of the plot 3028b.Fatigue-resistant properties of the MNA

[0217] According to some embodiments, histological analysis of MNAs may reveal a more uniform axonal distribution in a reinnervating nerve compared to that of native muscle. This difference in axonal biophysics may be hypothesized to enhance muscle fatigue resistance during repetitive and continuous actuation. Such a hypothesis may be evaluated based on fatigue responses of an MNA and a native LG muscle. According to some embodiments, fatigue responses of the MNA and native muscle may be evaluated and compared under two extreme conditions: sequential twitch and continuous force production.

[0218] FIG. 31 is a plot 3128 illustrating individual twitch force responses for a native muscle and a myoneural actuator, according to an example embodiment. Sequential twitch responses can be assessed over 450 cycles with 1 -second resting intervals between singlepulse stimulation protocols. The plot illustrates, for a given mouse (rat 1 - Rl), that signal traces of the native muscle 3129a may indicate significantly reduced twitch force when compared to signal traces of the MNA 3129b.

[0219] FIG. 32 is a plot 3228 illustrating sequential twitch force responses for native muscles and myoneural actuators, according to an example embodiment. The plot 3228 can comprise data from the plot 3128 of FIG. 31, which may be indicated as Rl, and can further comprise data from two other rats. The plot 3228 further illustrates normalized twitch force- 47 - 5040745. vlDocket No. 4889-1034001with respect to number of cycles and may indicate reduced twitch force in the native muscles when compared to the MNAs at increased cycle counts.

[0220] Restated, the plots 3128, 3228 of FIGS. 31 and 32 may indicate that the MNA cohort exhibited superior force preservation during sequential single-pulse actuation compared to the native cohort. Additionally, the MNA cohort may achieve an earlier balance between fatigue and recovery during single-pulse stimulation protocols, while the native cohort continued to experience progressive fatigue.

[0221] FIG. 33 is a plot 3328 illustrating twitch force loss per cycle based on the sequential twitch force responses of FIG. 32. Statistical analysis may comprise a paired two-tailed / -test and indicate that the MNA may be associated with a significantly lower rate of twitch force loss (P = 0.0061).

[0222] FIG. 34 is a plot 3428 illustrating cumulative probability densities of the twitch forces of native muscles and myoneural actuators of FIG. 32. The cumulative probability densities may be computed for the native muscle and the MNA of each rat (R1-R3) and may be normalized based on a maximum twitch force.

[0223] FIG. 35 is a plot 3528 illustrating twitch force variability of the twitch forces of native muscles and myoneural actuators of FIG. 32. Statistical analysis may comprise a paired two-tailed / -test and may indicate that MNAs exhibit significantly reduced twitch force variability (P = 0.030). FIGS. 34 and 35 may indicate that MNAs may display lower rate of twitch force loss and more stable actuation with reduced force variability.

[0224] The plots 3128-3528 of FIGS. 31-35 may together demonstrate that myoneural actuators comprising muscles reinnervated using sensory nerves exhibit different fatigue characteristics than native muscle. In particular, the myoneural actuators may demonstrate decreased rates of fatigue and decreased variability of twitch force over multiple actuation cycles.

[0225] FIGS. 36A and 36B are plots 3628a, 3628b illustrating force response of native muscles and myoneural actuators under continuous stimulation, according to an example embodiment. Force responses of 7 native muscles and 4 MNAs may be measured under constant and continuous stimulation (i.e., without resting periods). The plot 3628a, which illustrates normalized force over time, may indicate decreased rates of fatigue or fatiguability over time of the MNAs with respect to the native muscles. The plot 3628a may further indicate that a normalized force applied by the MNA after 60 seconds of continuous stimulation is greater than that of the native muscle. The plot 3628b illustrates a fatigue time,- 48 - 5040745. vlDocket No. 4889-1034001which may be a time to reach 75% normalized force (e.g., the dotted line in FIG. 36A). The MNA cohort may demonstrate a 260% improvement in fatigue resistance compared to the native cohort (Native: 5.19 ± 1.16 s; MNA: 18.67 ± 2.96 s). Statistical analysis may comprise an unpaired two-tailed / -test (P = 0.00067).

[0226] FIG. 37 is a plot 3728 illustrating fatigue dynamics of native muscles and MNAs based on the force response under continuous stimulation of FIGS. 36A and 36B. The fatigue dynamics may comprise a rate of fatigue, which may be computed from the normalized force. The plot 3728 may reveal fundamentally different fatigue dynamics in the MNA cohort compared to the native cohort. The MNA exhibited additional logarithmic fatigue dynamics, which can be indicative of fatigue resistance, alongside typical exponential fatigue dynamics. Such fatigue-resistant properties may allow MNAs to sustain higher forces more effectively during continuous actuation compared to native muscles. The plots 3628a, 3628b, 3728 may share a common legend, as shown in the plot 3628b.MNA closed-loop control

[0227] According to some embodiments, fatigue-resistant actuation provided by an MNA may enable extended force controllability compared to native muscles, even when utilized within a closed-loop control architecture, which may be designed to compensate for forcetracking errors. A closed-loop force control system can be implemented to evaluate this. The closed-loop force control system can be implemented with a fixed feedback gain for cohorts of native muscles and MNAs, with target force levels set at 30%, 50%, and 70% of maximum force.

[0228] FIG. 38 illustrates an example embodiment of a system 3800 comprising a closed-loop force control architecture 3850 that can be used to actuate a muscle 3802 of a myoneural actuator 3862. The system can be similar to the systems described hereinabove for transforming phenotypes of muscles and may be used for such applications. The system 3800 can comprise a sensor 3804, e.g., a force transducer configured to determine force feedback responsive to stimulations to a muscle 3802, wherein the force feedback can be indicative of a current state of the muscle 3802. A controller 3808 can be configured to implement a closed-loop control architecture 3850 to determine a stimulus pattern for the muscle based on the force feedback and a user-defined force (e.g., a desired force). The controller 3808 can be configured to delivery of the stimulus pattern determined to the muscle 3802 via a signal generator 3806 configured to stimulate the muscle.- 49 - 5040745. vlDocket No. 4889-1034001

[0229] FIG. 39 is a plot 3828 illustrating closed-loop control performance of native muscles (top) 3929a and myoneural actuators (bottom) 3929b across force target values, according to an example embodiment. As described hereinabove, the target levels comprise 30%, 50%, and 70% of a maximum force of the muscle, as indicated by the reference trajectories, wherein the system 3800 of FIG. 38 can be used to deliver stimulations configured to trigger contractions associated with the target levels. Control performance may be assessed over 45 sequential cycles for each target value. As indicated by the plot 3928, the native muscle 3929a may fatigue rapidly and exert forces significantly lower than the corresponding reference trajectories. By contrast, the MNA 3929b may be capable of prolonged contractions at or close to a target force level, particularly for the lower target levels of 30% and 50%.

[0230] FIGS. 40 A and 40B are plots 4028a, 4028b illustrating quantification of the closed-loop control performance of the native muscles and the myoneural actuators of FIG.39 over stimulation cycles. The plot 4028a illustrates normalized force over the stimulation cycles and the plot 4028b illustrates a log-scale ratio of force applied versus target force. The plots 4028a, 4028b of FIG. 40 A and 40B may indicate that the closed-loop controller can, initially, e.g., within the first 15-20 cycles, effectively track the target force values for both native muscles and the MNA. However, as the number of cycles increased, native muscles may rapidly lose force controllability due to fatigue, regardless of the target force levels. In contrast, the MNA may consistently maintain force controllability across all target levels, even as the number of cycles increased. These findings may underscore limitations of native muscles under artificial computer control, where fatigue can significantly impair actuation performance despite closed-loop compensation. By contrast, the fatigue-resistant actuation enabled by MNAs can enhance the fundamental capabilities of muscle-actuated systems, preserving force controllability over extended periods of operation.MNA actuation with isolation from the central nervous system

[0231] According to some embodiments, modulating an MNA through nerve stimulation may activate neural pathways that extend both towards a distal (MNA) and a proximal (CNS) end of a nerve. In some embodiments, this neural activation can lead to unintended sensations, pain, or disruption of essential neural signaling for neuromotor control if it reaches the CNS. Therefore, isolating the proximal signaling pathway during MNA modulation can allow for actuation without triggering adverse neural activation and for- 50 - 5040745. vlDocket No. 4889-1034001expansion of ranges of applicable stimulation parameters and protocols. As described hereinabove, a reversible electrical nerve block can be integrated into MNA technology.

[0232] According to an example embodiment, efficacy of a nerve block may be verified by applying it distally to a nerve stimulation for MNA control or modulation. For example, the nerve block may be configured to be delivered between the nerve stimulator and the muscle. According to some embodiments, the nerve block can comprise an electrical signal, for example, a KHFAC as described hereinabove.

[0233] FIG. 41 illustrates schematically an example embodiment of a system 4100 comprising a myoneural actuator 4162 with a nerve block 4156. The myoneural actuator 4162 can comprise a muscle 4102 reinnervated using a sensory nerve 4112, which may be connected to a CNS 4116 of a human or animal. The myoneural actuator 4162 may comprise an interface 4114 wherein the MNA can receive a stimulus, for example, an electrical stimulus for a signal generator. The MNA can further comprise a nerve block site 4115, wherein the nerve block 4156, e.g., a reversible electrical nerve block, can be delivered for neural isolation. As described hereinabove, the nerve block site 4115 can be positioned distal to the interface 4114 for verification of nerve block efficacy. The muscle 4102 may interface a sensor, e.g., a force transducer (not shown), to detect force generated by the muscle 4102.

[0234] FIG. 42 is a plot 4228 illustrating normalized force applied by a myoneural actuator, similar to the MNA schematically illustrated in FIG. 41, with or without a nerve block. The plot includes traces 4229a, 4229b representative of stimulation potentials at the muscle with (4229a) or without (4229b) a nerve block applied. Subplot 4229c of may further indicate that activation of the nerve block can successfully divert a distal neural pathway, which may be reflected by the substantial changes of normalized force between the nerve block off and on conditions. As such, the nerve block may effectively silence the MNA response to nerve stimulation, confirming that the nerve block could isolate the neural pathway at the point of application.

[0235] FIG. 43 illustrates schematically an example embodiment of a system 4300 comprising myoneural actuator 4362 with a nerve block 4356 and a closed-loop control architecture 4350. Closed-loop control may be implemented in the myoneural actuator 4362 comprising a nerve block 4356 applied proximally (e.g., at a nerve block site 4315between a site of delivery 4314 of a stimulation pattern for actuating the MNA 4362 and the CNS 4316). The system 4300 may be similar to the system 3800 described hereinabove with respect to FIG. 38. The system 4300 can comprise sensors 4304 configured to detect signals- 51 - 5040745. vlDocket No. 4889-1034001from a muscle 4302 of the myoneural actuator 4362 and can compute neuromuscular dynamics such as force feedback. A controller can be configured to implement the closed-loop control architecture 4350 to use the force feedback, which may be indicative of a current state of the muscle, and a target or user-defined force to determine a stimulus pattern to excite the muscle 4302 via a nerve 4312 innervating the muscle (e.g., a sensory nerve). The myoneural actuator 4362 of FIG. 43 can allow for assessment of MNA modulation while ensuring neural isolation from the CNS.

[0236] FIGS. 44 A and 44B are plots 4428a, 4428b illustrating force modulation of the myoneural actuator illustrated schematically in FIG. 43 under nerve block. The plot 4428a illustrates normalized force applied by the myoneural actuator over time for varying target force levels (30%, bottom; 50%, middle; 70%, top) over a course of multiple cycle.Reference square waves are also illustrated in the plot 4428. The plot 4428b illustrates the normalized force applied versus target force across cycles for all target levels (30%: n = 89 cycles, 50%: n = 85 cycles, 70%: n = 33 cycles; large white dots, medians; boxes and whiskers, interquartile ranges and adjacent values).

[0237] The plots 4428a, 4428b may show that the MNA could achieve various target force levels while under nerve block, with a high degree of linearity observed between the target and output force. These findings may further confirm that the MNA technology could provide muscle actuation while being neurally isolated from the CNS during operation.Biohybrid organ systems using the MNA

[0238] According to some embodiments, the ability to control organ actuation can enable modulation of human biology. For instance, the actuation of muscle-tendon stretch through a serially coupled MNA allows modulation of neural afferents responsible for limb perception, thereby providing proprioceptive feedback for bionic or virtual limbs. Such applications, which may be referred to as a proprioceptive mechanoneural interface, have been previously reported.

[0239] According to some embodiments described herein, myoneural actuators may applied toward biological organ systems, for example, to form a biohybrid organ system, to replicate organ mechanics. A non-limiting example of such a biohybrid organ system can comprise a biohybrid intestinal system, wherein intestinal contractions can be driven by neural or virtual cues, to support impaired organ function and offering physiological feedback, such as appetite regulation.- 52 - 5040745. vlDocket No. 4889-1034001

[0240] It should be further noted that various organs support essential functions through mechanical movements. For example, contraction of the thoracic diaphragm enables respiration, while the contractions of the bladder, small intestine, and vascular system facilitate excretion, digestion, and hemodynamic regulation, respectively. Additionally, these organs can directly influence emotions and behaviors, such as stress and feelings of satiation. The ability to control organ mechanics can offer the potential to create technologies that can address organ failure, manage physiological conditions, and even provide physiological feedback in virtual environments, such as the metaverse.

[0241] FIG. 45 illustrates schematically an example embodiment of a biohybrid organ system 4580 for the small intestine. The biohybrid organ system 4580 can comprise an organ system of a human or animal (e.g., the gastrointestinal system 4582). The biohybrid organ system 4580 can further comprise a myoneural actuator 4562 comprising a transplanted muscle 4502 of a human or animal. The transplanted muscle 4502 of the myoneural actuator 4562 can be in coupled arrangement with a portion of the organ system, e.g., a small intestine 4583 (or a portion thereof) of the gastrointestinal system. Contraction of the transplanted muscle 4502 responsive to a stimulus, e.g., from a stimulus source 4506 can mechanically actuate the portion 4583 of the organ system to perform a physiological function of the organ system 4582. According to some embodiments the stimulus source can comprise a CNS of the human or animal or from an external control input, and the physiological function of the GI organ system can comprise contraction of the small intestine 4583. In some embodiments, one or more myoneural actuators can be used in combination to actuate the organ system, which can include, for non-limiting example, sequential contraction of the small intestine to replicate peristaltic motion. Example embodiments of implementations of biohybrid organ systems are further described hereinbelow.

[0242] FIGS. 46A-46J illustrate an example embodiment of a method for implementing a myoneural actuator, which may be used as part of a biohybrid organ system. FIG. 46A illustrates components that may be used to create a myoneural actuator in a rodent model, which can comprise a lateral gastrocnemius muscle. The LG muscle can comprise a native tibial nerve and may be later reinnervated using the sural nerve. FIG. 46B illustrates anaesthetization of the animal and access of the aforementioned components via a small incision on a lateral side of the body. FIG. 46C illustrates eliminating native motor control of the LG muscle by transecting the TN. In some embodiments, the TN can be fixed to a surrounding, non-targeted muscle, e.g., a biceps femoris muscle. FIG. 46D illustrates an- 53 - 5040745. vlDocket No. 4889-1034001alternative view of the LG muscle, particularly with respect to a medial gastrocnemius (MG) muscle, and denervation of the LG muscle. Redirecting and suturing a severed branch of the TN to a nearby muscle, e.g., the BF muscle, may be useful for preventing unintended reinnervation by the native motor nerve. FIG. 46E illustrates reinnervation of the LG muscle, which comprises transecting of the sural nerve and insertion of a portion of the transected SN nerve into the LG muscle. FIGS. 46F-46H illustrate a step-by-step process of the reinnervation of FIG. 46E, comprising transecting the SN and tunneling the SN to reposition the nerve (FIG. 46F), forming a muscle pocket in the LG muscle (FIG. 46G), and inserting the SN into the muscle pocket (FIG. 46H). FIG. 461 illustrates an example reinnervated myoneural actuator and a timeline for MNA implementation and characterization. According to some embodiments, for example, those described hereinabove with reference to at least FIGS. 29-3 OB, sustainability of the myoneural actuator may be evaluated by withholding additional intervention such as electrical stimulation to the sural nerve and muscle of the myoneural actuator.

[0243] FIG. 46J illustrates schematically the myoneural actuator implemented by the foregoing process of FIGS. 46A-46I. The myoneural actuator comprises the LG muscle with native signaling from the CNS disrupted due to a transected tibial nerve. The MNA is subsequently reinnervated by a pure sensory nerve, which may still be coupled to the CNS. As described hereinabove, such a design of a myoneural actuator may introduce a nerve to sustain the MNA while eliminating motor control from the CNS and may enable neural manipulation to customize a base composition of the MNA. The MNA can subsequently be coupled to a system for characterization or closed loop control, or be incorporated into a biohybrid organ system.

[0244] FIGS. 47A-47H illustrate example embodiments of methods for characterization and control of a myoneural actuator. The characterization and control may be performed subsequent to the implementation of FIGS. 46A-46J and may be similar to those described hereinabove with reference to the system for transforming a phenotype of a muscle. FIG. 47A illustrates separation of the MNA from surrounding tissue, which can comprise transecting a tendon connected to the LG muscle at a distal end tunneling of a reinnervated sural nerve. FIG. 47B illustrates an alternative view of the separating the MNA from a soleus muscle and a medial gastrocnemius muscle. The TN, which has been separated from the LG muscle, is also shown. FIG. 47C illustrates an example apparatus for characterization and control of the MNA. The apparatus can comprise a heated bed, a muscle tensioner, and a force transducer (a- 54 - 5040745. vlDocket No. 4889-1034001sensor). The muscle tensioner can further comprise a screw and spring and can be configured to position the bed to pre-tension the MNA. The apparatus can also comprise a subsystem for anaesthetizing the animal, e.g., a rat. FIG. 47D illustrates an example setup of a myoneural actuator in the apparatus of FIG. 47C. A patellar tendon and anterior talofibular ligament of the myoneural actuator can be mechanically anchored to the heated bed and a distal end of the MNA construct can be attached to the force transducer.

[0245] FIG. 47E illustrates schematically an example system 4700a for MNA characterization. The system 4700a can be similar to the system 3800 of FIG. 38 and can comprise a signal generator 4706a electrically coupled to a nerve 4712a via a hook electrode 4714a and configured to deliver stimulations to the nerve 4712a via the electrode 4714a. MNA force outputs responsive to the stimulations can be recorded by the force transducer 4704a and transmitted to a processor 4708a for processing. The signal generator 4706a, the force transducer 4704a, and the processor 4708a (e.g., a computer-based system) can be communicatively coupled to transmit information and signals for data collection.

[0246] FIG. 47F illustrates schematically an example system 4700b for closed-loop control of the myoneural actuator 4762b. The system 4700b can comprise the signal generator 4706b, the hook electrode 4714b, the MNA 4762b, and the processor 4708b, and may be similar to the system 4700a of FIG. 47E. The processor 4708b may be configured to implement a closed-loop controller architecture 4750a to utilize data from the force transducer 4704b, e.g., force feedback, in combination with a user-defined force, to determine stimulation patterns to be delivered to a muscle 4702b of the MNA 4762b via the signal generator 4706b.

[0247] FIGS. 47G and 47H illustrate example systems 4700c, 4700d for verification of a nerve block 4756, 4756b and for closed-loop control of a myoneural actuator 4762c, 4762d with the nerve block. The system 4700c of FIG. 47G may correspond with the system 4100 of FIG. 41 and the system 4700d of FIG. 47H may correspond with the system 4300 of FIG.43. The systems may be similar, comprising the myoneural actuator 4762c, 4762d coupled to a force transducer 4704c, 4704d configured to measure isometric forces. A signal generator can be configured to stimulate a muscle 4702c, 4702d of the MNA 4762c, 4762d via a nerve 4712c, 4712d, which can be coupled to a signal generator at a stimulation site 4714a, 4714b and a nerve block site 4715a, 4715b. The system 4700c can be coupled to MNA characterization architecture, which may be similar to the architecture the system 4700a of- 55 - 5040745. vlDocket No. 4889-1034001FIG. 47E. The system 4700d can be coupled to closed-loop control architecture, which can be similar to the architecture of the system 4700b of FIG. 47F.

[0248] According to an example embodiment, a biohybrid organ system can comprise a biohybrid small intestine, which can be designed by wrapping an MNA around a small intestine to replicate a mechanical squeezing function of the small intestine. Such a biohybrid organ system may be controlled using neural signals from the nervous system, artificial computer control, or a combination thereof. Furthermore, multiple MNAs can be used to emulate, for example, sequential actuation such as peristalsis. Evaluation of the biohybrid organ system can be performed using optical flow analysis.

[0249] FIGS. 48A-48J illustrate schematically example embodiments of a method of constructing and validating a biohybrid gastrointestinal organ system. The biohybrid organ system can be similar to the biohybrid organ system 4500 of FIG. 45. FIGS. 48 A and 48B illustrate example functional mechanism designs of the biohybrid organ system 4880a, 4880b- 1, 4880b-2. The biohybrid organ system 4880a of FIG. 48 A can comprise a portion of a native organ system, e.g., a gastrointestinal system, and in particular a small intestine 4883. The biohybrid organ system 4880 can further comprise a myoneural actuator 4862a with a transplanted muscle 4802 in coupled arrangement with the small intestine 4883a, for example, encircling the small intestine 4883a. The MNA 4862a can comprise a reinnervated nerve 4812 and a tendon ground 4884, which can provide a mechanical grounding for the MNA 4862a and which may be useful for ensuring that contraction of the MNA 4862a applies a force to the small intestine 4883a. FIG. 48B further illustrates the biohybrid organ system in an ON state 4880b- 1, wherein contraction of MNA can apply a squeezing force to the small intestine9as indicated by the arrows), and an OFF state 4880b-2, wherein relaxation of the MNA can enable recoil of the small intestine (as indicated by the arrows).

[0250] The MNA can be implemented using a process similar to those described in FIGS.46A-46J. FIGS. 48C and 48D illustrate preparation of components of example embodiments of a biohybrid organ system. FIG. 48C illustrates an example embodiment a method of preparing components for a biohybrid organ system. The method can comprise performing an incision on a rat to access a target organ, i.e., the small intestine 4883b. The small intestine 4883b may be isolated and a distal end 4885-2 and a proximal end 4885-1 of a portion of the small intestine 4883b can be tied. Fluid can be injected into the portion of the small intestine 4883 between the tied proximal end 4885-1 and the tied distal end 4885-2 prior to tying off both ends. FIG. 48D illustrates mechanical coupling of a myoneural actuator 4862b to the- 56 - 5040745. vlDocket No. 4889-1034001small intestine 4883c by folding the MNA 4862b around the target organ. FIG. 48F illustrates an example embodiment of a biohybrid organ system comprising a small intestine 4883d and an MNA 4362c.

[0251] FIG. 48F illustrates schematically an example system 4800 for control and actuation of an example embodiment of a biohybrid organ system 4880c-l. The system 4800 comprises the biohybrid organ system 4880c- 1 coupled to a signal generator 4806 configured to stimulate a myoneural actuator 4862d of the biohybrid organ system 4880c-l via a nerve cuff 4814 on nerve 4812 innervating the MNA 4862d. For example, the signal generator can deliver stimulations to the myoneural actuator to control a state of the biohybrid organ system (e.g., an ON state 4880c-2 or an OFF state 4880c-3), which may be similar to the states 4880b- 1, 4880b-2 of FIG. 48B. The system 4800 can further comprise a camera 4886 for analysis of the biohybrid organ system.

[0252] FIG. 48G illustrates example camera (top) and segmented (bottom) images of an example embodiment of a biohybrid organ system. The segmentation, which can comprise identifying the MNA and the organ system, can be used for analysis of, for example, optical flow measurements. FIG. 48H illustrates example frames from optical flow analysis if an example embodiment of a biohybrid organ system, including movement of the MNA and the small intestine. The flow analysis may be based on the segmentation illustrated in FIG. 48G.

[0253] FIG. 481 is a plot 4828a illustrating root-mean-square (RMS) velocity values for regions of interests corresponding to the MNA and the small intestine over time. The plot 4828a may comprise traces of MNA movement and small intestine movement, as determined by optical flow analysis, and may indicate that movement of the MNA is associated with movement of the small intestine. The numbers associated with the peaks the plot 4828a correspond to the numbered frames of FIG. 48H.

[0254] FIGS. 48 J and 48K are plots 4828b, 4828c illustrating organ velocity traces of the small intestine (4828b) and MNA (4828c) based on optical flow analysis. The traces of the plots 4828b, 4828c may indicate synchronized movements between the MNA and the small intestine over time along x- and y- components of optical images.

[0255] FIG. 48L is a scatter plot 4828d of RMS organ velocity versus RMS MNA velocity. Points on the scatter plot 4828d illustrate a relationship between the RMS organ velocity and the RMS MNA velocity at a plurality of time points (each dot being a separate instance), which may represent time points along traces of optical flow analysis of the biohybrid organ system (e.g., the plots 4828a-c of FIGS. 48I-48K). The plot comprises n =- 57 - 5040745. vlDocket No. 4889-1034001186 points and exhibits a Pearson correlation coefficient of r = 0.76. The traces of FIG. 48K and the plot of FIG. 48L may indicate strong correspondence between the myoneural actuator and the small intestine mechanically coupled together and may indicate that myoneural actuators can be used for driving biohybrid organ systems.

[0256] According to some embodiments, the MNA an enable actuation with complete bidirectional isolation from the central nervous system (CNS), which can enable creation of a variety of biohybrid organ systems, ranging from autonomic modulation to neural feedback.

[0257] FIG. 49 illustrates example embodiments of biohybrid organ systems. Such biohybrid organ systems can comprise implanted or external signal generators and sensors. For non-limiting examples, a biohybrid cardiac system (1) can comprise MNAs be wrapped around the heart to provide mechanical actuation, with ventricular systole determined from electrocardiogram (ECG) leads and respiratory signals to synchronize nerve stimulation for closed-loop control of an MNA-based biohybrid heart. A biohybrid respiratory system (2) can comprise MNAs be serially coupled to diaphragm muscle attachment points and intercostal muscles to facilitate contraction during inspiration, synchronized with respiratory and ECG signals. A biohybrid neuromuscular system (3) can comprise MNAs coupled to end organ muscles to modulate muscle dynamics for proprioceptive afferent signaling. Strain and tension sensing from these muscles can be used to estimate muscle state feedback, while data from bionic and virtual limbs can help compute reference trajectories, guiding stimulation patterns for MNA closed-loop control. A biohybrid dermal system (4) can comprise MNAs wrapped around innervated skin grafts for cutaneous afferent signaling, where strain and tension sensing from the skin can provide state feedback, and bionic or virtual limb states can inform stimulation patterns. A biohybrid urinary system (5) can comprise MNAs wrapped around the detrusor muscle of the bladder to provide mechanical compression, assisting with urinary control. Inputs from implanted pressure sensors, brain-computer interfaces, and external controls like smartphone commands can be used to inform stimulation patterns for closed-loop control of an MNA-based biohybrid bladder. A biohybrid gastrointestinal system (6) can comprise a series of MNAs can be wrapped around the intestine to replicate peristaltic motion and modulate the gut-brain axis. According to some embodiments, the biohybrid organ systems can be actuated by a nervous system of a human or animal or by external signaling.

[0258] FIG. 50 illustrates an example embodiment of a methodology for design of a biohybrid organ systems using MNAs. A design workflow can begin with selection of a- 58 - 5040745. vlDocket No. 4889-1034001target organ to be controlled (1 - target biology manipulation). Factors such as biological principles 5001 can inform motion dynamics, including contractile and fatigue characteristics, attachment points, and tension levels. Additionally, type of force 5002 required, for example, torsional, squeezing, peristaltic, or contraction, can guide the necessary dynamics. Force scale 5003 can determine a magnitude and duration of contractile activity needed to meet functional demands. Together, these factors can define the design of a functional mechanism (2 - Functional mechanism design).

[0259] Once the mechanism design has been finalized, surgical planning (3 - surgical design) may be carried out and may consider factors such as tissue interface and coupling. For example, tissue interface and coupling considerations 5007 can comprise, as a nonlimiting example, an MNA interacting with cardiac tissue, which may require a material that isolates tissue conduction may be required. Surgical constraints 5009 may help determine which tissue can be used to create the MNA. Myoneural considerations 5011 can dictate a type and size of muscle used, a sensory nerve, and required numbers of said muscles and sensory nerves needed to meet functional demands. Neuromodulation capacity 5013 can specify stimulation parameters needed to achieve a desired functionality. Finally, anatomical factors 5015 can guide optimal locations for obtaining MNA components and an optimal surgical approach to construct the biohybrid system.

[0260] The surgical design can then be executed and evaluated (4 - Implementation and Evaluation). If necessary, revision surgeries can be conducted to further refine the biohybrid organ system. Functional biohybrid systems can be stored in a database 5017, which can later inform new mechanism designs.Discussion

[0261] According to some embodiments, a fully biointegrable actuator capable of high-performance continuous actuation could facilitate modulation of various biological functions, ranging from restoring organ function to generating neural afferent feedback for neuroprostheses and virtual experiences. Despite significant advancements in synthetic actuator and biofabrication technologies, such an actuator with the aforementioned capabilities has not yet been realized. As described herein, example embodiments of an MNA were introduced as an engineered construct, which may combine bioimplantable and actuation properties of skeletal muscle while demonstrating fatigue-resistant continuous actuation under artificial computer control.- 59 - 5040745. vlDocketNo. 4889-1034001

[0262] Example embodiments of MNAs may be built on a myoneural framework that transforms native skeletal muscle into an actuator for artificial computer control by redirecting the central nervous system's (CNS) motor control through denervation of the native motor nerve while improving fatigue resistance through sensory reinnervation.Morphological and electrophysiological evidence may demonstrate that the sensory nerve successfully reinnervates muscle fibers, preserving muscle function without the need for external interventions and enabling control of muscle fibers via the natural neuromuscular pathway. The MNA may exhibit significantly enhanced fatigue resistance in both open-loop and closed-loop settings, and may further exhibit different fatigue dynamics compared to native muscles that lacked the myoneural manipulation.

[0263] Construction of MNAs can involve the reinnervation of a pure sensory nerve, which may be guided by a dual design rationale: to preserve muscle function, as supported by sensory protection interventions

[0013] , and to modify the recruitment biophysics under electrical nerve stimulation. When peripheral nerves with varying axon sizes, such as motor nerves, are subjected to extraneural stimulation, the lower resistance of large-diameter axons may lead to preferential activation of motor units that typically innervate fast-twitch muscle fibers, which may thereby accelerate muscle fatigue. It is hypothesized that the more uniform fiber size distribution of sensory nerves, compared to motor nerves, may contribute to the enhanced fatigue resistance observed in MNAs, offering a potential solution to this persistent limitation. This hypothesis may align with recent evidence suggesting that by uniformly expressing light-sensitive ion channels in nerves, muscles can be optically controlled in a more orderly, natural manner, improving recruitment profiles, controllability, and fatigue resistance [4], Such enhanced muscle modulation can result from greater access to axons in the nerve, allowing for the activation of a broader range of fibers, not just the large ones.

[0264] Seminal studies on cross-reinnervation have provided insights into muscle fiber type transformations [7], while more recent work has demonstrated that peripheral motor nerve transfers can induce a donor-specific shift toward slower fiber types, potentially enhancing fatigue resistance

[0014] , However, these transfers can also lead to higher conduction velocities, which may suggest that motor units contain large-diameter fibers that are more susceptible to fatigue. These studies may imply that a sensory nerve transfer might induce fiber type transformation; however, no significant shift was observed in the current study. An absence of motor control may prevent fiber transformation, as shown in studies where the lack of spinal cord signals or changes in the pattern of impulse activity reaching the muscle- 60 - 5040745. vlDocket No. 4889-1034001had no significant effect on fiber transformation [6], Alternatively, the lack of fiber transformation may be due to the absence of innervation signaling from sensory nerves that may drive myosin heavy chain protein expression and fiber type changes. A prevailing hypothesis may be that the observed fatigue resistance arises from the axonal biophysics of the sensory nerve; however, the possibility of fiber type transformation may not be entirely ruled out. Such fiber type transformation could potentially be revealed with an alternative immune-histological procedure

[0015] ,

[0265] Example embodiments of MNA technology, as described herein, can facilitate a variety of biohybrid applications in medicine. For non-limiting examples, biohybrid organ systems may demonstrate how MNAs can modulate organ mechanics. Conditions such as ileus, diabetic enteropathy, and Crohn's Disease can disrupt intestinal contractions, hindering the movement of food and nutrients through the gastrointestinal (GI) tract. Embodiments such as those of FIGS. 45-48J may show how MNAs can interface with the small intestine to control its mechanics on demand, which may establish a groundwork for a biohybrid organ system that could restore GI tract function. Additionally, mechanical actuation of the small intestine and the GI tract can trigger a series of physiological effects, including gene expression from mechanosensor activation, hormone release such as GLP-1, and modulation of neurotransmitters and enzymes. These changes may improve nutrient absorption, energy balance, gut-brain axis modulation, and influence neurological functions like mood, anxiety, and cognitive function. According to such embodiments, MNAs could serve as tools to modulate organ mechanics and explore their downstream effects.

[0266] In some embodiments, due to its neural isolation from the CNS, the MNA may present a promising tool for controlling autonomic functions. In addition to modulating neural sensory afferents and the GI tract, autonomic functions such as respiration and urinary control could be achieved through mechanical modulation of the lungs and bladder. For respiratory modulation in individuals with ventilatory insufficiency, multiple MNAs could be coupled to the diaphragm and intercostal muscles, enabling coordinated contraction during inspiration, with respiratory sensor signals guiding the timing and intensity of the actuation. For urinary modulation, one or more MNAs could be wrapped around the detrusor muscle of the bladder to generate compression, aiding in urinary excretion through control inputs, such as smartphone commands.

[0267] While biohybrid system applications may hold great promise, understanding tissue interactions between the MNA and various tissues, such as smooth or cardiac muscle, can be- 61 - 5040745. vlDocketNo. 4889-1034001essential. Insights gained from dynamic cardiomyoplasty, in which skeletal muscle is wrapped around the heart to assist with pumping, have shown that differences in electrical conduction properties between skeletal and cardiac muscle can lead to ventricular arrhythmias and fibrillation

[0011] , According to some embodiments, MNAs can be applied towards assisting modulation of heart mechanics through engineering tissue interfaces to electrically isolate the MNA from the myocardium. In contrast to dynamic cardiomyoplasty, where native skeletal muscle requires extensive training to become fatigue-resistant

[0011] , the MNA could provide fatigue resistance without such conditioning. Additionally, a muscle used in cardiomyoplasty may retain its original motor nerve, which can limit neural isolation from the CNS and hinder artificial control. In some example embodiments, nerve stimulation in the MNA could offer greater control over muscle fibers. Synchronizing nerve stimulation with ventricular systole could enable closed-loop assistance for the MNA. A computational framework based on functional demands and tissue properties may be valuable for determining appropriate design choices for the MNA in specific biohybrid applications, such as size, tissue coupling modality, and stimulation parameters.

[0268] According to some embodiments, because MNAs may be based on a reconstructive surgical technique that engineers skeletal muscle along with a stimulation interface, e.g., standard nerve cuff electrodes, MNAs may be highly translatable for various medical applications. Similar surgical techniques are commonly used in facial nerve and brachial plexus surgeries, as well as in reconstructive methods for bionic integration

[0017] , According to some embodiments, a wireless pulse generator and nerve electrodes, both commonly used, can also be implanted during a same procedure. Furthermore, as described hereinabove, MNAs may not require external interventions for reinnervation and long-term functionality. The sustainability data (presented at least in FIGS. 29-3 OB) may represent extreme cases, with external interventions, such as stimulation protocols through the implanted stimulator, having the potential to enhance MNA sustainability and capacity, as well as to alter muscle phenotype for desired functionality [7], According to some embodiments, the same stimulator can be utilized for reversible electrical nerve block

[0010] or antidromic conduction block to prevent unintended signaling during operation. In some embodiments, nerve block may only be necessary if the stimulation parameters for MNA control can cause undesired or painful sensations. The relationship between stimulation parameters and evoked sensations could be explored in patients with sensory protection

[0013] to determine if nerve block is needed.- 62 - 5040745. vlDocket No. 4889-1034001

[0269] According to some embodiments, use of an autograft skeletal muscle in the MNA procedure can eliminate issues such as infection, rejection, and immunosuppression that may typically be associated with allografts, xenografts, or engineered tissues. Furthermore, the MNA procedure can be relatively straightforward and safe compared to implantable mechanical assist devices or organ transplants. In some embodiments, skeletal muscle tissue from the same limb, such as those from reconstructive surgery, that would otherwise be discarded, can become available for creating constructs like the MNA. Unlike procedures such as cardiomyoplasty, where the muscle must be in close proximity to the heart due to the requirement for an intact neurovascular bundle

[0011] , MNAs can be constructed from skeletal muscles sourced from any location. According to some embodiments, muscles like the latissimus dorsi from the back, rectus abdominis from the abdomen, or pectoralis major from the chest, which may commonly be used as autografts in other procedures, can be used for MNA construction. In some embodiments, sensory nerves that innervate skin areas near the organ of interest can be utilized to construct the MNA. Notably, it may be important to consider selection of muscle and nerve size for the MNA to meet specific functional clinical requirements.ApplicationsFatigue-resistant FES control of muscles

[0270] As previously stated, FES of muscles can cause unnatural recruitment of muscle fibers, which may lead to rapid fatigue and limiting chronic use of FES for muscle control. Moreover, after neurological conditions like SCI, muscle fibers can undergo gene expression changes towards fast-twitch fibers. According to some embodiments, a closed-loop adaptive architecture can be used to transform a fast-twitch muscle from individuals with neurological conditions into a slow-twitch muscle fiber. After the muscle has been transformed, the same neuromuscular system can be used to artificially control the muscle.

[0271] An example embodiment of such a closed-loop architecture is described hereinabove with reference to FIG. 11. In some embodiments, a system identification procedure, which can be similar to the one described hereinabove with reference to FIGS. 5A-5D, can serve to derive a mathematical model of the muscle. Stimulation from signal generators, e.g., the IPG, can cause a muscle to contract and muscle dynamics can be sensed by the sensors. Subsequently, muscle states or neuromuscular dynamics can be output, which can comprise, as non-limiting examples, to muscle activation, muscle length, muscle velocity,- 63 - 5040745. vlDocketNo. 4889-1034001and muscle force. These muscle states, which can be used together with a reference signal coming from a higher-level controller (e.g., the reference block), can be fed to the feedback controller, such as but not limited to a proportional integrative derivative controller. The higher-level controller (the reference block) can determine a reference signal, which can originate from systems such as but not limited to brain-computer interfaces, wearable sensors that estimate gait phase, and neuro-refl exive architectures. The reference signal can be fed to the controller previously described to estimate the stimulation signal to be applied to the muscle. The signal estimated by the inverted model (feedforward signal) can be added to the signal estimated by the feedback controller, to generate a signal that informs the IPG with a stimulation pattern.

[0272] According to some embodiments, such an architecture can run at any desired bandwidth. In some embodiments, bandwidths of at least 100 Hz are desirable may be usable for accurate muscle control. In patients with SCI, muscles may not receive signals from the CNS. This can be advantageous to phenotype transformation as, after the muscle is transformed, there may be no signals that can reverse a phenotype of the muscle. However, a phenotype evaluator can be used to detect any potential deviations.Transformation to fast-twitch fibers in elderly individuals

[0273] As healthy individuals age, muscle function and mass may start to decline. This decrease in function is commonly referred to as sarcopenia and may begin in approximately the fifth decade of life

[0018] , From studies on human muscle biopsies, a decrease of fast-twitch and an increase in slow-twitch muscle fibers may be observed. In some embodiments, the closed-loop adaptive architecture can be used to transform muscles into fast-twitch muscles from elderly individuals. After the muscle has been transformed, the muscle can be volitionally used for normal operation. In some embodiments, due to the neural signals from the CNS used during normal operation, the phenotype evaluation architecture may e necessary to monitor the phenotype of the muscle and adjust if necessary.

[0274] In some embodiments, the closed-loop adaptive architecture with the phenotype evaluation can be used in healthy individuals before muscles start undergoing fiber type changes. The phenotype evaluation can detect an onset of muscle fibers shifting to a slow-twitch phenotype, and start stimulation protocols to revert this change. This preventive embodiment might prove more effective as the number of fibers in a muscle that need to be- 64 - 5040745. vlDocket No. 4889-1034001transformed is likely smaller compared to when the muscle already underwent significant transformation due to aging.Muscle-based biohybrid organ systems

[0275] As described hereinabove, example embodiments of methods and systems described herein can be directed to biohybrid organ systems and closed-loop adaptive architectures for further engineering of myoneural actuators. In some embodiments, after a muscle has been denervated and reinnervated with a sensory nerve, a signal generator, e.g., an IPG can be used to stimulate the muscle with patterns to accelerate a reinnervation process and augment muscle capacity. Additionally, with the neuromuscular system, muscle contractions can be sensed and muscle states can be computed to inform a reference estimator. The reference estimator, together with a desired phenotype, can inform the phenotype controller which determines the stimulation parameters to the IPG for phenotype transformation. An example embodiment of such a closed-loop architecture, which can be used for further engineering muscles and MNAs for different functional requirements, is described hereinabove with reference to FIG. 10. In some embodiments, because MNA may not receive signals from the CNS, a possibility of phenotype reversal may be very unlikely. However, depending on stimulation parameters used to control the MNA, especially if they differ from that of an intended phenotype, phenotype evaluation can be required to maintain the muscle in its desired phenotype.

[0276] FIG. 51 illustrates schematically example embodiments of myoneural actuators 5162a, 5162b, 5162c, 5162d in stages of construction, refinement, and application. The myoneural actuator 5162a can be constructed from a base muscle 5102, e.g., a skeletal muscle, and a sensory nerve 5112. A native motor nerve 5113 of the muscle may be transected. The MNA 5162b may undergo phenotype transformation, which may comprise stimulation 5106 with stimulation patterns and sensing 5104a of muscle dynamics in a closed loop architecture. The MNA 5162c, which may be transformed to a desired or target phenotype, can be applied towards muscle restoration, which can be controlled in a closed-loop architecture comprising sensors 5104b, a signal generator 5106b, and a controller 5108. Alternatively, the myoneural actuator 5162d can be used to form a biohybrid organ system by mechanically coupling the myoneural actuator to an organ system of a human or animal and may be operated in conjunction with other sensing modalities, e.g., electrocardiography for a biohybrid cardiac system or respiration for a biohybrid respiratory system.- 65 - 5040745. vlDocketNo. 4889-1034001Transformation of fibers for compensatory activities

[0277] Neurological conditions that result in the loss of neural input, such as peripheral neuropathies, stroke, multiple sclerosis, motor neuron disease, SCI, cerebral palsy, can result in fiber type transformations. This may be due to the changes in neural signals that muscles receive. Recently, the success of weight loss drugs such as Ozempic, may reveal that muscle mass and function can be severely impacted during the rapid weight loss process

[0019] , In these situations, example embodiments of closed-loop adaptive architectures, which can be implemented by systems for transforming a phenotype of a muscle or myoneural actuator, can be used to transform a muscle to a desired phenotype. In other embodiments, phenotype transformation can be initiated in preparation of such situations. For example, an example embodiment of a method for transforming a phenotype of a muscle can be initiated as soon as weight loss drugs are taken.Transformation of fibers for high-performance activities

[0278] Interestingly, exercise of muscles may not result in the same level of phenotype transformation as when electrical stimulation is used [7], During sustained regimes of endurance exercise, changes in myosin isoform composition may usually be restricted to a fast subclass, which can include changes only within type II fibers. Changes in type I fibers may tend to be very small or not detectable at all. This behavior may be due to recruitment of fibers during exercise being limited to only a population of fibers that undergo a significant change in exercise conditions. As described hereinabove, electrical stimulation can generate an unnatural recruitment order, which may indicate that electrical stimulation can indiscriminately activate a lot of the fibers. Additionally, electrical stimulation can be applied for extended durations, even 24 hours a day, whereas exercise may be more intermittent. A closed-loop adaptive stimulation architecture can be applied to particular muscles for high-performance activities, for example, to perform sprinting. Additionally, the stimulation protocols can be performed together with training regimes to get the body acclimated to the transformed muscle.

[0279] Computer Support

[0280] FIG. 52 is a schematic view of a computer network in which embodiments may be implemented. Client computer(s) / devices 50 and server computer(s) 60 provide processing,- 66 - 5040745. vlDocketNo. 4889-1034001storage, and input / output (I / O) devices executing application programs and the like. Client computer(s) / device(s) 50 can also be linked through communications network 70 to other computing devices, including other client device(s) / processor(s) 50 and server computer(s) 60. The communications network 70 can be part of a remote access network, a global network (e.g., the Internet), cloud computing servers or service, a worldwide collection of computers, local area or wide area networks, and gateways that currently use respective protocols (e.g., TCP / IP, Bluetooth®, etc.) to communicate with one another. Other electronic device / computer network architectures are also suitable.

[0281] FIG. 53 is a block diagram illustrating an example embodiment of a computer node (e.g., client processor(s) / device(s) 50 or server computer(s) 60) in the computer network 70 of FIG. 52. Each computer node 50, 60 contains system bus 79, where a bus is a set of hardware lines used for data transfer among components of a computer or processing system. The system bus 79 is essentially a shared conduit that connects different elements of a computer system (e.g., processor, disk storage, memory, I / O ports, network ports, etc.) that enables transfer of information between the elements. Attached to the system bus 79 is an I / O devices interface 82 for connecting various input and output devices (e.g., keyboard, mouse, display(s), printer(s), speaker(s), etc.) to the computer node 50, 60. A network interface 86 allows the computer node to connect to various other devices attached to a network (e.g., the network 70 of FIG. 52). A memory 90 provides volatile storage for computer software instructions 92a and data 94a used to implement embodiments of the present disclosure (e.g., the method of FIGS. 5A-5D, etc.). A disk storage 95 provides non-volatile storage for the computer software instructions 92b and data 94b used to implement an embodiment of the present disclosure. A central processor unit (CPU) 84 is also attached to the system bus 79 and provides for execution of computer instructions.

[0282] In an embodiment, the processor routines 92a-92b and data 94a-94b are a computer program product (generally referenced as 92), including a non-transitory, computer readable medium (e.g., a removable storage medium such as DVD-ROM(s), CD-ROM(s), diskette(s), tape(s), etc.) that provides at least a portion of the software instructions for the disclosure methods. The computer program product 92 can be installed by any suitable software installation procedure, as is well known in the art. In another embodiment, at least a portion of the software instructions may also be downloaded over a cable, communication, and / or wireless connection. In other embodiments, the disclosure programs are a computer program propagated signal product embodied on a propagated signal on a propagation- 67 - 5040745. vlDocket No. 4889-1034001medium (e.g., a radio wave, an infrared wave, a laser wave, a sound wave, or an electrical wave propagated over a global network such as the Internet, or other network(s)). Such carrier medium or signals provide at least a portion of the software instructions for the present disclosure routines / program 92.

[0283] In alternative embodiments, the propagated signal is an analog carrier wave or digital signal carried on the propagated medium. For example, the propagated signal may be a digitized signal propagated over a global network e.g., the Internet), a telecommunications network, or other networks (such as the network 70 of FIG. 52). In one embodiment, the propagated signal is a signal that is transmitted over the propagation medium over a period of time, such as the instructions for a software application sent in packets over a network over a period of milliseconds, seconds, minutes, or longer. In another embodiment, the computer readable medium of the computer program product 92 is a propagation medium that the computer system 50 may receive and read, such as by receiving the propagation medium and identifying a propagated signal embodied in the propagation medium, as described above for computer program propagated signal product.

[0284] Generally speaking, the term “carrier medium” or transient carrier encompasses the foregoing transient signals, propagated signals, propagated medium, storage medium, and the like.

[0285] In other embodiments, the program product 92 may be implemented as a so-called Software as a Service (SaaS), or other installation or communication supporting end-users.

[0286] Embodiments or aspects thereof may be implemented in the form of hardware including but not limited to hardware circuitry, firmware, or software. If implemented in software, the software may be stored on any non-transient computer readable medium that is configured to enable a processor to load the software or subsets of instructions thereof. The processor then executes the instructions and is configured to operate or cause an apparatus to operate in a manner as described herein.

[0287] Further, hardware, firmware, software, routines, or instructions may be described herein as performing certain actions and / or functions of the data processors. However, it should be appreciated that such descriptions contained herein are merely for convenience and that such actions in fact result from computing devices, processors, controllers, or other devices executing the firmware, software, routines, instructions, etc.

[0288] It should be understood that the flow diagrams, block diagrams, and network diagrams may include more or fewer elements, be arranged differently, or be represented- 68 - 5040745. vlDocketNo. 4889-1034001differently. But it further should be understood that certain implementations may dictate the block and network diagrams and the number of block and network diagrams illustrating the execution of the embodiments be implemented in a particular way.

[0289] Accordingly, further embodiments may also be implemented in a variety of computer architectures, physical, virtual, cloud computers, and / or some combination thereof, and, thus, the data processors described herein are intended for purposes of illustration only and not as a limitation of the embodiments.

[0290] The teachings of all patents, published applications and references cited herein are incorporated by reference in their entirety.While example embodiments have been particularly shown and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope of the embodiments encompassed by the appended claims.References

[0291] 1. Roberts, T. J. et al. The Multi-Scale, Three-Dimensional Nature of Skeletal Muscle Contraction. Physiology 34, 402-408 (2019).

[0292] 2. Llewellyn, M. E., Thompson, K. R., Deisseroth, K. & Delp, S. L. Orderly recruitment of motor units under optical control in vivo. Nat. Med. 16, 1161-1165 (2010).

[0293] 3. McDonnall, D., Clark, G. A. & Normann, R. A. Interleaved, multisite electrical stimulation of cat sciatic nerve produces fatigue-resistant, ripple-free motor responses. IEEE Trans. Neural Syst. Rehabil. Eng. 12, 208-215 (2004).

[0294] 4. Herrera-Arcos, G. et al. Closed-loop optogenetic neuromodulation enables high-fidelity fatigue-resistant muscle control. Sci. Robot. 9, eadi8995 (2024).

[0295] 5. Wan, J., Qin, Z., Wang, P., Sun, Y. & Liu, X. Muscle fatigue: general understanding and treatment. Exp. Mol. Med. 49, e384-e384 (2017).

[0296] 6. Salmons, S. & Sreter, F. A. Significance of impulse activity in the transformation of skeletal muscle type. Nature 263, 30-34 (1976).

[0297] 7. Salmons, S. Adaptive change in electrically stimulated muscle: A framework for the design of clinical protocols. Muscle Nerve 40, 918-935 (2009).

[0298] 8. Taylor, C. R. et al. Magnetomicrometry. Sci. Robot. 6, eabg0656 (2021).

[0299] 9. Taylor, C. R. et al. Untethered muscle tracking using magnetomicrometry. Front. Bioeng. Biotechnol. 10, 1010275 (2022).- 69 - 5040745. vlDocketNo. 4889-1034001

[0300] 10. Yeon, S. H., Landis, C., Herrera-Arcos, G., Song, H. & Herr, H. M. Compact Reversible Nerve Block System for Wearable Bioelectronics, in 202446th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) 1-5 (IEEE, Orlando, FL, USA, 2024). doi: 10.1109 / EMBC53108.2024.10782179.

[0301] 11. Carpentier, A. et al. Dynamic cardiomyoplasty at seven years. J. Thorac. Cardiovasc. Surg. 106, 42-54 (1993).

[0302] 12. Cederna, P., Nghiem, B., Hu, Y., Sando, I. & Urbanchek, M. Sensory protection to enhance functional recovery following proximal nerve injuries: current trends. Plast. Aesthetic Res. 2, 202 (2015).

[0303] 13. Adi dharma, W. et al. Sensory nerve regeneration and reinnervation in muscle following peripheral nerve injury. Muscle Nerve 66, 384-396 (2022).

[0304] 14. Bergmeister, K. D. et al. Peripheral nerve transfers change target muscle structure and function. Sci. Adv. 5, eaau2956 (2019).

[0305] 15. Bloemberg, D. & Quadrilatero, J. Rapid Determination of Myosin Heavy Chain Expression in Rat, Mouse, and Human Skeletal Muscle Using Multicolor Immunofluorescence Analysis. PLoS ONE 7 , e35273 (2012).

[0306] 16. Herr, H. & Carty, M. J. The agonist-antagonist myoneural interface. Tech. Orthop. Rockv. Md36, 337-344 (2021).

[0307] 17. Shu, T., Herrera-Arcos, G., Taylor, C. R. & Herr, H. M. Mechanoneural interfaces for bionic integration. Nat. Rev. Bioeng. 2, 374-391 (2024).

[0308] 18. Cruz-Jentoft, A. J. & Sayer, A. A. Sarcopenia. The Lancet 393 , 2636-2646 (2019).

[0309] 19. Sheridan, C. Innovators want pills to treat sickle cell disease. Can they match gene therapy? Nat. Biotechnol. 42, 347-350 (2024).- 70 - 5040745. vl

Claims

Docket No. 4889-1034001CLAIMSWhat is claimed is:

1. A system for transforming a phenotype of a muscle of a human or animal, the system comprising:a sensor configured to detect signals associated with a muscle of a human or animal;a signal generator configured to stimulate the muscle; anda processor configured to:determine a stimulus pattern based on neuromuscular dynamics computed from the signals detected, the neuromuscular dynamics computed indicative of a current phenotype of the muscle; anddeliver the stimulus pattern determined to the muscle via the signal generator, the stimulus pattern determined and delivered causing a transformation of the muscle toward a target phenotype.

2. The system of Claim 1, wherein the sensor is configured to detect the signals responsive to a stimulus delivered to the muscle, the stimulus being delivered to the muscle by one or more of a nervous system of the human or animal or by the signal generator.

3. The system of Claim 2, wherein the processor is further configured to cause the signal generator to deliver the stimulus.

4. The system of Claim 3, wherein the stimulus delivered comprises a waveform having one or more of (i) a low frequency short duration stimulus, (ii) a high frequency long duration stimulus, or (iii) a stimulus having one or more frequencies.

5. The system of Claim 2, wherein, in an iterative manner, the sensor is configured to detect the signals associated with the muscle and the processor is configured to (i) determine the stimulus pattern based on the neuromuscular dynamics, the neuromuscular dynamics being computed from signals detected responsive to the- 71 - 5040745. vlDocketNo. 4889-1034001stimulus to the muscle, and (ii) deliver the stimulus pattern generated to the muscle via the signal generator until the target phenotype of the muscle is achieved.

6. The system of Claim 1, wherein the processor is further configured to compute the neuromuscular dynamics from the signals detected.

7. The system of Claim 6, wherein the neuromuscular dynamics computed include one or more of:muscle activation, muscle length, muscle velocity, muscle force, tendon length, tendon force, muscle fatigue, muscle fatigue rate, and peak muscle state over a dynamic frequency range.

8. The system of Claim 7, wherein the processor is further configured to:compute the muscle force by forming a muscle-tendon model based on muscle activation, muscle length, and muscle velocity; andcompute the tendon force through a tendon dynamic model based on the tendon length.

9. The system of Claim 1, wherein the processor is further configured to infer the current phenotype of the muscle based on the neuromuscular dynamics computed and wherein the processor is configured to generate the stimulus pattern based on the current phenotype inferred.

10. The system of Claim 9, wherein a phenotype of the muscle includes a fast twitch phenotype or a slow twitch phenotype.

11. The method of Claim 9, wherein the system further comprises an auxiliary sensing system and wherein the processor is configured to infer the current phenotype based on the neuromuscular dynamics computed and data acquired from the auxiliary sensing system, the auxiliary sensing system comprising one or more of magnetic resonance imaging, computed tomography, metabolic profiling, genotyping, and myosin heavy chain isoform profiling.- 72 - 5040745. vlDocket No. 4889-103400112. The system of Claim 1, wherein the sensor comprises one or more of:an implantable electrode, a surface electrode, an implantable magnetic sensor, an ultrasound sensor, or an optical sensor.

13. The system of Claim 1, wherein the sensor is configured to detect the signals associated with the muscle using one or more of:electromyography, magnetomicrometry, ultrasound imaging, near-infrared imaging, laser doppler vibrometry, or optical coherence tomography.

14. The system of Claim 1, wherein the signal generator is configured to stimulate the muscle using one or more of:a transcutaneous electrical stimulation, a fine-wire stimulation, a nerve cuff stimulator, an ultrasound stimulation, or a magnetic stimulation.

15. The system of Claim 1, wherein the processor is configured to determine the stimulus pattern based on the neuromuscular dynamics computed and one or more of (i) a model estimating the current phenotype of the muscle based on a repository of muscle stimulus data or (ii) a reference library comprising stimulation patterns versus muscle phenotypes.

16. The system of Claim 1, wherein the signal generator is coupled electrically to a nerve innervating the muscle and wherein the signal generator delivers the stimulus pattern to the muscle through the nerve.

17. The system of Claim 16, wherein the signal generator is coupled electrically to a nerve at a first site and a second site, the first site being proximal to the brain with respect to the second site, and wherein the signal generator is configured to apply a nerve block at the first site.

18. The system of Claim 17, wherein the nerve block applied by the signal generator comprises a kilohertz-frequency alternating current and wherein the nerve block- 73 - 5040745. vlDocketNo. 4889-1034001applied blocks propagation, in a direction of the first site, of a stimulus delivered to the second site by the signal generator.

19. The system of Claim 1, wherein the processor is further configured to:actuate the muscle by delivering a stimulus to the muscle via the signal generator, the muscle performing a native function of the muscle responsive to the actuating.

20. A method of transforming a phenotype of a muscle of a human or animal, the method comprising:detecting signals associated with a muscle of a human or animal; computing neuromuscular dynamics of the muscle based on the signals detected, the neuromuscular dynamics indicative of a current phenotype of the muscle;determining a stimulus pattern based on the neuromuscular dynamics computed; anddelivering the stimulus pattern determined to the muscle, the stimulus pattern determined and delivered to the muscle causing a transformation of the muscle toward a target phenotype.

21. The method of Claim 20, wherein detecting the signals is responsive to a stimulus delivered to the muscle by a nervous system of the human or animal or by a signal generator configured to stimulate the muscle.

22. The method of Claim 21, further comprising causing delivery of the stimulus to the nerve.

23. The method of Claim 22, further comprising encoding the stimulus delivered with a temporally varying pattern.

24. The method of Claim 21, wherein the detecting the signals associated with the muscle, the computing the neuromuscular dynamics, the determining the stimulus- 74 - 5040745. vlDocket No. 4889-1034001pattern, and the delivering the stimulus pattern are repeated in an iterative manner until the target phenotype of the muscle is achieved.

25. The method of Claim 20, wherein detecting the signal comprises:detecting one or more of an electrical signal, a magnetic signal, an ultrasound signal, or an ultrasound signal.

26. The method of Claim 20, wherein computing the neuromuscular dynamics comprises:computing, based on the signal detected, one or more of muscle activation, muscle length, muscle velocity, muscle force, tendon length, tendon force, muscle fatigue, muscle fatigue rate, and peak muscle state over a dynamic frequency range.

27. The method of Claim 26, wherein computing the neuromuscular dynamics further comprises:computing the muscle force by forming a muscle-tendon model based on muscle activation, muscle length, and muscle velocity; andcomputing the tendon force through a tendon dynamic model based on the tendon length.

28. The method of Claim 20, further comprising inferring the current phenotype of the muscle based on the neuromuscular dynamics computed and wherein determining the stimulus pattern is based on the current phenotype inferred.

29. The method of Claim 28, wherein inferring a phenotype of the muscle includes a fast twitch phenotype and a slow twitch phenotype.

30. The method of Claim 28, further comprising:augmenting the neuromuscular dynamics using data collected from an auxiliary sensing system, the auxiliary sensing system comprising one or more of magnetic resonance imaging, computed tomography, metabolic profiling, genotyping, and myosin heavy chain isoform profiling; andwherein inferring the current phenotype is based on the neuromuscular dynamics computed and augmented.- 75 - 5040745. vlDocketNo. 4889-103400131. The method of Claim 20, determining the stimulus pattern is based on the neuromuscular dynamics computed and one or more of (i) a model estimating the current phenotype of the muscle based on a repository of muscle stimulus data or (ii) a reference library comprising stimulation patterns versus muscle phenotypes.

32. The method of Claim 20, wherein delivering the stimulus pattern includes delivering:a transcutaneous electrical stimulation, fine-wire stimulation, an electrical nerve cuff, ultrasound stimulation, or magnetic stimulation.

33. The method of Claim 20, wherein delivering the stimulus pattern to the muscle comprises delivering the stimulus pattern to a nerve innervating the muscle.

34. The method of Claim 33, wherein deliver the stimulus pattern to the nerve further comprises delivering a nerve block signal to a first site along a nerve and delivering the stimulus pattern at a second site along the nerve, the nerve block preventing a transmission of the stimulus pattern delivered beyond the first site from the second site.

35. The method of Claim 20, further comprising:for the muscle of the target phenotype, actuating the muscle by delivering a stimulus to the muscle.

36. The method of Claim 20, further comprising:for the muscle of the target phenotype, monitoring the current phenotype of the muscle by detecting the signals associated with the muscle; andresponsive to a change of the muscle from the target phenotype, determining the stimulus pattern based on the neuromuscular dynamics computed from the signals detected and delivering the stimulus pattern determined.

37. A biohybrid organ system, the biohybrid organ system comprising:an organ system of a human or animal; and- 76 - 5040745. vlDocketNo. 4889-1034001a myoneural actuator comprising a transplanted muscle of the human or animal, the transplanted muscle in coupled arrangement with a portion of the organ system, a contraction of the transplanted muscle responsive to a stimulus mechanically actuating the portion of the organ system to perform a physiological function of the organ system.

38. The biohybrid organ system of Claim 37, wherein the myoneural actuator further comprises:a signal generator functionally coupled to the transplanted muscle, the signal generator configured to deliver the stimulus to the muscle.

39. The biohybrid organ system of Claim 38, wherein the signal generator is functionally coupled to the muscle via a nerve of the human or animal and wherein the signal generator is configured to cause the delivery of stimulus to the muscle via the nerve.

40. The biohybrid organ system of Claim 39, wherein:the muscle is decoupled from a native efferent nerve of the muscle; and the nerve comprises a transected portion of a sensory nerve of the human or animal, the muscle being reinnervated using the nerve, wherein the muscle is coupled to the nerve via a restored neuromuscular junction.

41. The biohybrid organ system of Claim 38, further comprising:a sensor configured to detect signals associated with the transplanted muscle; anda processor configured to:compute neuromuscular dynamics of the transplanted muscles based on the signals detected and to transform a phenotype of the transplanted muscle based on the neuromuscular dynamics computed; orcause the signal generator to deliver the stimulus of the muscle based on a function of the organ system.- 77 - 5040745. vlDocket No. 4889-103400142. The biohybrid organ system of Claim 37, wherein the transplanted muscle of the myoneural actuator is configured to be fatigue resistant based on properties of a nerve innervating the transplanted muscle or on a phenotype of the transplanted muscle.

43. The biohybrid organ system of Claim 37, wherein:the transplanted muscle is configured to receive the stimulus from a nervous system of the human or animal.

44. The biohybrid organ system of Claim 37, wherein the myoneural actuator is a first myoneural actuator and wherein the biohybrid organ system comprises at least one additional myoneural actuator in coupled arrangement with the organ system, the myoneural actuator and the at least one additional myoneural activator configured to actuate the organ system to perform a physiological function of the organ system.

45. The biohybrid organ system of Claim 37, wherein the organ system comprises:a digestive system and wherein the myoneural actuator encircles a portion of an intestine;a cardiac system and wherein the myoneural actuator wraps around a portion of a heart;a respiratory system and wherein the myoneural actuator is mechanically coupled to a diaphragm or an intercostal muscle;a urinary system and wherein the myoneural actuator encircles a portion of a bladder;a sphincter system;a smooth muscle system; orskin and wherein the myoneural actuator is mechanically coupled to a skin graft.

46. A method of actuating an organ system of a human or animal, the method comprising:surgically coupling a muscle to an organ system of the human or animal to position the muscle in coupled arrangement with a portion of the organ system; and- 78 - 5040745. vlDocketNo. 4889-1034001causing the muscle surgically coupled to the organ system to contract by stimulating the muscle, the muscle contracting mechanically actuating the portion of organ system to perform a physiological function of the organ system.

47. The method of Claim 46, further comprising:transecting a nerve of the human or animal;reinnervating the muscle using the nerve transected, the stimulus to the muscle being deliverable via the nerve; andstimulating the muscle via the nerve.

48. The method of Claim 47, further comprising:selecting the nerve based on one or more axonal biophysical properties of the nerve, and wherein the nerve is a peripheral nerve or a sensory nerve.

49. The method of Claim 47, further comprising:delivering a nerve block to the nerve transected, the nerve block configured to prevent transmission of a stimulus of the nerve in an afferent direction.

50. The method of Claim 46, further comprising:transecting a native efferent nerve of the muscle extracted and operatively coupled, the transecting the native efferent nerve disconnecting the muscle from an efferent signal from a nervous system of the human or animal to the muscle.

51. The method of Claim 46, further comprising:delivering one or more stimulus patterns to the muscle, a stimulus pattern of the one or more stimulus patterns causing a transformation of a phenotype of the muscle toward a target phenotype, the target phenotype being associated with fatigue resistance.

52. The method of Claim 46, wherein surgically coupling the muscle extracted to the organ system comprises, using the muscle:encircling a portion of an intestine;wrapping a portion of a heart;- 79 - 5040745. vlDocket No. 4889-1034001mechanically coupling to a portion of a diaphragm or an intercostal muscle; encircling a portion of a bladder;mechanically coupling to a sphincter system;mechanically coupling to a smooth muscle system; ormechanically coupling to a skin graft.

53. The method of Claim 46, wherein the muscle is a first muscle and the method further comprises:surgically coupling at least one additional muscle and causing first muscle and the at least one additional muscle to contract in coordination to perform the physiological function of the organ.5040745. vl

Images