Prosthetic communication controller

Abstract

A system includes a prosthetic device including an actuatable joint, one or more implants, each of the one or more implants including an electrode configured to detect electromyography (EMG) signals and positional data from an Inertial Measurement Unit (IMU), and a communication controller including a hardware processor, a memory for storing instructions, and a wireless communications device, the communication controller being configured to: receive wireless signals from each of the one or more implants, the wireless signals carrying EMG signals, process the EMG signals and positional data received from each of the one or more implants or sensors, generate a prosthetic control signal based on the received EMG signals, and send the prosthetic control signal to the actuatable joint.

Description

PROSTHETIC COMMUNICATION CONTROLLERTECHNICAL FIELD

[0001] The present disclosure generally relates to controlling prosthetic devices and more specifically to processing data received from implants to control prosthetic devices.BACKGROUND

[0002] There are currently hundreds of millions of individuals worldwide with mobility impairments resulting from aging and / or physical disabilities. Robotic limb prostheses, active orthotics, and exoskeletons can help replace and / or augment the motor function of amputated or impaired biological limbs and allow users to perform daily activities that require the use of motorized orthopaedic technologies.SUMMARY

[0003] Implementations of the present disclosure are directed to controlling prosthetic devices. More particularly, implementations of the present disclosure are directed to using a communication controller for processing data received from implants to control prosthetic devices.

[0004] In some implementations, a system includes: one or more prosthetic devices including one or more actuatable joints, one or more implants, each of the one or more implants including one or more electrodes configured to detect electromyography (EMG) signals, one or more Inertial Measurement Units (IMUs) configured to detect positional data of the implants, one or more IMUs on the communication controller to detect positional data of the communication controller, one or more wireless IMUs located on the prosthetic device or exoskeleton to detect positional data of the prosthetic device or exoskeleton, and a communication controller including a hardware processor, a memory for storing instructions, and a wireless communications device, the communication controller being configured to: receive wireless signals from each of the one or more implants and IMUs, the wireless signals carrying EMG and IMU signals, process the EMG signals and IMU positional data received from each of the one or more implants and IMUs, generate a prosthetic control signal based on the received EMG and IMU signals, and send the resulting control signal to the actuatable joints.

[0005] The foregoing and other implementations can each optionally include one or more of the following features, alone or in combination. In particular, implementations can include but are not limited to the following features:

[0006] In a first aspect, combinable with any of the previous aspects, the system further includes a user device wirelessly coupled to the communication controller, the communication controller being configured to transmit, for display, to the user device, data associated with a setting of the communication controller. The communication controller processes EMG signals using a machine learning algorithms trained in a learning mode that is initiated by a user device (e.g., user’s companion mobile phone device). Some features of the learning mode include prosthetic control scenarios selected, through a user interface of the user device, presented as instructional videos, training the system to interpret synchronized received EMG signals. Machine learning training, where the EMG data is processed locally on the communication controller or on a remote server to train a machine learning model. Training data validation ensures that the training dataset improves system accuracy by subsampling rest periods,wherein the user is not performing any movement, assuring poor quality data is not added into the system.

[0007] The communication controller connects to each implant using a separate radio frequency (RF) link (optionally a Near-Field for power and a 2.4GHz Mid-Field or Far-Field Link for EMG / IMU data) and the communication controller connects to the user device, during user operations such as configuration, learning, or testing, using a Wi-Fi link (or BLE) separate from the Implant link(s). The communication controller is configured to receive, from the user device, prosthetic control scenarios for the machine learning algorithm to translate the EMG signals into machine learning coefficients that the machine learning algorithms subsequently translate to the prosthetic control signals.

[0008] The prosthetic control scenarios are selected (e.g. gestures or movements) by the user through the user’s companion mobile phone device during the learning mode. The prosthetic control scenarios include movement videos displayed on the user’s companion mobile phone device in synchronization with recording of the EMG signals on the communication controller, that are processed by the communication controller, or alternatively the remote server system, for training the machine learning algorithm to calibrate a complete trained machine learning algorithm.

[0009] The communication controller stores the trained machine learning algorithm for use during a normal control mode. During learning mode the training data is evaluated in-line (sub-sampling rest periods of the data set) to assure that the present training dataset improves the quality of the overall training dataset. In normal control mode, gesture decoding is performed using a majority voting mechanism to determine, with high certainty, the command sent to the prosthesis. During the normal control mode, the system calculates a confidence level for each gesture command. If the confidence level exceeds a predetermined threshold, the majority voting mechanism is bypassed, and the command is sent to the prosthesis with reduce latency. The user device can, at any time, scan the communication controller, and subsequently display a monitored characteristic including EMG signal quality, implant power levels, and battery level, thereby assisting the user in operation of the device.

[0010] The communication controller operates and monitors one or more of the implants using secure near field communication. The implants and the communication controller securely exchange identifiers and network credentials for wireless connections. In some implementations, a phantom key server can provide secure keys to a communication controller, which can relay the keys to connected implants via the near field connection. Thecommunication controller is configured to be connected to and powered by the prosthesis battery.

[0011] Through the near-field RF link the communication controller is configured to supply transcutaneous radio frequency power to one or more of the implants through an external power coil located overlying the receiver coil on a given implant. For implants that are under the prosthesis socket, the external power coils are applied to the inside or outside of the socket in a location overlying the location of the receiver coil on each implant. Each implant has a dedicated external power coil that powers that given implant. In some implementations, one or more implants that are distant from the communication controller, to minimize inconvenient cables, can be powered by a wireless power module that includes a battery and a near-field RF link with an integrated external power coil. The power module is placed overlying the location of the implant. The communication controller is configured to include a power link to supply power and control signals to one or more prosthetic devices.

[0012] In the setting of an exoskeleton device, the communication controller along with wireless power modules processes the EMG signals and actuates motorized joints within the exoskeleton device. In some implementations, the communication controller includes an IMU to detect the position of the communication controller. Additional wireless IMU modules wirelessly connected to the communication controller with BLE can be attached to prosthetic surfaces (e.g. back of hand) or exoskeleton surfaces to detect position data of corresponding prosthetic surfaces to optimize prosthetic response (e.g. control the position of the wrist to be within normal human expectations).

[0013] Other implementations of the aspect include corresponding systems, apparatus, and computer programs, configured to perform the actions of the methods, encoded on computer storage devices.

[0014] The present disclosure also provides a computer-readable storage medium coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations in accordance with implementations of the methods provided herein.

[0015] In some implementations, a computer implemented method includes: determining, by one or more processors, a connection status between a communication controller, a one or more of implant devices, and a prosthetic device, the one or more of implant devices being configured to record electromyography (EMG) signals, transmitting, by the one or more processors, the connection status to initiate a training of the communication controller,transmitting, by the one or more processors, the EMG signals received in response to movements displayed by a user device, receiving, by the one or more processors, a machine learning algorithm trained using the EMG signals associated to the movements displayed by a user interface, and initiating, by the one or more processors, a normal operation of the communication controller.

[0016] In some aspects, combinable with any of the previous aspects, the normal operation of the communication controller includes: receiving, by the one or more processors, a first set of the EMG signals recorded by the one or more of implant devices, and generating, by the one or more processors, by processing the first set of the EMG signals using the machine learning algorithm, a first actuating signal for the prosthetic device. The computer- implemented method further includes: receiving, by the one or more processors, a second set of EMG signals recorded by a one or more of implant devices, determining, by the one or more processors, that the second set of EMG signals corresponds to a chaotic behavior of the prosthetic device, and generating, by the one or more processors, a second actuating signal for the prosthetic device to prevent the chaotic behavior of the prosthetic device. The chaotic behavior of the prosthetic device includes a repetitive actuation of one or more actuatable joints of the prosthetic device. Processing, by the one or more processors, the first set of EMG signals includes applying a machine learning algorithm trained in a learning mode. The machine learning algorithm is trained using prosthetic control scenarios. Each of the prosthetic control scenarios includes an actuation of one or more actuatable joints of the prosthetic device in response to an EMG signal pattern. Determining, by one or more processors, that the second set of EMG signals corresponds to the chaotic behavior of the prosthetic device includes determining a similarity between a first prosthetic control scenario and a second prosthetic control scenario of the prosthetic control scenarios. Preventing the chaotic behavior of the prosthetic device includes stopping an actuation of one or more actuatable joints of the prosthetic device. The training includes adapting a machine learning model for gesture recognition in the prosthetic device, wherein adapting the machine learning model includes an adjustment of coefficients by adding additional training based on a real-time variability of userspecific signal patterns and environmental conditions. Adapting the machine learning model includes an adjustment based on user inputs flagging operation errors of the prosthetic device, the adjustment prioritizing gestures according to an accuracy and a reliability of an operation of the prosthetic device. The computer-implemented method further includes: in response to determining that a first signal breaches an integrity threshold, activating a fallback controlmode for uninterrupted device operation. The computer-implemented method further includes: cross-validating the first signal of a first type with a second signal of a second type. The computer-implemented method further includes: performing adaptive noise filtering using Fast Fourier transform-based notch filters.

[0017] The present disclosure further provides a system for implementing the methods provided herein. The system includes one or more processors, and a computer-readable storage medium coupled to the one or more processors having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations in accordance with implementations of the methods provided herein.

[0018] It is appreciated that methods in accordance with the present disclosure can include any combination of the aspects and features described herein. That is, methods in accordance with the present disclosure are not limited to the combinations of aspects and features specifically described herein, but also include any combination of the aspects and features provided.

[0019] Implementations described in the present disclosure, provide multiple technical advantages over traditional prosthetic devices. The described technology provides an artificial intelligence (Al) optimized control of prosthetic devices to enable a vast range of motions and degrees of movement freedom for neurorehabilitation (e.g., to help refine neuroplasticity -based training for recovering motor functions), assistive robotics (e.g., to command robotic systems in medical, industrial, or personal contexts), and advanced control systems (e.g., to combine decoded gestures with neural commands allows for a hybrid control model, where cognitive intent and physical gestures jointly control devices). The described prosthetic devices are controlled by a trainable communication controller that enables continuous adaptation and adjustment to optimize personalized movement functions. The training is based on gesture scenarios that enable personalized parametrization of the communication controller. The described communication controller provides a testing mode to correct detected errors and retraining to minimize deviations from user intended movements. Configurations of the communication controller enable secure data communication between the communication controller, the prosthetic device, the implant, and external devices. The described secure data communication system is designed to prevent malicious external interferences of third parties. Furthermore, the described data communication system includes a local private network around the body using low power transmitter settings to preserve power and provide compliance with absorption rate (SAR) limit restrictions. Another advantage of the described technology is thatthe communication controller allows users to activate prosthetic devices in an operational mode independent of any external devices.

[0020] The details of one or more implementations of the subject matter of the specification are set forth in the accompanying drawings and the description below. Other features, aspects, and advantages of the subject matter can become apparent from the description, the drawings, and the claims.DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, show particular aspects of the subject matter disclosed herein and, together with the description, help explain some of the principles associated with the disclosed implementations. In the drawings,

[0022] FIG. 1 is a block diagram of an example system for controlling a prosthetic device, in accordance with some example implementations;

[0023] FIG. 2 is a block diagram of an example system for controlling prosthetic devices, in accordance with some example implementations;

[0024] FIG. 3A depicts a schematic diagram illustrating an example of a training system configuration, in accordance with some example implementations;

[0025] FIG. 3B depicts a schematic diagram illustrating an example of an operational system configuration, in accordance with some example implementations;

[0026] FIG. 4 depicts a schematic diagram illustrating an example implant configuration, in accordance with some example implementations;

[0027] FIG. 5 depicts a schematic diagram illustrating an example communication controller configuration, in accordance with some example implementations;

[0028] FIG. 6A depicts a schematic diagram illustrating an example power supply system, in accordance with some example implementations;

[0029] FIG. 6B depicts a schematic diagram illustrating another example power supply system, in accordance with some example implementations;

[0030] FIG. 7A depicts a schematic flow diagram illustrating an example training architecture, in accordance with some example implementations;

[0031] FIG. 7B depicts a schematic flow diagram illustrating another example training architecture, in accordance with some example implementations;

[0032] FIG. 8 depicts a schematic flow diagram illustrating an example operational architecture, in accordance with some example implementations;

[0033] FIG. 9 depicts a schematic flow diagram illustrating an example testing architecture, in accordance with some example implementations;

[0034] FIG. 10 depicts a schematic diagram illustrating an example power module electronic architecture, in accordance with some example implementations;

[0035] FIG. 11 depicts a schematic diagram illustrating an example implant electronic architecture, in accordance with some example implementations;

[0036] FIG. 12 depicts a schematic diagram illustrating an example communication controller electronic architecture, in accordance with some example implementations;

[0037] FIG. 13 A depicts a schematic diagram illustrating an example communication port electronic architecture, in accordance with some example implementations;

[0038] FIG. 13B depicts a schematic diagram illustrating an example communication system electronic architecture, in accordance with some example implementations;

[0039] FIGS. 14A-14D illustrate different views of an electronic module of the implantable device, in accordance with some example implementations;

[0040] FIG. 15 depicts a flowchart illustrating an example process for prosthetic device control, in accordance with some example implementations; and

[0041] FIG. 16 depicts a block diagram illustrating a computing system, in accordance with some example implementations.

[0042] When practical, like labels are used to refer to same or similar items in the drawings.DETAILED DESCRIPTION

[0043] Implementations of the present disclosure are directed to techniques and tools for controlling prosthetic devices. More particularly, implementations of the present disclosure are directed to controlling prosthetic devices by using a communication controller to process data received from implants. The described implementations provide a prosthetic device including an actuatable joint, one or more implants, a communication controller, one or more inertial measurement units (IMU), and a user device. Each of the implants includes an electrode configured to detect electromyography (EMG) signals and positional data from one or more IMUs.

[0044] The communication controller includes a hardware processor, a memory, a wireless communications device, one or more IMUs, and a user interface. The communication controller can also be referred to as a fusion port. The hardware processor facilitates training and execution of machine learning algorithms for prosthetic device control, according to the system configuration. The memory stores training data, machine learning models, and system configuration including instructions for prosthetic device control. The wireless communications device for the implant includes near-field communication (NFC), mid-field communication (proprietary RF), and far-field RF (Wi-Fi) for communication with remote server systems. The communication controller receives, from the implants, wireless signals that include EMG signals. The communication controller receives positional data from integrated IMUs. The communication controller processes the received wireless signals to generate a prosthetic control signal that is sent to the prosthetic device to actuate one or more joints of the prosthetic device. The communication controller includes a user interface to facilitate direct interaction with the communication controller, without the use of a user device. Wireless IMUs are placed on a limb or body part to provide contextual data in addition to EMG signals to provide signals to the communication controller. The wireless IMUs are placed on the prosthetic device or exoskeleton to detect the spatial position to enhance the prosthetic control algorithms (e.g. wrist rotation).

[0045] The user device is wirelessly coupled to the communication controller. For example, the user device connects to the communication controller using Wi-Fi or Bluetooth Low Energy (BLE) during a setup of system configuration, learning, or testing. The user device facilitates training and testing of machine learning algorithms. The user device displays data associated with the settings of the communication controller and configuration of the system. For example, the user device facilitates monitoring of monitor EMG signal quality, implantpower levels, and battery status. The user device can bridge a communication between the communication controller and remote server systems.

[0046] The implementations described in the present disclosure, provide multiple technical advantages. For example, the described communication controller includes a circuitry and a software package that facilitates an artificial intelligence (Al) optimized control of prosthetic devices to generate a vast range of motions and degrees of movement freedom. Another advantage of the implementations described in the present disclosure is that the described prosthetic devices are controlled by a trainable communication controller that enables continuous adaptation and adjustment to optimize personalized movement functions. The training is based on gesture scenarios that facilitate personalized parametrization of the communication controller and personalized improved control of the prosthetic device. For instance, configuring the prosthetic device to perform particular actions in response to signals recorded by implants can lead to errors manifested as incorrect or uncontrollable repeated movements performed by the prosthetic device. The personalized control of prosthetic devices is expected to evolve, requiring continuous adjustments to correct and to refine movements of the prosthetic device. The described communication controller provides a testing mode to correct detected errors and retraining to minimize deviations from user intended movements. Configurations of the communication controller manages secure data communication between the communication controller, the prosthetic device, the implant, and external devices. The described secure data communication system is designed to prevent malicious external interferences of third parties. For example, the communication controller can be configured to prevent other devices from accessing the wireless connections. Furthermore, the described data communication system includes a local private network around the body using low power transmitter settings to preserve power and provide compliance with absorption rate (SAR) limit restrictions. Another advantage of the described technology is that the communication controller allows users to activate prosthetic devices in an operational mode independent of any external devices. Other advantages of the prosthetic device control techniques are described with reference to FIGS. 1-16.

[0047] System for Controlling a Prosthetic Device

[0048] FIG. 1 is a block diagram illustrating an example system 100 for controlling prosthetic devices in accordance with some implementations of the present disclosure. Specifically, the illustrated example system 100 includes or is communicably coupled with a communication controller 102, a user device 104, a network 106, a server system 108, aprosthetic device 110, a support system 112, an implant system 114, and a power supply system 116.

[0049] The communication controller 102 includes a hardware processor 126, a memory 128 for storing instructions, a wireless communications device (as described in detail with reference to FIG. 5), one or more IMUs 119, and a user interface 120. The hardware processor 126 facilitates training and execution of machine learning algorithms for prosthetic device control, according to the system configuration. The memory 128 stores training data, machine learning models, and system configuration including instructions for prosthetic device control. The wireless communications device for the implant includes near-field communication (NFC), mid-field communication (proprietary RF), and far-field RF (Wi-Fi) for communication with remote server systems. During operation, the communication controller 102 receives wireless signals (including EMG signals) from one or more of implants 118A, 118B, 118C of the implant system 114. The communication controller 102 receives positional data from integrated IMUs 119. The communication controller 102 processes the received wireless signals to generate a prosthetic control signal that is sent to the prosthetic device to actuate one or more joints of the prosthetic device. The communication controller 102 includes a user interface 120 to facilitate direct interaction with the communication controller 102, without the use of a user device. Wireless IMUs 119 are placed on a limb or body part to provide contextual data in addition to EMG signals to provide signals to the communication controller 102. The wireless IMUs 119 are placed on the prosthetic device or exoskeleton to detect the spatial position to enhance the prosthetic control algorithms (e.g. wrist rotation). The communication controller 102 processes the received wireless signals and generates output signals that the communication controller 102 can transmit to the user device 104 (e.g., in a training mode) or to the prosthetic device 110 (as control signals, in an operational usage mode).

[0050] The user device 104 can be communicatively coupled to the communication controller 102 and the server system 108, through the network 106. In some implementations, the network 106 can support a short-range communication network, managed by the communication controller 102, and a wide range communication network, accessible through the user device 104. The short-range range communication network can include radio frequency (RF) based network (e.g., using a 2.4 GHz RF link), Bluetooth, Wi-Fi, and / or other such transceiver modules (as described in detail with reference to FIGS. 13A and 13B). The wide range communication network can include a wireless local-area network (WLAN), a localarea network (LAN), a wide area network (WAN), the Internet, a cellular network, a telephone network or an appropriate combination thereof connecting any number of user devices 104 and server systems 108 (e.g., during a training mode of the communication controller 102 for controlling the prosthetic device 110). Data exchanged over the network 106, is transferred using any number of network layer protocols, such as Internet Protocol (IP), Multiprotocol Label Switching (MPLS), Asynchronous Transfer Mode (ATM), and Frame Relay. Furthermore, in implementations where the network 106 represents a combination of multiple sub-networks, different network layer protocols are used at each of the underlying subnetworks. In some implementations, the network 106 represents one or more interconnected internetworks, such as the public Internet.

[0051] The user device 104 can be any computing device operable to connect to or communicate in the network(s) 108 using a wireline or wireless connection. In general, the user device 104 includes an electronic computer device operable to receive, transmit, process, and store any appropriate data associated with the system 100 of FIG. 1, such as data received from the communication controller 102 and the server system 108, as described with reference to FIG. 15. The user device 104 is generally intended to encompass any client computing device such as a laptop / notebook computer, wireless data port, smart phone, personal data assistant (PDA), tablet computing device, one or more processors within these devices, or any other suitable processing device. The user device 104 includes interface(s), processor(s), memory, and a graphical user interface 120. The user device 104 can include one or more applications. The user device 104 can be configured to execute an application that allows the user device 104 to request and view content on the user device (e.g., initiate a training mode to train the communication controller 102 to control the prosthetic device 110). For example, the user device 104 can include a computer that includes an input device, such as a keypad, touch screen, or other device that can accept user information, and an output device that conveys information associated with a machine learning (ML) model 122 of the server system 108, or the user device itself, including digital data, visual information, or a GUI 120, respectively. The GUI 120 can interface with at least a portion of the system 100 for any suitable purpose, including generating a visual representation of the prosthetic control scenarios 124. Generally, the GUI 120 provides the user device 104 with an efficient and user-friendly presentation of training data provided by or communicated within the system 100 during a training mode. The GUI 120 can include multiple customizable frames or views having interactive fields, pulldown lists, and buttons operated by the user. The GUI 120 can include any suitable graphicaluser interface, such as a combination of a generic web browser, intelligent engine, and command line interface (CLI) that processes information and efficiently presents the results to the user visually. There can be any number of user devices associated with, or external to, the system 100. Additionally, there can also be one or more additional user devices external to the illustrated portion of system 100 that are capable of interacting with the system 100 using the network(s) 108. Further, the term “client,” “user device,” and “user” can be used interchangeably as appropriate without departing from the scope of the disclosure. Moreover, while user device can be described in terms of being used by a single user, the disclosure contemplates that many users can use one computer, or that one user can use multiple computers.

[0052] In the example of FIG. 1, the server system 108 is intended to represent various forms of servers including, but not limited to a web server, an application server, a proxy server, a network server, and / or a server pool. In general, server systems 108 accept requests for application services and provides such services to any number of user devices 104 (e.g., the user device 104 over the network 106). In accordance with implementations of the present disclosure, and as noted above, the server system 108 can host a solution environment that can be a cloud environment providing software applications, systems, and services that can be consumed by customers as a service. In some instances, the server system 108 can support training of the communication controller 102 to control the prosthetic device 110. The server system 108 includes a processor 126, a memory 128, and an interface. The processor 126 included in the server system 108 or the user device 104 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or another suitable component. Generally, the processor 126 included in the server system 108 (or the user device 104) executes instructions and manipulates data to perform the operations of the server system 108 or the user device 104, respectively. Specifically, the processor 126 executes the functionality required to process / send requests to perform training operations. The processor 126 can be a central processing unit (CPU), a blade, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or another suitable component. As used in the present disclosure, the term “computer” is intended to encompass any suitable processing device. For example, although FIG. 1 illustrates a single server system 108 and a single user device 104, the system 100 can be implemented using a single, standalone computing device, two or more servers 108, or multiple user devices. The server system 108 and the user device 104 can include any computer or processing device such as, forexample, a blade server, general-purpose personal computer (PC), Mac®, workstation, UNIXbased workstation, or any other suitable device. In other words, the present disclosure contemplates computers other than general purpose computers, as well as computers without conventional operating systems. Further, the server system 108 and the user device 104 can be adapted to execute any operating system or runtime environment, including Linux, UNIX, Windows, Mac OS®, Java™, Android™, iOS, BSD (Berkeley Software Distribution) or any other suitable operating system. According to one implementation, the server system 108 can also include or be communicably coupled with an e-mail server, a Web server, a caching server, a streaming data server, and / or another suitable server.

[0053] The memory 128 can include ML model 122 and prosthetic control scenarios 124 used for training and updating the training of the communication controller 102 to control the prosthetic device 110. The memory 128 can include any type of memory or database module and can take the form of volatile and / or non-volatile memory including, without limitation, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), removable media, or any other suitable local or remote memory component. The memory 128 can store various objects or data, including caches, classes, frameworks, applications, backup data, application objects, jobs, web pages, web page templates, database tables, database queries, repositories storing application data and / or dynamic information, and any other appropriate information including any parameters, variables, algorithms, instructions, rules, constraints, or references thereto associated with the purposes of the server system 108, the communication controller 102, and the user device 104, respectively.

[0054] Regardless of the particular implementation, “software” can include computer- readable instructions, firmware, wired and / or programmed hardware, or any combination thereof on a tangible medium (transitory or non-transitory, as appropriate) operable when executed to perform at least the processes and operations described herein. Indeed, each software component can be fully or partially written or described in any appropriate computer language including C, C++, Java™, JavaScript®, Visual Basic, assembler, Perl®, ABAP (Advanced Business Application Programming), ABAP OO (Object Oriented), any suitable version of 4GL, as well as others.

[0055] In some implementations, the communication controller 102 is powered by the power supply system 116. For example, the communication controller 102 is powered by an external battery pack 130 connected to a power connector 132 of the communication controller 102. The external battery pack 130 can include a primary battery comprising a prismatic cellwith built-in protection circuitry and a secondary battery comprising a lithium-ion battery. The communication controller 102 supplies power, such as transcutaneous radio frequency (RF) power, to the implants 118A, 118B, 118C of the implant system 114. For example, the communication controller 102 supplies power, through the power coil 134, to the implant(s) 118A, 118C attached to (a bottom layer of) the support system 112 (prosthesis socket), in a pocket formed within the subcutaneous or sub-adipose or subfascial anatomical planes of a subject. In some implementation, the communication controller 102 can be connected to and powered by the external battery pack 130 included in the prosthetic device 110. The communication controller 102 can connect to the prosthetic device 110 and can route power from the battery pack 130 to the prosthetic device 110, supplying the prosthetic device 110 and other components of the example system 100 with the necessary power for operation.

[0056] The communication controller 102 can also supply power to remote implants 118B (positioned at a distance greater than 2 cm from the communication controller 102) using a power module 136 including a power coil 134. The power coils 134 can be integrated within openings formed into the socket 112A that can be removably secured to a limb of the subject, directly over the implants 118A, 118C. The socket 112A can be formed from light weight material that is resistant to radially inward compression, such as thermoplastic-fiber composite materials. Some materials forming the socket 112A can include a polymer matrix of polypropylene, polyethylene terephthalate (PET), acrylic, and / or polymethylmethacrylate (PMMA). In some implementations, the thermoplastic material of the struts can include a fiber embedded within a polymer matrix, and the fiber may be formed from carbon, glass, or any other suitable material.

[0057] For implants distant from the communication controller 102, wireless power modules 136 with integrated external power coils and batteries 130 can be used. The power modules 136 minimize cable usage and can be placed over the implant locations. The power modules 136 can be positioned directly over implants 118B and held in place by an arm band 112B that can be removably secured to a portion of a limb of the subject. The powered implants can be configured to detect, by using integrated electrodes, EMG signals transmitted by respective nerves to which they are connected, condition and time stamp the detected EMG signals, position data, add internal data, and transmit the data over the short-range network 106 (Wi-Fi) to the communication controller 102. The communication controller 102 can process the data with ML models 122 that are tuned in the training mode. The ML models 122 translate the EMG signals into prosthesis commands that are stored by the communication controller102 for use during operational mode. The prosthetic device 110 can be powered either by the communication controller 102 over a power line 138 (or wireless) or by a separate battery. The prosthesis commands can be transmitted by to the communication controller 102 to the prosthetic device 110 over the power line 138 over the short-range network 106 (e.g., using 2.4 GHz RF link). During the operational mode, one or more parameters (e.g., battery level, connection quality) of the communication controller 102, of the implant system 114 and of the power supply system 116 can be displayed by the GUI 120 of the user device 104 that is connected with the communication controller 102 over the short-range network 106 (e.g., using 2.4 GHz RF link). The near-field RF link can operate from more than a meter distance between the implant and communication controller. In some implementations, in settings of high electromagnetic interference, a wired antenna can be routed from the communication controller to the location of the implant for a stable and robust RF link that can minimize the impact of electromagnetic interference on the signal quality. For example, in environments with high electromagnetic interference or when the implant is shielded by a carbon fiber socket, a wired antenna can be connected from the communication controller to the implant for a more reliable RF link.

[0058] The prosthetic device 110 can be an active prosthetic device, configured as a wearable robotic device controlled by the communication controller 102. The active prosthetic devices 110 described herein incorporate parallel mechanisms to improve the performance of the motions. The parallel mechanisms couple springs and motors in a parallel kinematically redundant arrangement to configure the prosthetic devices to optimize replication of human muscular behavior. For example, the motors 140A-140J are linked to linking members 1421- 142 J to form a kinematic chain made up of bodies connected by various joint types. The joint types include revolute joints, prismatic joints, screw-type joints, or other joint types. The joint type may further include one or more higher pair joint types, which are represented by a combination of revolute joints, prismatic joints, screw-type joints, or other joint types. The linking members 142I-142J include actuating, compliant, passive, and / or damping elements. Actuating linking members include one or more of the joints and are moved by an active component, such as a respective motor actuated by a control signal received from the communication controller 102. Compliant linking members include one or more of the joints configured as a compliant element, such as a spring and can generally be moved in association with a movement of an actuating linking member. Passive linking members can include passive joints that are independent of a controlling element, missing an associated motor. Dampinglinking members can include one or more of the joints configured to be controlled by a damping component, such as a dashpot.

[0059] During training mode, the user device 104 can be connected to the communication controller 102 with another communication link (e.g., RF link or Wi-Fi link) separate from the communication link used by the communication controller 102 to communicate with the implants 118A, 118B, 118C. The user device 104 can include prestored prosthetic control scenarios or can make a connection to the server system 108, using a cell phone link, to provide access to training mode anywhere the cell phone service is available, to access prosthetic control scenarios. To train the communication controller 102, the GUI 120 of the user device 104 displays prosthetic control scenarios 124 (e.g., movement videos) that the user attempts to execute to generate corresponding EMG signals. The corresponding EMG signals are processed by the communication controller 102 to train the ML models 122 for generating prosthetic device commands for each of the displayed prosthetic control scenarios 124. In some implementations, the corresponding EMG signals are sent to the server system 108 that is configured to use the EMG signals to train the ML models 122 for generating prosthetic device commands for each of the displayed prosthetic control scenarios 124.

[0060] In response to determining, by the communication controller 102, that training is complete, the machine learning parameters including the prosthetic device commands are stored by the communication controller 102 for use during operational mode as control signals. In the example context in which the EMG data is processed by the server system 108, in response to determining that training is complete, the user device 104 can be configured to interrupt the connection to the server system 108, which is not used during operational mode. During the operational mode, the communication controller 102 can transmit the control signals patching particular EMG signals to the prosthetic device 110. The prosthetic device 110 can actuate, in response to the received control signals, one or more motors 140A-140J of the prosthetic device 110 to perform one or more movements (e.g., a series of coordinated movements) corresponding to the prosthetic control scenarios 124.

[0061] The system incorporated additional satellite IMUs 119, which can be located on the communication controller 102, surfaces of the prosthetic device 110, or exoskeleton components. These IMUs 119 provide supplementary positional and motion data to enhance the control system’s accuracy and efficiency. One or more IMUs 119 integrated into the communication controller 102, mounted on the arm or leg can serve as an alternative to implant based IMUs 119 to reduce power consumption and minimize RF bandwidth requirements asthe IMU 119 of the communication controller 102 provides similar positional data to that of the implants. An IMU 119 mounted on the back of the prosthetic hand provide signals that can be processed by the communication controller 102 to prevent over-rotation, under-rotation, or improper alignment of the wrist. The IMU 119 mounted on the back of the prosthetic hand can be particularly beneficial for systems lacking direct positional feedback mechanisms. An IMU positioned on the exoskeleton provides critical feedback to the communication controller, facilitating the system to maintain a targeted posture or position, such as keeping the user upright during movement or stabilization tasks. The additional IMUs 1196 enhance the system’s ability to provide precise responsive control, improving both functionality and user safety.

[0062] System for Controlling Multiple Prosthetic Devices

[0063] FIG. 2 depicts a block diagram of an example system 200 for controlling prosthetic devices 210A, 210B, in accordance with some example implementations. The illustrated example system 200 can include any of the components of the example system 100, described in detail with reference to FIG.1 (e.g., a communication controller 102, a user device 104, a network 106, a server system 108, a prosthetic device 110, a support system 112, an implant system 114, and a power supply system 116) arranged in a different configuration. In particular, example system 200 includes a configuration that describes the use of multiple communication controllers 202A, 202B, 202C (e.g., similar to the communication controller 102, described in detail with reference to FIG.l), a user device 204 (e.g., similar to the user device 104, described in detail with reference to FIG.l), a network 206 (e.g., similar to the network 106, described in detail with reference to FIG.1), a server system 208 (e.g., similar to the server system 108, described in detail with reference to FIG.l), multiple prosthetic devices 210A, 210B (e.g., similar to the prosthetic device 110, described in detail with reference to FIG.l), support systems 212A, 212B (e.g., similar to the support systems 112, described in detail with reference to FIG. l), implant systems 214A, 214B (e.g., similar to the implant systems 114, described in detail with reference to FIG.l), one or more power supply systems 216A, 216B and, optionally, an exoskeleton 244. In exoskeleton applications, the communication controller 202A, 202B, 202C processes EMG signals to actuate respective motorized joints. Wireless IMUs attached to the exoskeleton 244 can provide additional positional data for more accurate control.

[0064] The prosthetic devices 210A, 210B can be configured to replace a missing limb of a subject 201 and the exoskeleton 244 can be configured to assist a movement of multipleactuatable joints 240A (e.g., elbow), 240B (e.g., wrist), 240C (e.g., finger), 240D (e.g., finger) of an existing limb of the subject 201 using a respective communication controller 202C. In some implementations, the prosthetic devices 210A, 21 OB include a robotic foot orthotic, a robotic leg orthotic, a robotic ankle orthotic, a robotic knee brace, a robotic arm brace, a robotic leg brace.

[0065] In some implementations, the exoskeleton 244 is, but is not limited to, a hip exoskeleton, a knee exoskeleton, an ankle exoskeleton, and / or a multiple joint exoskeleton. In some implementations, the exoskeleton 244 is, but is not limited to, a soft wearable robot composed of a textile. In some implementations, the exoskeleton 244 includes an external rigid structure 246A, 246B, 246C, 246D that can be attached to at least a portion of an elastic structure 248 configured to comfortably cover all or a part of the subject's body. The rigid structure 246A, 246B, 246C, 246D can include one or more sensors 248 and the muscle actuation interface 250. The sensor(s) 248 can detect electrical signals and / or other information generated by the nerves when the subj ect 201 moves or attempts to move a body area of interest. For example, the sensor(s) 248 may detect a neuronal action potential (hereinafter referred to as a “nerve signal”) generated by the subject 201. Alternatively, or additionally, one or more of the sensors 248 may detect the user's pulse rate, blood pressure, temperature, combinations thereof, muscle response, and the like. While not limiting, all or some of the sensors 248 can be configured to detect neural signals generated by the subject 201 that are simultaneously measured with EMG data and IMU signals to improve accuracy and adaptability of control decoding. The sensors 248 operate to detect a neural signal generated by the subject 201 when the subject 201 moves or attempts to move a part of his or her body by operating one or more skeletal muscles and / or muscle groups. The sensor(s) 248 can transmit the detected signals to the communication controller 202C that generates control signals for the muscle actuation interface 250. The muscle actuation interface 250 generally functions to receive actuation signals from the controller 202C and apply these actuation signals to one or more muscles / muscle groups within the body area of interest. In particular, the muscle actuation interface 250 transmits the actuation signal from the controller 202C to one or more muscles / muscle groups participating in the movement of the body region of interest, for example, through the actuation of one or more muscles. The muscle operation interface 250 can transmit electrical signals to one or more motor nerves of a muscle / muscle group participating in movement and / or stabilization of a body region of interest.

[0066] As shown in FIG. 2, the example system 200 includes multiple prosthetic devices 210A, 21 OB and an exoskeleton 244 attached to a subject 201. Each of the plurality of prosthetic devices 210A, 210B can be controlled by a respective communication controller 202A, 202B based on wireless EMG signals received from a respective set of one or more implants of the implant systems 214A, 214B. The communication controller 202 A, 202B can be configured to process the EMG signals received from the respective implants and positional data from IMUs integrated in a respective communication controller 202A, 202B, respective implants of the implant systems 214A, 214B or a respective on the prosthetic device 210A, 210B. The communication controller 202A, 202B can be configured to generate independent prosthetic control signals based on the received EMG signals and positional data and send the prosthetic control signal to a respective prosthetic device 210A, 210B, in an operational mode. In some implementations, the communication controllers 202A, 202B, 202C of the example system 200 facilitate positional awareness of the prosthetic devices 210A, 210B and / or the exoskeleton 244. A BLE-connected Inertial Measurement Unit (IMU) can be attached to the prosthetic devices 210A, 210B and / or the exoskeleton 244 to determine the respective positions and movements, either in absolute terms or relative to the respective communication controllers 202A, 202B, 202C. The IMU can be strategically placed on various parts of the prosthetic devices 210A, 210B, such as the wrist, back of the hand, or even a specific finger (e.g., the ring finger), to provide accurate motion and location data. The additional feedback enhances the overall control system’s accuracy and responsiveness, enabling more natural and intuitive prosthetic movement.

[0067] The short-range communication network can be configured to prevent signal interferences (including, but not limited to cross-talk) between implant systems 214A, 214B and non-associated communication controllers 202B, 202A as well as signal interferences between communication controllers 202 A, 202B and non-associated prosthetic devices 210B, 210A. The prevention of signal interference can include transmission channel selection, signal transmission gating based on signal frequency and / or time modulation.

[0068] Although communication controllers 202A, 202B, 202C can provide independent (non-interfering) control of respective prosthetic devices 210A, 210B or the exoskeleton 244, in some implementations, actuation of two or more prosthetic devices 210A, 210B and / or the exoskeleton 244 can be provided by a single communication controller 202A or can be coordinated by a single communication controller 202A designated as a master communication controller 202A, in a coordinated operation mode. In some implementations,the coordinated operation mode can be enabled by the subject 201 in a training mode, by selecting training using prosthetic control scenarios corresponding to the coordinated operation mode, involving actuation of a combination of prosthetic device(s) 210A, 21 OB and / or the exoskeleton 244. In the coordinated operation mode, the master communication controller is configured to process the EMG signals and to actuate at least a portion of the exoskeleton 244 (e.g., motorized braces) to execute an augmented movement synchronized with an actuation of one or more prosthetic device 210A, 21 OB. In some implementations, the power supply system 216 provides power for all communication controllers 202A, 202B, 202C or the power supply system 216 can include multiple power supply systems 216A, 216B, 216C (including multiple external battery packs) separately providing power the communication controllers 202A, 202B, 202C, to minimize wired connections between different regions of the subject 201. The communication controllers 202A, 202B, 202C can be configured to supply transcutaneous radio frequency power to one or more of their respective implant system 214A, 214B through a power coil and can include power lines to supply power to the prosthetic device 210A, 21 OB, as described in FIG. 1.

[0069] System Configuration for Training Mode

[0070] FIG. 3A depicts a schematic diagram illustrating an example of a training system configuration 300 A, in accordance with some example implementations. The training system configuration 300 A can include any of the components of the example system 100, described in detail with reference to FIG.1 (e.g., a communication controller 102, a user device 104, a network 106, a prosthetic device 110, a support system 112, an implant system 114, a power supply system 116, and, optionally, a server system 108) arranged in a different configuration. In particular, example system 300A includes a configuration that describes the setting and the training of a communication controller 302 (e.g., similar to the communication controller 102, described in detail with reference to FIG.l) that is coupled to a user device 304 (e.g., similar to the user device 104, described in detail with reference to FIG.1), a network 306 (e.g., similar to the network 106, described in detail with reference to FIG. l), a server system 308 (e.g., similar to the server system 108, described in detail with reference to FIG. l), a prosthetic device 310 (e.g., similar to the prosthetic device 110, described in detail with reference to FIG. l), IMUs 311A, 31 IB, 311C (e.g., similar to the IMUs 119, described in detail with reference to FIG.1), implants 318A, 318B, 318C (e.g., similar to the implants 118A-118C, described in detail with reference to FIG. l), a power source 330 (e.g., similar to the battery 130, described in detail with reference to FIG.l), and a power coil 334 (e.g., similar to the 1power coil 134, described in detail with reference to FIG. l). A satellite IMU 360 connects to the communication controller 302 with a near-field communication (NFC) or a mid-field communication (proprietary RF, such as Bluetooth® Low Energy link).

[0071] The training system configuration 300A can be assembled (setup) after the implants 318A, 318B, 318C are attached to a nervous system of a subject and the inflammation subsided. In the case where a prosthesis socket is required and the implants 318A, 318B, 318C are attached to a support system 312 (e.g., a socket), the support system 312 can be included in the training system configuration 300A. The support system 312 can include one or more IMUs 311A. In some implementations, one or more implants 318A, 318B and IMUs 311A, 31 IB, 311C can be away from the socket, being attached to an arm band or another retaining device to hold a respective power module 336A, 336B, 336C proximal to (approximately above) the respective implant 318A, 318B, 318C. In some implementations, one or more IMUs 31 IB, 311C can be away from the support system 312, being attached to the communication controller 302 and / or the prosthetic device 310. The controller’s IMU 31 IB and additional wireless IMUs 311C (e.g., on the back of the hand) factor into the control solution to replicate natural human movement, such as wrist positioning.

[0072] The communication controller 302 and / or the user device 304 can be used to scan each component of the training system configuration 300A including the implants 318A, 318B, 318C and IMUs 311A, 31 IB, 311C using a near field communication (NFC) of the network 306. Each component can have an NFC identifier tag for the purposes of cybersecurity and communications. For components, such as a third-party prosthetic device 110 that does not have a tag, a passive tag can be provided that identifies the device in use, in that way only approved devices are integrated within the training system configuration 300 A. Once all the unique identifiers are identified by the communication controller 302 and / or the user device 304, the communication controller 302 can store the unique identifiers. If the power coils 334 have not been inserted into the socket, the GUI of the user device 304 can display a visual guide to align and place the coils in the proper location. In addition, the GUI of the user device 304 can also support alignment of the power modules 336A, 336B, 336C held in place by arm bands.

[0073] The user device 330 can display prosthetic control scenarios that can be stored by the user device 330. The selected prosthetic control scenarios can be accompanied by instructional videos to synchronize EMG signal and position data acquisition. A machine learning training process can include locally processing, by the communication controller 302,or remotely processing, by the remote server system 308, the EMG data and the position data to train the machine learning model. The communication controller 302 can perform a training data validation. Inline validation ensures system accuracy by subsampling rest periods (when the user is not moving) to exclude poor-quality data.

[0074] For the remote processing, the training system configuration 300A can be assembled (setup) to configure the components to securely communicate with each other and ensure compatibility of the system. During remote processing, the setup of the training system configuration 300 A can be performed using a server system 308 (e.g., configured to act as a phantom key server and configurator). The server system 308 can provide a cloud service to configure the training system configuration 300 A and provide secure keys and connection (RF channels and / or Wi-Fi addresses) for the components of the training system configuration 300 A. After the unique identifiers to the server system 308 are stored by the server system 308, any firmware updates can also be performed and the channels and / or addresses can be updated.

[0075] In response to determining that the setup of the training system configuration 300A was successfully completed, a training mode can be initiated by the user device 304.The user device 304 can be connected to the communication controller 302 with a fast Wi-Fi link of the network 306. The user device 304 can make a connection to the server system 308 with a cell phone link, enabling the user to train anywhere where cell phone service is available. To train the communication controller 302, the user device 304 can present demonstration videos that the subject follows along with. The corresponding tagged EMG data is sent up the server system 308 (cloud) where the ML algorithms are trained. Once training is complete, the ML parameters are stored by the communication controller 302 for use during operational mode (also referred to as run mode). The communication controller 302 does not need connection to the server system 308 during normal control mode, as described with reference to FIG. 3B.

[0076] System Configuration for Normal Control (Operational) Mode

[0077] FIG. 3B depicts a schematic diagram illustrating an example of an operational system configuration 300B, in accordance with some example implementations. The operational system configuration 300B can include any of the components of the example system 100, described in detail with reference to FIG. 1 (e.g., a communication controller 102, a user device 104, a network 106, a server system 108, a prosthetic device 110, a support system 112, an implant system 114, and a power supply system 116) arranged in a different configuration. In particular, example operational system configuration 300B includes aconfiguration that describes the setting and the training of a communication controller 302 (e.g., similar to the communication controller 102, described in detail with reference to FIG.1) that is coupled to the user device 304 (e.g., similar to the user device 104, described in detail with reference to FIG.l), a network 306 (e.g., similar to the network 106, described in detail with reference to FIG.l), a prosthetic device 310 (e.g., similar to the prosthetic device 110, described in detail with reference to FIG. l), implants 318A, 318B, 318C (e.g., similar to the implants 118A-118C, described in detail with reference to FIG. l), a power source 330 (e.g., similar to the battery 130, described in detail with reference to FIG.l), and a power coil 334 (e.g., similar to the power coil 134, described in detail with reference to FIG.l).

[0078] In run mode once the implants 318A, 318B, 318C are powered, each implant 318A, 318B, 318C can measure EMG with their integrated electrodes, condition and time stamp the data, and transmit the collected data over a near range network (Wi-Fi) to the communication controller 302. The communication controller 302 can process the data with machine learning parameters that were generated in the training mode (described with reference to FIG. 3A). The communication controller 302 can use the machine learning parameters to translate the EMG signals to prosthetic device commands in that also control the prosthetic device 310. The communication controller 302 and prosthetic device 310 may use a common or individual power source 330 (batteries). During the run mode the user can monitor the communication controller 302 (e.g., battery level) over the user device 304 that is connected using a near range communication provided by the communication controller 302. The operational system configuration 300B does not need a connection to the server system 308 during run mode.

[0079] In run mode once the implants 318A, 318B, 318C are powered, gesture decoding is performed using a majority voting mechanism for high-certainty command determination. A confidence-based execution includes execution of prosthetic device commands based on gesture recognition confidence levels, incorporating a fallback mechanism to mitigate low-confidence misinterpretations. For example, the confidence level for each gesture command can be simultaneously calculated. If the confidence level exceeds a predetermined threshold, the communication controller 302 bypasses the majority voting mechanism, sending commands directly to the prosthetic device 310 to reduce latency and perform matching actions. For example, in exoskeleton applications, the communication controller 302 processes EMG signals to actuate motorized joints. Wireless IMUs 311Cattached to the exoskeleton provide additional positional data for more accurate movement control.

[0080] Design for Implant Communicably Coupled to a Communication Controller

[0081] FIG. 4 depicts a schematic diagram illustrating an example implant electronic design 400, in accordance with some example implementations. The example implant configuration 400 can include a receiver coil 402, a rectifier 404, a capacitor 406, a step-down converter 408, low dropout linear regulator (LDO) 410, microprocessor control unit (MCU) 412, a NFC wireless communication receiver front end (NFC-COM) 414, a Wi-Fi module 416, a Wi-Fi antenna 418, inertial measurement unit (IMU) 420, analog front end (AFE) 422, a set of monopolar epimysial electrodes (EX) 424, a return electrode (ER) 426, and an over temperature shutdown (OTS) 428.

[0082] The example implant electronic design 400 is configured for voltage stabilization within an implantable device. The receiver coil 402 can be a resonant coil that receives power from an external power coil, through impedance matching, to power of the example implant electronic design 400. The rectifier 404 converts 13.56MHz alternating current (AC) to direct current (DC) voltage that can be up to 15 volts. The capacitor 406 is charged up at a rate limited by the receiver coil impedance. The step-down converter 408 reduces Vrect to Vdigwith high efficiency. The LDO 410 regulates the Vrect voltage to Vadc to supply noise sensitive circuits in the implantable device.

[0083] The MCU 412 receives voltages from the step-down converter (or combination step-up and step-down converter) and the LDO 410. The MCU 412 is responsible for coordinating all the functions in the example implant electronic design 400, the functions including EMG signal recording. The MCU 412 notifies the NFC poller through the NFC Listener that power can increase. For example, MCU 412 can regulate power to NFC-COM 414 can preemptively increase power to avoid brownout conditions in the implant, a near field wireless communication receiver front end including a near field communication tag providing serial communications with a communication controller, through a Wi-Fi module and a Wi-Fi antenna.

[0084] The NFC-COM 414 implements NFC tag, serial communications with wearable / poller and security access. The Wi-Fi module 416 can be configured to provide WiFi communication (wireless signal transmission). The Wi-Fi antenna 418 can be configured to provide transmission of signals. The IMU 420 detects a force, an angular rate, and an orientation of the example implant electronic design 400, using a combination ofaccelerometers, gyroscopes, and in some implementations, magnetometers. The IMU 420 sends the position signals to the MCU 412. The AFE 422 can include analog amplifiers, (e.g., operational amplifiers), filters, and application-specific integrated circuits to provide a configurable and flexible electronics functional block used to interface a variety of sensor signals for the MCU 412. The EX 424 are implanted on top of muscle surface (being coupled to the epimysium of the subject) to detect EMG signals transmitted by the nerves to muscle fibers. The ER 426 is implanted proximal to the EX 424 to remove current from the patient safely. The OTS 428 is configured to provide protection for the example implant electronic design 400 by shutting down the power supply if the internal temperature exceeds a safe value.

[0085] In some implementations, the receiver coil 402 in the example implant electronic design 400 is a 13.56 MHz resonant receiver coil that receives RF power from the external power module / power coil. The receiver coil provides AC voltage to the rectifier 404 after impedance matching and EMI filtering. The rectifier 404 converts the voltage (Vrect) to a DC level with a voltage up to 15 V. The Vrect charges up a large capacitor 406 at a speed limited by the impedance of the receiver coil 402. The impedance of the receiver coil 402 can be optimized for the maximum load of the implant while running normally. The step-down converter 408 reduces the voltage of Vrect to Vdigital at a lower voltage (3.3V) but provides higher current. The capacitor 406 has a charge capacity configured to supply the high current peaks required by the Wi-Fi module 416. The step-down converter can smooth out the current, or normalizes the impedance received by the receiver coil 402 making the implant powering significantly more efficient. Vdigital supplies the power for approximately entire example implant electronic design 400, except for the sensitive analog front end 422. The low dropout linear regulator 410 supplies power to the noise sensitive components, such as the sensitive analog front end 422. The MCU 412 is powered by Vrect. The MCU 412 controls the entire implant including the WiFi 416, the AFE 422, the initial measurement unit 420 and communicates to an NFC WLC implemented in the external power module with the NFC- COM 414. The NFC-COM (7) establishes a serial link with the NFC WLC (Power Module) and implements NFC Tag functionality to be used during system configuration and for dynamic power optimization. The MCU 412 can turn on the NFC tag functions unless the NFC poller is an NFC WLC. A mobile phone NFC tag does not have enough power to allow the Wi-Fi to be powered. The example implant electronic design 400 can include a power module (NFC WLC) configured to power Wi-Fi 418, AFE 422, and the IMU 420.

[0086] The analog front end 422 can read the voltages on 16 monopolar electrodes against 1 of 2 available references for the ER 426. The MCU 412 converts the voltages to digital values and transmits the signals to the communication controller over the Wi-Fi 418. To save power the Wi-Fi 418 turns on for up to 2 mS at particular time intervals (e.g., every 20 mS) to reduce power usage. The Wi-Fi 418 has a bandwidth of about 80 Mbps and at a particular duty cycle can transmit approximately 2 Mbps. The MCU 412 is also responsible for determining the position have acceleration of the example implant electronic design 400, by reading the IMU 420. The NFC tag of a user device (e.g., user device 104, 204, 304 described with reference to FIGS. 1-3) does not have enough power to allow the Wi-Fi 418 to be powered.

[0087] Communication Controller Configuration

[0088] FIG. 5 depicts a schematic diagram illustrating an example communication controller configuration 500. The example communication controller configuration 500 includes the example communication controller 502 (e.g., communication controller 102, 202, 302 described with reference to FIGS. 1-3) relative to external components, in accordance with some example implementations. The example communication controller 502 includes an MCU 504, a DSP core 508 for running DSP algorithms, a neural core 510 for running machine learning algorithms, a power supply 510, a secure private network controller 512, a public network connection controller 514, and NFC wireless chargers (WLC) 516A, 516BB. The example communication controller electronic design 502 can be powered by the power supply 510 including a protected secondary lithium-ion battery. The communication controller 502 can route power to one or more implants 517 and a prosthetic device 518. The communication controller 502 can include a serial interface to control the prosthetic device 518. The communication controller 502 can support multiple (e.g., two) power coils 520A, 520B. The power coils 520A, 520B are connected for impedance matching with termination boards inside the FP along with a temperature sensor on the coil in the case that the coil is within the socket where heat can build up. The communication controller 502 includes multiple (e.g., two) synchronized NFC WLC 516A, 516BB to drive the power coils 520A, 520B. The power coils 520A, 520B can power implants, as described with reference to FIG. 4.

[0089] The implants 517 can be connected to the secure private network, as managed by the network controller 512, implemented on the communication controller 502. The communication controller 502 can be configured to prevent other devices to join the secure private network to assure that it is using clear channels. The communication controller 502 caninclude a RF switch to route high frequency signals through transmission paths, within allocated channels. Congested channels can increase the communication time potentially beyond the power limits of the implants 517. The communication controller 502 verifies the channels and in response to identifying interference significantly impacting the quality of communication, the communication controller 502 can establish communication with the respective implant 517 using a new channel for the transmission of a signal packet.

[0090] The communication controller 502 selectively activates the public network connection controller 514 to link to a server system, by using the user device as a communication bridge or. For example, during training mode (as described with reference to FIG. 3A), the communication controller 502 activates the 2.4 GHz RF link such that RAW EMG signals can be transmitted to the user device and the server system. The RF Link operates over distances exceeding one meter between the implant and communication controller. During operational mode (as described with reference to FIG. 3B), the communication controller 502 activates the RF port for controlling the prosthetic device 518. For increasing the security of the network, the communication controller 502 can act as a server to the user device, thereby minimizing a tampering risk of the communication controller 502.

[0091] Implant Powering Configuration

[0092] FIG. 6A depicts a schematic diagram illustrating an example implant powering configuration 600, in accordance with some example implementations. The example implant powering configuration 600 includes a power coil 600 A and a power module 600B. The example implant powering configuration 600 includes a polyamide power coil 602, a micro coax connector 604, a power supply 606, a printed circuit board (PCB) 608, and a ferrite shield 610. The example implant powering configuration 600 receives power through the power coil 600A. The power coil 600A is implemented from a printed circuit on a polyimide substrate 602. The power coil 600A can either connect to the communication controller (e.g., communication controller 102, 202, 302, 502 described with reference to FIGS. 1-3 and 5) through a micro coax cable 604 or can be integrated into the power module 600B. The power module 600B can include the power coil 600 A with the thin ferrite shield 610 to increase the efficiency of the power coil 600A and shield it from the PCB 608 and the power supply 606 (battery) that are layered into a cylindrical device covered with a housing composed of a durable material (e.g., plastic or polymer). The ferrite shield 610 can provide contact (of the limb including the implants) with a conductive (metallic) surface without disrupting an operation of the power coil 600 A. The power coil 600 A can have a temperature sensor 612configured to monitor temperature and compare the measured temperature to a safety temperature threshold to power off the example implant powering configuration 600, if a thermal dose threshold (according to CEM43 guidelines) is exceeded.

[0093] In some implementations, the example implant powering configuration 600 includes two separate coils for wireless power (e.g., the power receiver coil 600A) and wireless communications. The power receiver coil 600A is wound on a bobbin with the same outline as the PCB 608 and sits directly on it. In some implementations, the power receiver coil 600A is embedded within the PCB 608 itself. In some implementations, the power receiver coil 600A is embedded within a housing (as described with reference to FIGS. 14A- 14D). The communications coil is a smaller solenoid-style coil mounted on a location inward on the PCB 608. The implantable device uses a Near Field Magnetic Induction (NFMI) link to communicate with the wearable device. Sensor data (e.g., EMG data and / or motion data) is configured to be primarily sent from the implantable device to the wearable device over this link. Command signals and control signals are also configured to be transmitted over this link; for example, the supply voltage and current of the implantable device can be transmitted to the wearable device, and the wearable device can update settings for the wireless power transmitter over this link. Additionally, the wearable device is configured to transmit data to the implantable device over the NFMI link. In some implementations, data is transferred directly over the power link (via a radiofrequency modulation scheme). In some implementations, communication between the wearable device and the implantable device is accomplished via other suitable methods including, but not limited to, methods using galvanic, capacitive, ultrasound, optical, and molecular components.

[0094] In some implementations, the implantable device is powered over a wireless power system using a magnetic link. In some implementations, there is no significant energy storage on the implantable device; thus, the wireless link is configured to be on constantly while the system is in use. In some implementations, the output voltage of the power receiver coil 600A is rectified and smoothed, resulting in an unregulated voltage from which all other power supplies are generated. In some implementations, the electronics module further includes an integrated current, voltage, and power measurement circuit configured to measure the voltage received by the implantable device and the current drawn by it. In some implementations, measuring the voltage received and the current drawn enables an alignment assistance function of the wearable device and closed-loop power control, if necessary. In some implementations, the electronic module further includes a MCU (e.g., MCU 504described with reference to FIG. 5) configured to capture data from an analogue front-end and forward it to the NFMI chip, along with system configuration and monitoring functions. In some implementations, the MCU is a part of the NFMI chip. In some implementations, the MCU is a component that is separate from the NFMI chip.

[0095] In some implementations, the electronic module further includes an analog front end in order to perform analog signal processing such as filtering, noise reduction, and / or digitization of the signals. For example, signal processing can include adaptive noise filtering including identification and removal of narrowband noise within the EMG frequency range that can interfere with the classifier’s accuracy and reliability. To mitigate the interference, the electronic module can detect, using notch filters and / or adaptive algorithms, the noise signals either during startup calibration or in real time. In response to noise identification, the noise can be selectively eliminated using Fast Fourier transform-based notch filters, which can accurately target and suppress the unwanted frequencies. By default, the electronic module removes 50 / 60 Hz powerline interference and its harmonics, ensuring optimized signal processing and improved classifier performance. In some implementations, the electronics module further includes anti-aliasing circuits and / or buffers, multiplexers, and averaging circuits. In some implementations, the electronics module may include additional components for digital signal processing.

[0096] FIG. 6B depicts a schematic diagram illustrating an example implant powering configuration 600, in accordance with some example implementations. The example implant powering configuration 600 includes a power coil 600A.

[0097] Example Training Architecture Coupling a Communication Controller to a User Device

[0098] FIG. 7A depicts a schematic flow diagram 700A illustrating an example training architecture 700A, in accordance with some example implementations. The schematic flow diagram 700A can be performed using a training system configuration 300A, as described with reference to FIG. 3A. The schematic flow diagram 700A illustrates a training process performed using a communication controller 702 (including or similar to the communication controller 102, 202, 302 described with reference to FIGS. 1-3) and a user device 704 (including or similar to the user device 104, 204, 304 described with reference to FIGS. 1-3). The communication controller 702 includes a training session module 706 configured to receive signals from the implants and to process the signals. The training session module 706 includes an EMG data receiver 708 configured to receive EMG data from the implants, an IMUdata receiver 710 configured to receive IMU data from the implants, a tagging processor 710 configured to tag received data, a training file module 712 configure to generate training files including training parameters, a ML training algorithm 714 configured to be trained in recognizing actions and gestures associated to the received data, and a ML parametrization file module 716 configured to generate control signals for the prosthetic device. The user device 704 includes a gesture module 718 that includes a gesture video module 720, a gesture data file module 722, and a test result module 724. Each gesture, stored by the gesture module 718, includes of a gesture video that is performed with a particular arm position and force. The communication controller 702 can support multiple gestures that are associated to respective EMG data stored by the EMG data receiver 708 and are also associated to respective IMU data stored by the IMU data receiver 710. The test results stored by the test result module 724 are accumulated for each gesture the user performs and accepts.

[0099] The training files, stored by the training file module 712, correspond to respective training sessions. The training files can provide a mapping between raw EMG data with gesture title, position, force and the IMU data. All the training files can be run sequential through the ML training algorithm to produce a ML parameterization file that includes resulting coefficients and a model accuracy. A record of the training files can be stored by the training file module 712, for a limited amount of time so that particular data can be ‘edited’ separate from the ML parameterization file, stored by the ML parametrization file module 716, by rerunning selected set of training files.

[0100] Example Training Architecture Coupling a Communication Controller to Implants

[0101] FIG. 7B depicts a schematic flow diagram illustrating an example training architecture 700B, in accordance with some example implementations. The example training architecture 700B can include at least a portion of the example training architecture 700A, described with reference to FIG. 7A, such as the communication controller 702 (including or similar to the communication controller 102, 202, 302 described with reference to FIGS. 1-3). The example training architecture 700B shows the communication controller 702 coupled to implants 730 A, 730B. The communication controller 702 includes multiple components (software packages and / or hardware modules) configured to operate in training mode. For example, the communication controller 702 includes a deserializer 732, a crosstalk filter 734, a wavelet analyzer (e.g., tunable Q-factor wavelet transform (TQWT) module) 736, a logging module 738, a posture interpolator 740, a posture coefficient extractor 742, a posturecoefficient validator 744, a posture spatial filter 746, ALDA coefficient extractor 748, a coefficient validator 750, a Latent Dirichlet Allocation (LDA) module 752, and a voting module 754. The implants 730A, 730B include sampler 756, a filter 758, a decimator 760, an IMU 762, and a serializer 764.

[0102] The sampler 756 can be configured to receive EMG signals and sample them using a 16 channel at 20 kHz at 14 bits. The sampler 756 sounds of the sampled signals to the filter 758 which filters the sample signals at 50 or 60 Hertz. The filtered signals are sent, by the filter 758, to the decimator 760, which decimates the filtered signals according to a set ratio (e.g., to 1 / 2 or 1 / 8). The serializer 764 receives decimated signals, from the decimator 760, and position data, from IMU 762. The serializer 764 generates serial data from the decimated signals and the position data. The serializer 764 sends the serial data to the communication controller 702.

[0103] The deserializer 732 is configured to receive signals from the implants 730A, 730B, comfort the data from the implants to serial data and send signals to the crosstalk filter 734. The crosstalk filter 734 filters the signals and sends filtered signals to the wavelet analyzer 736 that can have a set quality factor to generate a wavelet with characteristics within a set range. The wavelet can be processed by the logging module 738 to extract a mean absolute value, a slope sign change, and the waveform length. The posture interpolator 740 can be configured to interpolate data from multiple postures. The interpolated posture data can be sent, by the posture interpolator 740, to the posture coefficient extractor 742 that is configured to extract the posture coefficients. The posture coefficients are sent, by the posture coefficient extractor 742, to the posture coefficient validator 744. The posture coefficient validator 744 can be configured to validate the received posture coefficients. The posture coefficient validator 744 sends the validated posture coefficients to the posture spatial filter 746 to filter involuntary movements from intended prosthetic device movements using the mean absolute value, a slope sign change, and the waveform length received from the log RMS 738. The posture spatial filter 746 sends the filtered postures to the LDA module 752. The LDA module 752 processes the filtered postures using validated coefficients are received from the coefficient validator 752 to generate movement control scenarios. The LDA module 752 sends the movement control scenarios to the voting module 754 to identify, through majority voting, a control signal to be sent to the prosthetic device.

[0104] Each time additional training is performed before the coefficients are updated, the posture coefficients are validated with a confusion matrix. The posture coefficients can onlybe updated if the model stays the same or provides better results with higher accuracy of identifying correct control signal. For each gesture (that maybe set) the confusion matrix can be the same or can improve the new posture coefficients. Once (one) gesture is accurately generated, in response to EMG signals received by the implants 730A, 730B, the gesture may be used for posture training, triggering an improvement of the posture transform, wherein the posture coefficients are accepted.

[0105] Example Operational Architecture Coupling a Communication Controller to a User Device

[0106] FIG. 8 depicts a schematic flow diagram illustrating an example operational architecture 800, in accordance with some example implementations. The example operational architecture 800 illustrates an operational training process performed using a communication controller 802 (including or similar to the communication controller 102, 202, 302, 702 described with reference to FIGS. 1-3 A, 3B, 7A, and 7B) and a user device 804 (including or similar to the user device 104, 204, 304, 704 described with reference to FIGS. 1-3 A, 3B, 7A, and 7B). In the example operational architecture 800, the user device 804 includes ML parametrization file module 806 that stores ML parametrization files 808. The user device 804 can archive ML parameterization files to reuse previous settings.

[0107] The communication controller 802 includes a ML parametrization file receiver 810, ML inference algorithm module 812, an EMG data module 814, and a prosthesis interface 816. The communication controller 802 can be powered on, to power the implants, start EMG data collection, using the EMG data module 814. The ML parametrization file receiver 810 receives from the ML parameterization file, from the user device 804. The ML inference algorithm module 812 can execute a ML inference algorithm, to process, the EMG data using the ML parameterization file received, from the ML parametrization file receiver 810 and to determine corresponding gestures. The prosthesis interface 816 can convert the gestures to a format required by a particular prosthetic device.

[0108] Example Testing Architecture Coupling a Communication Controller to a User Device

[0109] FIG. 9 depicts a schematic flow diagram illustrating an example testing architecture 900, in accordance with some example implementations. The example testing architecture 900 illustrates a testing process performed using a communication controller 902 (including or similar to the communication controller 102, 202, 302, 702, 802 described with reference to FIGS. 1-3 A, 3B, 7A, 7B, and 8) and a user device 904 (including or similar to theuser device 104, 204, 304, 704, 804 described with reference to FIGS. 1-3A, 3B, 7A, 7B, and 8). In the example testing architecture 900, the communication controller 902 includes a gesture module 906, an EMG data module 908, ML inference algorithm module 812, and a prosthesis interface 912. The user device 904 is shown to include a user interface 916.

[0110] The user device 904 can receive, through the user interface 916, a user input indicating a request to activate a testing process. The user input can include a selection of a gesture to be tested, triggering the test mode. The ML inference algorithm module 910 can execute a ML inference algorithm, to process, the EMG data corresponding to the gesture, retrieved from the gesture module 906 and to perform gesture tests. The prosthesis interface 912 can convert the tested gestures to a format required by a particular prosthetic device. The ML inference algorithm module 910 can also process arbitrary gestures that are labeled by user inputs received, through the user interface 916. Before the ML inference algorithm can be used to process and classify incoming data, the ML inference algorithm is trained. In some implementations, training of the ML inference algorithm initially takes place on a user device connected to a server system (as described with reference to FIG. 3 A). Once the ML inference algorithm was successfully trained, the trained algorithm is configured to process the input data in real-time, to control a peripheral device through the prosthesis interface 912. The trained algorithm is configured to receive data and produce control outputs for the peripheral device, such as, but not limited to, continuous joint angles, discrete gestures, or other control parameters. In some implementations, the peripheral device is a prosthetic device, an exoskeleton, an orthotic, and / or an exosuit. In some implementations, the prosthesis is, but is not limited to, a robotic limb prosthesis (e.g., a robotic arm or leg prosthesis), a robotic hand prosthesis, and / or a robotic foot prosthesis. In some implementations, the exoskeleton is, but is not limited to, a hip exoskeleton, a knee exoskeleton, an ankle exoskeleton, and / or a multiple joint exoskeleton. In some implementations, the orthotic is, but is not limited to, a robotic foot orthotic, a robotic leg orthotic, a robotic ankle orthotic, a robotic knee brace, a robotic arm brace, a robotic leg brace. In some implementations, the exosuit is, but is not limited to, a soft wearable robot composed of a textile. In some implementations, the exosuit excludes an external rigid structure.

[0111] Example Power Module Electronic Architecture

[0112] FIG. 10 depicts a schematic diagram illustrating an example power module electronic architecture 1000, in accordance with some example implementations. The example power module electronic architecture 1000 includes a NFC wireless charging transmitter 1002,receiver paths 1004A, 1004B, an electromagnetic compatibility (EMC) filter 1006, a matching circuit 1008, an antenna coil 1010, variable wired connection 1012, a power coil 1014 (similar to power coil 600 A, described with reference to FIGS. 6 A and 6B) and a temperature sensor 1016. The NFC wireless charging transmitter 1002 includes an electronic device (e.g., CTN730) that is used to transfer electric power wirelessly to the wireless charging receiver paths 1004 A, 1004B.

[0113] The process to matching the antenna coil 1010 to the NFC wireless charging transmitter 1002 is enabled by the matching circuit 1008 in a reader mode or a card mode. The NFC wireless charging transmitter 1002 can select the reader mode matching when the example power module electronic architecture 1000 is on reader mode and the card mode matching when the example power module electronic architecture 1000 is on card mode. The matching procedure can include determining characteristics of the antenna coil 1010, determining a filter cutoff frequency of the EMC filter 1006, determining the matching circuit 1008 between the antenna coil 1010 and the EMC filter 1006 for reader mode, adapting the matching for card mode, and determining the reception block. The EMC filter 1006 and the matching circuit 1008 can transform the antenna impedance Zmatch(f) to the required matching resistance Rmatch at the operating frequency of f =13.56 MHz. The current transmitted to the power coil 1014 can be regulated by the temperature sensor 1016 to ensure safety of the example power module electronic architecture 1000. For example, the temperature sensor 1016 can be configured to monitor temperature and compare the measured temperature to a safety temperature threshold to power off the example power module electronic architecture 1000, if a thermal dose threshold is exceeded.

[0114] Example Implant Electronic Architecture

[0115] FIG. 11 depicts a schematic diagram illustrating an example implant electronic architecture 1100, in accordance with some example implementations. The example implant electronic architecture 1100 includes an antenna coil 1102, a matching circuit 1104, an EMC filter 1106, a rectifier 1108, a buck converter 1110, and a NFC wireless charging transmitter 1112. The NFC wireless charging transmitter 1112 includes an electronic device (e.g., CTN730) that is used to transfer electric power wirelessly between the antenna coil 1102 and the EMC filter.

[0116] The NFC wireless charging transmitter 1112 includes a NFC interface that can include a NFC forum type 2 tag, a unique 7 byte UTD, a module configured for anidentification of a chip type and supported features, and an input capacitance of approximately 50 pF.

[0117] The example implant electronic architecture 1100 includes a host interface 1114 that can be integrated in or coupled to the NFC wireless charging transmitter 1112. The host interface 1114 including a I2C target interface for a reliable and fast data transfer. The host interface 1114 provides configurable event detection pin based on open-drain implementation to synchronize pass-through data transfer. The host interface 1114 provides NFC field detection in response to a wake-up trigger received by the example implant electronic architecture 1100.

[0118] The NFC wireless charging transmitter 1112 includes a (Ik bytes) electrically erasable programmable read-only memory (EEPROM), having a 64 bytes SRAM buffer for transfer of data between NFC and 12 C interfaces with memory mirror or pass-through mode. The NFC wireless charging transmitter 1112 provides arbitration between NFC and memory access according to a set communication path. The NFC wireless charging transmitter 1112 provides data transfer including a Pass-through mode with 64 byte SRAM buffer, Multiple page read and write NFC commands for higher data throughput. The NFC wireless charging transmitter 1112 provides secure memory-access management including full, read-only, or no memory access from NFC interface, based on 32-bit password, full, read-only, or no memory access from 12C interface, and NFC disable feature.

[0119] Example Communication Controller Electronic Architecture

[0120] FIG. 12 depicts a schematic diagram illustrating an example communication controller electronic architecture 1200, in accordance with some example implementations. The example communication controller electronic architecture 1200 includes a coil 1202, a capacitor 1204, a capacitor 1206, a coil 1208, a multiplexer (MUX) 1210, coils 1212A, 1212B, capacitors 1214A, 1214B, 1216A, 1216B, 1218, an instrumentation amplifier (INA) 1220, resistors 1222, 1226, 1228, 1236, 1240, capacitances 1224, 1233, diodes 1230, 1242, a transistor 1234, a variable resistance 1238, and a processor 1244.

[0121] The processor 1244 can include a pulse width modulation (PWM), one or more digital to analog converters (DAC) and an analog to digital converter. The processor 1244 can include an analog front end (AFE) that digitizes the differential voltages present across 16 monopolar EMG channels. The AFE design aims at finding the optimum compromise for good enough performance and minimal power consumption and size.

[0122] The reference electrode for all channels is selectable among 2 available reference electrodes: one on top of the hermetic capsule and the other on top layer of the flexible array. The configuration of the example communication controller electronic architecture 1200 allows to have a backup reference electrode for reliability as well as gives the ability to evaluate in-vivo if different positioning of the reference electrode impacts signal quality significantly.

[0123] The example communication controller electronic architecture 1200 is configured to provide a minimum through per-channel to be sent to the MUX 1210, which serializes the digitization of each EMG channel. The front of the MUX 1210 can be connected to a single pole anti-aliasing filtering (LPF, with a cut off frequency of approximately 8 kHz) and a high pass filter (with a cut off frequency of approximately 15Hz) used to attenuate the DC offset and low frequency baseline drift to maximize the available dynamic range. The resistors 1212A, 1212B can be 20 kQ resistors included in a low pass filter was chosen to limit the impact of the electrode-tissue impedance on the cutoff frequency and filter mismatch between active and reference electrode

[0124] A common mode rejection ratio (CMRR) can be achieved by using minimal filtering upstream the INA 1220 and using 1% component tolerance for those passives. In order to compensate for charge injection generated by MUX 1210 switching operation, a DC bias is applied separately on each inputs of the INA 1220. A non-inverting programmable gain amplifier is included in the example communication controller electronic architecture 1200, wherein the example communication controller electronic architecture 1200 can adjust to a wide range of signal amplitudes in-vivo. The example communication controller electronic architecture 1200 can include a simple circuit to detect if an electrode pair is connected or not. A PWM DAC of the processor 1244 injects a known amplitude signal at the input of the INA 1220 (ex: 1kHz sine wave with Vpp=50mV). The INA 1220 voltage inputs is then recorded and represents a voltage divided version of the injected voltage: V ina = V inj x (40k+R_contact) / (100k+40k+R_contact) and can be used to extract the approximate electrodetissue contact impedance.

[0125] Example Communication Port Electronic Architecture

[0126] FIG. 13 A depicts a schematic diagram illustrating an example communication port electronic architecture 1300 A, in accordance with some example implementations. The example communication port electronic architecture 1300 A includes a Wi-Fi multi -protocol wireless module 1302, a Berkeley Packet Filter (BPF) 1304, an electromechanical relaylatching switch 1306, a Wi-Fi transceiver module 1308, an external oscillator 1310, and a serial communication interface 1312.

[0127] The Wi-Fi multi -protocol wireless module 1302 provides high throughput and an extended range with power-optimized performance. The Wi-Fi multi-protocol wireless module 1302 is used to communicate at up to 2 Mbps per implant with up to 4 implants. The Wi-Fi multi -protocol wireless module 1302 can transmit at 80 Mbps @ 200mA, but since a maximum of 2 Mbps is required and the Wi-Fi multi -protocol wireless module 1302 can be duty cycled the actual power required is 200 * 2 / 80 = 5mA per implant. At 5mA and if NFC has a 20% efficiency, 25mA can be used to off power, extending a 4-hour battery life lOOmAhr of capacity for each implant.

[0128] The actual data rates can be lower reducing the power further.

[0129] The BPF 1304 can be configured to analyze network traffic. The electromechanical relay latching switch 1306 can be configured to control (either switch off or amplify) a large amount of DC current flow up to 18 GHz, up to 90W, 28V. The Wi-Fi transceiver module 1308 can be configured to provide a comprehensive multi -protocol wireless connectivity solution including 802.11 a / b / g / n (2.4 / 5 GHz), dual-mode Bluetooth 5. The WiFi modules offer high throughput, extended range with power-optimized performance. The external oscillator 1310 can be configured to provide a frequency that allows to any possible device to operate within a set frequency range. The serial communication interface 1312 can be configured to provide storage and data transfer for the example communication port electronic architecture 1300 A.

[0130] FIG. 13B depicts a schematic diagram illustrating an example communication system electronic architecture 1300B, in accordance with some example implementations. The example communication system electronic architecture I 300B includes a wireless modem 1322, a security module 1324, a thread arch wireless processor 1326, a power management unit 1328, a wireless processor peripheral 1330, a memory 1332, and a host interface module 1334. The wireless modem 1322, the security module 1324, the thread arch wireless processor 1326, the power management unit 1328, the wireless processor peripheral 1330, the memory 1332, and the host interface module 1334 can be interconnected using a system bus 1336 that includes a communication system that transfers data between components of the example communication system electronic architecture 1300B.

[0131] Example Implantable Devices

[0132] FIGS. 14A-14D illustrate different views 1400A, 1400B, 1400C, MOOD (perspective, side and back views) of an electronic module of the implantable device 1400, in accordance with some example implementations.

[0133] As shown in FIGS. 14A-14D, the implantable system can include two or more implantable substrates 1403 that can work as a cohesive system where the two or more implantable substrate 1403 can wirelessly connect to, be powered by, and be recharged by two or more wearable devices. In this manner, the wearable devices can wirelessly transmit data (e.g., EMG data, motion data, and / or other types of sensor data) to and from the two or more implantable substrates 1403 via a wireless induction link, or other type of wireless communication system. In some implementations, all of the wearable devices of the system can wirelessly transmit data to and from all of the implantable substrates 1403 of the system. In some implementations, a specific wearable device in an implantable system including two or more wearable devices can be configured to wirelessly transmit data only to and from all of the implantable devices 1402 of the implantable system. For example, in some implementations, an implantable system includes first, second, and third wearable devices and first, second, and third implantable substrates, and the first wearable device is configured to wirelessly transmit data only to and from a first implantable device. The data from all of the implantable substrates 1403 present in the implantable device 1402 can be configured to be transmitted either via wireless link, or a hardwired link to the EPU 1408 where it can be processed and analyzed by one or more algorithms (e.g., a machine learning classifier or model). In some implementations, the two or more implantable substrates are in contact withtwo or more different muscles. In some implementations, the two or more implantable substrates are in contact with two or more different portions of the same muscle.

[0134] In some implementations, the implantable device 1402 is configured to be implanted in a subject, on the surface of one or more muscles of the subject. In some implementations, the muscle is a skeletal muscle or a portion thereof. In some implementations, the implantable device 1402 is implanted in the subject such that the implantable device 1402 is in direct contact with at least a portion of a fascia of the muscle, an epimysium of the muscle, a perimysium of the muscle, an endomysium of the muscle, a fascicle of the muscle, a muscle fiber, a tendon, a blood vessel of the muscle, a nerve of the muscle, or any combination thereof. In some implementations, the fascia is a deep fascia of the muscle. In some implementations, the deep fascia is an aponeurotic fascia and / or an epimysial fascia. In some implementations, the implantable device 1402 is implanted in the subject such that the implantable device 1402 is in direct contact with at least a portion of loose connective tissue of the muscle. In some implementations, the implantable device 1402 is implanted in the subject such that the implantable device 1402 is in direct contact with at least a portion of a surface of a fasciculus of the muscle.

[0135] In some implementations, the implantable device 1402 can be inserted under the skin through one or more small incisions (e.g., an incision having a length of about 0.5 centimeters (cm) to about 5 cm). For example, a small flexible camera can be placed at the tip of an insertion tool to provide the surgeon with a clear view of where the insertion tool is spatially located to ensure accuracy and safety during pocket formation through a limited number of (e.g., one or more) incisions. Once the proper implant pocket length is achieved, the implantable device 1402 can then be inserted into the implant pocket and deployed onto the surface of one or more muscles. In some implementations, the implantable device 1402 is not fixedly secured to the muscle. In some implementations, at least a portion of the implantable device 1402 can be secured in place via one or more sutures, surgical glues, or physical anchoring features of the implantable device 1402 used to fix the implantable device 1402 to the underlying or overlying tissues. In some implementations, the implantable device 1402 is configured to be sterilized (e.g., via autoclaving, gas sterilization, gamma radiation, etc.) prior to implantation.

[0136] The implantable device 1402 includes an implantable substrate 1403 and an electronic module 1404 that are configured to operatively connect to each other, for example via a connector and / or a feedthrough architecture. The implantable substrate 1403 is anelongated, generally flat substrate or strip having a proximal end and a distal end. The implantable substrate 1403 (e.g., an implantable sensor array substrate) includes one or more sensors (e.g., EMG sensors), one or more reference electrodes, and an interconnect to electrically bond the one or more sensor pads (e.g., EMG sensors or EMG electrodes) at the distal end. In some implementations, the implantable substrate does not include one or more reference electrodes and / or biasing electrodes. In some implementations, the reference electrodes are biasing electrodes. In some implementations, the implantable substrate 1403 includes one or more reference electrodes and one or more biasing electrodes. In some implementations, the sensors can be sensor pads. The electronic module 1404 includes an opposing, second mating connector (e.g., a male or female connector) or feature configured to connect to the mating portion of the first connector of the implantable substrate 1403. The electronic module 1404 can include a case 1406 that houses the electronic components. In some implementations, the electronic module may not include a case that houses the electronic components. Instead, the electronic module can include a protective coating using technologies such as Atomic Layer Deposition (ALD) or Parylene C coating.

[0137] The implantable substrate 1403 includes an electrode array 1408 having three rows and eight columns of sensors arranged in a grid configuration, for a total of twenty -four sensors. The first row and the third row of sensors can be laterally aligned while the second row of sensors can be longitudinally offset from the first and third rows. The electrode array 1408 further includes a pair of reference electrodes that are staggered between the first and second rows and a pair of reference electrodes that are staggered between the second and third rows, for a total of four reference electrodes. The reference electrodes can be configured to be used as reference and bias drive. In some implementations, the electrode connections can be reconfigured in situ. For example, in some implementations, this can be implemented using analogue switches and / or multiplexers, which are controlled by a microcontroller. In some implementations, the degree of reconfigurability depends on which exact components with suitable parameters can be sourced. For example, in some implementations, fewer electrode configuration options can be implemented with single-pole, double-throw switches compared to a full switch matrix. The center of each sensor can be about equidistant from the center of each of the neighboring sensors. In some implementations, two or more sensors are spaced equidistantly from a center of each of the two or more sensors. In some implementations, the center-to-center sensor spacing is about 10 mm. In someimplementations, the distance between the center of each sensor and the center of an immediately adjacent sensor is about 10 mm.

[0138] Alternative numbers of columns and rows may be employed. For example, in some implementations, 14 or more electrodes are distributed into multiple rows and multiple columns. Also, every row need not contain the same number of columns. For example, an implantable substrate can include a design having one or more rows that include 10 columns of electrodes while additional rows can include 14 or more rows of electrodes to provide a greater amount of electrical field resolution. The sensors 1410A, 1410B can be biocompatible, electroconductive electrodes that are configured to contact a surface of a muscle in a subcutaneous, subadipose, or subfascial area of the subject and are configured to measure an electrical biopotential of the muscle. In some implementations, the sensors 1410A, 1410B are EMG sensors. In some implementations, the electrode array includes about 14 to about 30 sensors 1410A, 141 OB. In some implementations, the electrode array includes about 35 to about 50 sensors 1410A, 1410B. In some implementations, the sensors 1410A, 1410B are platinum-iridium alloy electrodes. In some implementations, the sensors 1410A, 1410B are carbon-based electrodes. In some implementations, the sensors 1410A, 1410B are any other suitable type of biocompatible and / or bioinert metal such as titanium, or a biocompatible and / or bioinert polymer such as PEDOT (poly 3, 4-ethylenedi oxythiophene). In some implementations, the reference electrodes are platinum iridium electrodes. In some implementations, the reference electrodes are carbon-based electrodes. In some implementations, the reference electrodes are any other suitable type of biocompatible and / or bioinert metal such as titanium, or a biocompatible and / or bioinert polymer such as PEDOT (poly 3,4-ethylenedioxythiophene). In some implementations, the sensors 1410A, 1410B are configured to have an impedance ranging from about 0.4 kQ (kOhm) to about 1 MQ (e.g., about 0.4 kQ to about 0.5 kQ, about 0.4 kQ to about 0.6 kQ, about 0.4 kQ to about 0.7 kQ, about 0.4 kQ to about 0.8 kQ, about 0.4 kQ to about 0.9 kQ, or about 0.7 kQ to about 1 kQ, about 1 kQ to about 100 kQ, about 100 kQ to about 250 kQ, about 100 kQ to about 500 kQ, about 100 kQ to about 1 MQ, about 500 kQ to about 1 MQ, about 1 kQ to about 1 M Q, or about 100 kQ to about 500 kQ) at 1 kHz.

[0139] The sensors 1410A, 1410B and reference electrodes along with their wires 1412A, 1412B are embedded within the implantable substrate 1403. The implantable substrate 1403 is composed of a flexible and bioinert and / or biocompatible material. In some implementations, the implantable substrate 1403 is composed of silicone. Non-limitingexamples of materials that the implantable substrate can be composed of include polymer- based materials (such as but not limited to silicone, liquid crystal polymer, or shape memory polymer) and a thin-film substrate coated with one or more biocompatible insulators (such as but not limited to silicone-carbide, silicone-oxide, or silicone-nitride). In some implementations, the implantable substrate is configured to wrap around a muscle. In some implementations, the implantable substrate is configured to wrap around a tissue having a generally cylindrical or tubular structure (e.g., a muscle of a limb). In some implementations, the implantable substrate is configured to wrap around a circumference of one or more muscles of the subject when implanted.

[0140] The implantable substrate 1403 has a top surface and a bottom surface opposing the top surface. The top surface includes the sensors 1410A, 141 OB, and the bottom surface includes the reference electrodes. In some implementations, the sensors 1410A, 141 OB are embedded within the top surface, and the reference electrodes are embedded within the bottom surface. The top surface is configured to be in contact with the muscle of the subject and defines one or more holes to expose the sensors 1410A, 1410B, thereby facilitating sensor 1410A, 141 OB -to-muscle contact.

[0141] The implantable substrate 1403 typically has a length (e.g., in a direction extending from the proximal end of the implantable substrate to the distal end of the implantable substrate) of about 10 mm to about 300 mm and a width (e.g., extending across the lateral edges of the implantable substrate 1403 of about 10 mm to about 200 mm. The implantable substrate 1403 typically has a total thickness of about 0.5 mm to about 5 mm, providing the implantable device 1402 with a film-like substrate having increased flexibility, which may be less noticeable to the subject when the implantable device 1402 is implanted. The implantable substrate 1403 has a generally rectangular shape with rounded edges; however, the implantable substrate can have any other suitable shape. In some implementations, the implantable substrate 702 is sized to be wrapped around one or more muscles of a subject at a subcutaneous, subadipose, or subfascial depth.

[0142] Process Description Including Training Mode and Normal Operational Mode

[0143] FIG. 15 depicts a flowchart illustrating an example process 1500 for prosthetic device control for applications using large language models, in accordance with some example implementations. Referring to FIGS. 1 and 2, the process 1500 can be performed by any components of the example systems 100 and / or 200.

[0144] At 1502, a connection status between a communication controller, a one or more of implant devices, and a prosthetic device, is determined. The implant devices can be configured to record EMG signals and positional data from an inertial measurement unit (IMU). The connection status can be provided as a response to scanning and setting a communication between system components using a user device. A successfully validated connection status includes a confirmation of powering of implants and peripheral device(s) by the communication controller and receipt of EMG signals and IMU data, by the communication controller from the implant(s). A successfully validated connection status can include registering the unique identifiers of the implants and peripheral device(s).

[0145] At 1504, a successfully validated the connection status is transmitted, by the communication controller to the user device, to initiate a training of the communication controller through a training service supported by a remote server system (e.g., server system 108, 308 described with reference to FIGS. 1 and 3). The server system can provide a cloud service to configure the training system configuration and to provide secure keys and Wi-Fi addresses for the communication controller, the implants and peripheral device(s).

[0146] At 1506, the EMG signals and IMU data are received, by the communication controller from one or more of implant devices. The EMG signals can be received in response to gesture scenario displayed by a user device as an animated video. Each gesture, stored by the gesture module 718, includes of a gesture video that is performed with a particular arm position and force. The received EMG signals and IMU data are associated, using tags, to respective gesture scenarios that were displayed within a time period including a gesture scenario display time interval. The received EMG signals and IMU data including the associated tags can be stored as training files that correspond to respective training sessions. The training files can provide a mapping between raw EMG data with gesture title, position, force and the IMU data.

[0147] At 1508, EMG signals received, by the communication controller from one or more of implant devices are transmitted to the user device, which can transmit the training files to the server system that is configured to produce a ML parameterization file. The training files can be transmitted in real time, using a cellular network connection, between the user device and the server system.

[0148] At 1510, a ML parameterization file of a trained algorithm is received. The ML parameterization file is generated by using a machine learning algorithm (model) that is trained using the tagged EMG signals, IMU data, and prosthetic control scenarios. All the training filescan be run sequential through the ML training algorithm to produce a ML parameterization file that includes resulting coefficients and a model accuracy. The machine learning algorithm can be subjected to supervised pre-training, for example, to perform a mapping between EMG signals generated in response to prosthetic control scenarios displayed by a user interface of a user device. The machine learning algorithm can be fine-tuned to differentiate between EMG signals corresponding to similar prosthetic control scenarios. The machine learning algorithm can use Equation (1) below that shows that the joint probability of a sequence s = (s_l,...,s_n) of individual movements can be factorized into a product of conditional probabilities corresponding to a particular gesture. The machine learning algorithm can use the factorization to implement an efficient sampling strategy such as sequential top-k in which the machine learning model computes the probability of a movement being a subsequent movement in the movement sequence of a particular gesture before a random sampling is performed from a k quantity of the most-likely candidates of movements. The sampling may be discontinued when a maximum sequence length is reached or when a terminal movement is produced (e.g., maximum movement trajectory is reached).PO) = HF POn |S - , sn-i) (1)

[0149] The machine learning model, for example, the transformer decoder network (e.g., generative pretrained transformer 2 (GPT-2) and / or the like), may require fine-tuning in order to perform the EMG signals to prosthetic control scenario mapping. The fine-tuning may include the machine learning model performing a conditional generation of control signals given an annotated answer, received as a user input entered on a user interface of the user device. For example, during this training phase, the machine learning model may be provided a prosthetic control scenario context c along with an I quantity of movement-gesture tuples (cq, Qi), wherein the value of I may vary from context to context, atmay denote the ground truth gesture, and qt may denote the ground truth intended movement(s). For example, a context of grabbing a fragile object (e.g., glass or a cup) can be differentiated from the context of grabbing a heavy durable object (e.g., a chain or a couch). Furthermore, the length for the ground truth gesture (as a series of movements) atmay be denoted as= |q . The optimization of the machine learning model can include maximizing the likelihood Q over all contexts c and the corresponding tuple sets (af, uation (2) below:X = Ui .u{(Qi, ai), ... , (Qk, Qfc)}

[0150] wherein u may denote the context cardinality. Factorizing over all contexts c may yield Equation (3) below, where in contrast to Equation (1), conditioning may be extended by a context ckand a specific answer in the context ak j- .

[0151] In some implementations, collaborative training of the machine learning model can include adjusting the weights applied by the machine learning model when generating control signals for the prosthetic device in order to minimize the errors present in the EMG signals to prosthetic control scenario mapping. For example, the machine learning model adapts for gesture recognition in prosthetic devices by adjusting coefficients (weights) derived from additional training. The additional training can include domain adaptation, using techniques including domain adversarial training that help the machine learning model to better generate control signals across different contexts by minimizing the discrepancy between the source (training) and target (real-world) domains. The adaptive training can ne applied to match particular contexts (environmental conditions) for controlling the prosthetic devices in realtime. Real-time adjustment can be derived from continuous learning. The model can continuously learn from new data during real-time use. The continuous learning involves updating the weights based on the most recent surface electromyography (sEMG) signals to improve accuracy and responsiveness. The adaptive training method can facilitate adjustment based on user inputs that flag errors during operation. The flagged instances can be processed to determine which gestures can be incorporated into the machine learning training loop. The adaptive training process prioritizes gestures that contribute to improved model accuracy and reliability, ensuring the system evolves dynamically based on real-world user input. A feedback-driven error correction for prosthetic control can integrate user-flagged errors into subsequent training sessions, enhancing future accuracy through weighted optimization.

[0152] A record of the training files can be stored by a training file module, for a limited amount of time enabling data editing, by rerunning a selected set of training files. In some implementations, before machine learning coefficients are applied, they are validated with statistics to verify that the training set improves the quality of the training database. If the improvement level is below a set threshold, fine-tuning and retraining are repeated until a set minimum accuracy level is exceeded. The validation includes verification of signals received from sensors, such as sEMG sensors, that are continuously monitored.

[0153] Predefined thresholds (upper bounds or lower bounds) for signal quality (e.g., signal-to-noise ratio, amplitude consistency) can be set. In response to determining that the signals meet or exceed the thresholds, potential issues including sensor displacement, noise interference, or muscle fatigue can be identified. Machine learning algorithms, such as anomaly detection models, can be used to identify deviations from normal signal patterns. The machine learning models can be trained on signal data within thresholds to recognize what constitutes normal and abnormal signals.

[0154] Multiple sensors and / or redundant pathways can be used to cross-validate signal integrity (incorporating primary EMG data and secondary IMU data). If one sensor’s data is compromised, the compromised signal can be compared with data from other sensors to detect inconsistencies. In response to detecting a breach in signal integrity, an immediate switch to a fallback control mode can be initiated for uninterrupted prosthetic device operation. The fallback control mode is designed to maintain basic functionality of the prosthetic device, facilitating maintenance of the user control. For example, the fallback control mode can include a real-time error detection and a switch to a mode that uses gross muscle movements rather than fine motor control. The validation includes recalibration of the sensors and restore normal operation. The validation can include adjusting sensor positions or recalibrating signal processing algorithms to account for the detected anomalies.

[0155] At 1512, a normal operation mode is initiated, by the communication controller. The initiation of the normal operation mode can include transmitting, by the communication controller, a command signal to the user device to trigger an automatic interruption of communication coupling between the user device and the server system. During the normal operation mode, the communication controller powers the implants and the peripheral devices, as described with reference to FIGS. 3B, 8, and 9.

[0156] At 1514, a first set of EMG signals and positional data recorded by a one or more of implant devices; is received. The normal operation of the communication controller includes receiving, by the communication controller, a first set of the EMG signals and IMU data recorded by the one or more of implant devices.

[0157] At 1516, processing, by the communication controller, the EMG signals by using the ML parameterization file of the machine learning algorithm. The processing of the first set of the EMG signals and IMU data can include serializing the signals received from multiple implants, applying filters to remove background noise, and to generate filtered signals.The processing can include mapping the series of filtered signals to gesture scenarios using the ML parameterization file of the machine learning algorithm.

[0158] At 1518, control signals are generated, by the communication controller, to control the prosthetic device according to prosthetic control scenarios, each of the prosthetic control scenarios includes an actuation of one or more actuatable joints of the prosthetic device in response to an EMG signal pattern to trigger the prosthetic device to execute the series of motions defining the matched gesture scenario.

[0159] At 1520, a second set of EMG signals and IMU data recorded by a one or more of implant devices; is received, by the communication controller from the implants.

[0160] At 1522, chaotic behavior of the prosthetic device is identified, by the communication controller. The identification of the chaotic behavior includes an identification of includes a repetitive actuation of one or more actuatable joints of the prosthetic device exceeding a set threshold number of repetitive actuations during a set time interval that can exceed a time interval of an associated gesture. In some implementations, determining, by the communication controller, that the second set of EMG signals corresponds to the chaotic behavior of the prosthetic device includes determining a similarity between a first prosthetic control scenario and a second prosthetic control scenario of the prosthetic control scenarios.

[0161] At 1524, an actuating signal to prevent the chaotic behavior of the prosthetic device is generated and transmitted identified, by the communication controller to the prosthetic device. Preventing the chaotic behavior of the prosthetic device includes stopping an actuation of one or more actuatable joints of the prosthetic device. In some implementations, preventing the chaotic behavior of the prosthetic device includes shutting down (interrupting power supply to) at least a portion of the prosthetic device performing the chaotic behavior. In some implementations, preventing the chaotic behavior of the prosthetic device includes retraining the communication controller and testing the communication controller, the implants and the prosthetic device to avoid recurrent chaotic behaviors.

[0162] The example process 1500 allows configuring the control of peripheral (prosthetic) devices in a transparent way. Scenario configurations can be pre-defined by the software vendor, and are changeable by the users of the application, enabling personalization of prosthetic device control. The example process 1500 can ease access to (re-)training, testing, and operation of communication controllers through secure network connections. Furthermore, the example process 1500 allows correction of chaotic behavior, which allows automatically triggered optimization of peripheral (prosthetic) device control, continuous fine tuning of themachine learning models, and optimization of peripheral (prosthetic) device use with improved insight into actual control of peripheral (prosthetic) device use.

[0163] Example Computing System For Performing Described Processes

[0164] FIG. 16 depicts a block diagram illustrating a computing system 1600, in accordance with some example implementations. Referring to FIG. 1, the computing system 1600 can be used to implement the communication controller 102, the user device, 104, and / or any components therein.

[0165] As shown in FIG. 16, the computing system 1600 can include a processor 1610, a memory 1620, a storage device 1630, and input / output devices 1640. The processor 1610, the memory 1620, the storage device 1630, and the input / output devices 1640 can be interconnected using a system bus 1650. The processor 1610 is capable of processing instructions for execution within the computing system 1600. Such executed instructions can implement one or more components of, for example, the machine learning controller 110 and the natural language processing engine 120. In some implementations of the current subject matter, the processor 1610 can be a single-threaded processor. Alternately, the processor 1610 can be a multi -threaded processor. The processor 1610 is capable of processing instructions stored in the memory 1620 and / or on the storage device 1630 to display graphical information for a user interface provided using the input / output device 1640.

[0166] The memory 1620 is a computer readable medium such as volatile or non-volatile that stores information within the computing system 1600. The memory 1620 can store data structures representing configuration object databases, for example. The storage device 1630 is capable of providing persistent storage for the computing system 1600. The storage device 1630 can be a floppy disk device, a hard disk device, an optical disk device, or a tape device, or other suitable persistent storage means. The input / output device 1640 provides input / output operations for the computing system 1600. In some implementations of the current subject matter, the input / output device 1640 includes a keyboard and / or pointing device. In various implementations, the input / output device 1640 includes a display unit for displaying graphical user interfaces.

[0167] According to some implementations of the current subject matter, the input / output device 1640 can provide input / output operations for a network device. For example, the input / output device 1640 can include Ethernet ports or other networking ports to communicate with one or more wired and / or wireless networks (e.g., a local area network (LAN), a wide area network (WAN), the Internet).

[0168] In some implementations of the current subject matter, the computing system 1600 can be used to execute various interactive computer software applications that can be used for organization, analysis and / or storage of data in various (e.g., tabular) format (e.g., Microsoft Excel®, and / or any other type of software). Alternatively, the computing system 1600 can be used to execute any type of software applications. These applications can be used to perform various functionalities, e.g., planning functionalities (e.g., generating, managing, editing of spreadsheet documents, word processing documents, and / or any other objects), computing functionalities, or communications functionalities. The applications can include various add-in functionalities (e.g., SAP Integrated Business Planning add-in for Microsoft Excel as part of the SAP Business Suite, as provided by SAP SE, Walldorf, Germany) or can be standalone computing products and / or functionalities. Upon activation within the applications, the functionalities can be used to generate the user interface provided using the input / output device 1640. The user interface can be generated and presented to a user by the computing system 1600 (e.g., on a computer screen monitor).

[0169] One or more aspects or features of the subject matter described herein can be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs, field programmable gate arrays (FPGAs) computer hardware, firmware, software, and / or combinations thereof. These various aspects or features can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device. The programmable system or computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

[0170] These computer programs, which can also be referred to as programs, software, software applications, applications, components, or code, include machine instructions for a programmable hardware processor, and can be implemented in a high-level procedural and / or object-oriented programming language, and / or in assembly / machine language. As used herein, the term “machine-readable medium” refers to any computer program product, apparatus and / or device, such as for example magnetic discs, optical disks, memory, and Programmable Logic Devices (PLDs), used to provide machine instructions and / or data to a programmableprocessor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term “machine-readable signal” refers to any signal used to provide machine instructions and / or data to a programmable processor. The machine-readable medium can store such machine instructions non-transitorily, such as for example as would a non-transient solid-state memory or a magnetic hard drive or any equivalent storage medium. The machine-readable medium can alternatively or additionally store such machine instructions in a transient manner, such as for example, as would a processor cache or other random-access memory associated with one or more physical processor cores.

[0171] To provide for interaction with a user, one or more aspects or features of the subject matter described herein can be implemented on a computer having a display device, such as for example a cathode ray tube (CRT) or a liquid crystal display (LCD) or a light emitting diode (LED) monitor for displaying information to the user and a keyboard and a pointing device, such as for example a mouse or a trackball, by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well. For example, feedback provided to the user can be any form of sensory feedback, such as for example visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including acoustic, speech, or tactile input. Other possible input devices include touch screens or other touch-sensitive devices such as single or multi-point resistive or capacitive track pads, voice recognition hardware and software, optical scanners, optical pointers, digital image capture devices and associated interpretation software, and the like.

[0172] The preceding figures and accompanying description illustrate example processes and computer implementable techniques. The environments and systems described above (or their software or other components) may contemplate using, implementing, or executing any suitable technique for performing these and other tasks. It can be understood that these processes are for illustration purposes only and that the described or similar techniques may be performed at any appropriate time, including concurrently, individually, in parallel, and / or in combination. In addition, many of the operations in these processes may take place simultaneously, concurrently, in parallel, and / or in different orders than as shown. Moreover, processes may have additional operations, fewer operations, and / or different operations, so long as the methods remain appropriate.

[0173] In other words, although the disclosure has been described in terms of certain implementations and generally associated methods, alterations and permutations of these implementations, and methods can be apparent to those skilled in the art. Accordingly, theabove description of example implementations does not define or constrain the disclosure. Other changes, substitutions, and alterations are also possible without departing from the spirit and scope of the disclosure.

[0174] A number of implementations of the present disclosure have been described. Nevertheless, it can be understood that various modifications may be made without departing from the spirit and scope of the present disclosure. Accordingly, other implementations are within the scope of the following claims.

[0175] In view of the above-described implementations of subject matter this application discloses the following list of examples, wherein one feature of an example in isolation or more than one feature of said example taken in combination and, optionally, in combination with one or more features of one or more further examples are further examples also falling within the disclosure of this application.

[0176] Example 1. A system comprising: a prosthetic device comprising an actuatable joint, one or more implants, each of the one or more implants comprising an electrode configured to detect electromyography (EMG) signals and positional data from an Inertial Measurement Unit (IMU); and a communication controller comprising a hardware processor, a memory for storing instructions, and a wireless communications device, the communication controller being configured to: receive wireless signals from each of the one or more implants, the wireless signals carrying EMG signals, process the EMG signals and positional data received from each of the one or more implants, generate a prosthetic control signal based on the received EMG signals, and send the prosthetic control signal to the actuatable joint.

[0177] Example 2. The system of the proceeding example, wherein the communication controller connects to the one or more of implants using a RF link.

[0178] Example s. The system of any of the proceeding examples, further comprising a user device configured display prosthetic control scenarios.

[0179] Example 4. The system of any of the proceeding examples, wherein the communication controller processes the EMG signals using a machine learning algorithm trained in a learning mode according to each of the prosthetic control scenarios.

[0180] Example 5. The system of any of the proceeding examples, wherein the communication controller translates the EMG signals into machine learning coefficients that the machine learning algorithms translates to the prosthetic control signals.

[0181] Example 6. The system of any of the proceeding examples, wherein the prosthetic control scenarios comprise movement videos displayed by the user device, the movement videos indicating movement trajectories performable by at least a portion of the prosthetic device, during the learning mode.

[0182] Example 7. The system of any of the proceeding examples, wherein the communication controller records and processes the EMG signals for training the machine learning algorithm to generate a trained machine learning algorithm.

[0183] Example 8. The system of any of the proceeding examples, wherein the communication controller is configured to execute the trained machine learning algorithm for use during a normal control mode.

[0184] Example 9. The system of any of the proceeding examples, wherein the user device is used as a communication bridge to connect the communication controller to a remote server system.

[0185] Example 10. The system of any of any of the proceeding examples, wherein the communication controller is configured to transmit to the user device, for display, a monitored characteristic comprising a battery level of the communication controller during any of the learning mode and the normal control mode.

[0186] Example 11. The system of any of the proceeding examples, wherein the communication controller is configured to scan each of the one or more of implants using a near field communication enabling detection of identifiers of the one or more of implants.

[0187] Example 12. The system of any of the proceeding examples, wherein the communication controller is configured to provide radio frequency identifiers, secure keys, and channel addresses for the wireless connection to each of the implants.

[0188] Example 13. The system of any of the proceeding examples, wherein the communication controller is configured to be connected to a battery pack.

[0189] Example 14. The system of any of the proceeding examples, wherein the communication controller is configured to supply power to one or more of the one or more of implants through a power coil and is configured to comprise a power link to supply power to the prosthetic device.

[0190] Example 15. The system of any of the proceeding examples, wherein at least one of the one or more of implants is powered by the power coil.

[0191] Example 16. The system of any of the proceeding examples, wherein the power coil is integrated into a portion of a socket, proximal to a respective implant.

[0192] Example 17. The system of any of the proceeding examples, further comprising an exoskeleton module configured to process the EMG signals and to actuate motorized braces to execute a movement synchronized with an actuation of the prosthetic device.

[0193] Example 18. A system comprising: a prosthetic device attached to a subject, the prosthetic device comprising a plurality of actuatable joints, a plurality of set of implants, each set of implants being coupled to the prosthetic device, each implant of the plurality of set of implants comprising an electrode configured to detect electromyography (EMG) and positional signals; and a communication controller comprising a hardware processor, a memory for storing instructions, and a wireless communications device, the communication controller being configured to: receive wireless signals from respective one or more implants, the wireless signals carrying EMG signals, process the EMG signals received from the respective one or more implants, generate a prosthetic control signal based on the received EMG signals for one or more actuatable joints of the plurality of actuatable joints, and send the prosthetic control signal to the one or more actuatable joints.

[0194] Example 19. The system of the proceeding example, wherein the communication controller is configured to transmit to a user device wirelessly coupled, data associated with a status of the respective communication controller.

[0195] Example 20. The system of any of the proceeding examples, wherein the respective communication controller is configured to process the EMG and positional signals using a machine learning algorithm trained in a learning mode, while executing prosthetic control scenarios displayed by the user device, the machine learning algorithm being trained to translate the EMG and positional signals into the prosthetic control signals.

[0196] Example 21. The system of any of the proceeding examples, wherein the communication controller connects to the one or more of implants using a near field communication network comprising a RF link.

[0197] Example 22. The system of any of the proceeding examples, wherein the communication controller connects to each of the one or more of implants using separate RF links, by assigning a pair of communication channels per implant .

[0198] Example 23. The system of any of the proceeding examples, wherein the prosthetic control scenarios comprise movement videos displayed by the user device and recording of the EMG signals that are processed by the communication controller for training the machine learning algorithm to generate a trained machine learning algorithm.

[0199] Example 24. The system of any of the proceeding examples, wherein the communication controller stores the trained machine learning algorithm for use during a normal control mode.

[0200] Example 25. The system of any of the proceeding examples, wherein the user device is used as a communication bridge, to connect the communication controller to a remote server .

[0201] Example 26. The system of any of the proceeding examples, wherein the communication controller is configured to transmit for display, to the user device a monitored characteristic comprising a battery level of the communication controller during any of the learning mode and the normal control mode.

[0202] Example 27. The system of any of the proceeding examples, wherein the communication controller configured to scan each of the one or more of implants using a near field communication enabling detection of identifiers of the plurality of sets of implants.

[0203] Example 28. The system of any of the proceeding examples, wherein the communication controller generates and stores the identifiers of the one or more of implants for the wireless connection.

[0204] Example 29. The system of any of the proceeding examples, comprising an exoskeleton module configured to process the EMG signals and to actuate motorized braces to execute a movement synchronized with an actuation of the prosthetic device.

[0205] Example 30. The system of any of the proceeding examples, the communication controller is configured to connect to an external battery pack.

[0206] Example 31. The system of any of the proceeding examples, wherein the communication controller is configured to supply power to one or more of the plurality of sets of implants through a power coil and comprises a power line to supply power to the prosthetic device.

[0207] Example 32. The system of any of the proceeding examples, wherein at least one of the plurality of sets of implants is powered by a power module comprising a battery and a power coil.

[0208] Example 33. The system of any of the proceeding examples, wherein the power coil is integrated into a portion of a socket, located proximal to a respective implant.

[0209] Example 34. The system of any of the proceeding examples, wherein the power coil comprises an integrated ferrite to provide contact with a conductive surface without disrupting an operation of the power coil.

[0210] Example 35. A communication controller comprising: a communication module configured to establish a wireless connection with a one or more of implants, to communicate with a prosthetic device, wherein the prosthetic device comprises a inertial measurement unit to determine a position and a movement of the prosthetic device; a battery; and one or more processors powered by the battery, configured to receive electromyography (EMG) and positional signals recorded by the one or more of implants and configured to process the EMG and positional signals to generate prosthetic control signals that actuate the prosthetic device.

[0211] Example 36. A communication controller comprising one or more processors, a memory for storing instructions, and a wireless communications device, the communication controller configured to: receive wireless signals from each of a one or more of implants; process the wireless signals received from each of the one or more of implants; generate a prosthetic control signal based on the received wireless signals; and wirelessly send the prosthetic control to an actuatable joint of a prosthetic.

[0212] Example 37. The communication controller of the proceeding example, wherein the one or more processors comprise microcontroller unit for controlling operations of the communication controller, a digital signal processor for processing the EMG signals, and a neural core for executing machine learning algorithms.

[0213] Example 38. The communication controller of any of the proceeding examples, further comprising one or more converters for conditioning a first power value from the battery into a second power value for charging the power coils.

[0214] Example 39. The communication controller of any of the proceeding examples, further comprising a near field positioned over one of the power coils configured for near field wireless charging.

[0215] Example 40. The communication controller of any of the proceeding examples, wherein the power coils are implemented from a printed circuit on a polyamide substrate.

[0216] Example 41. The communication controller of any of the proceeding examples, wherein the power coils connect to the communication controller through a micro coax cable or are integrated into a power module.

[0217] Example 42. The communication controller of any of the proceeding examples, wherein the power module comprises a power coil comprising a ferrite shield separating the power coil from a printed circuit board.

[0218] Example 43. The communication controller of any of the proceeding examples, wherein the one or more processors are configured to route power from the battery of the communication controller to the prosthetic device.

[0219] Example 44. The communication controller of any of the proceeding examples, wherein the wireless connection comprises a secure private network implemented on the communication module.

[0220] Example 45. The communication controller of any of the proceeding examples, wherein the one or more processors are configured to prevent other devices from accessing the wireless connection.

[0221] Example 46. A system configured for voltage stabilization within an implantable device, the system comprising: a receiver coil included in the implantable device comprising a resonant coil that receives power from an external power coil; a voltage rectifier configured to generate a rectified voltage by converting an alternating current to a direct current; a capacitor charging at a rate limited by a receiver coil impedance; a step-down converter reducing the rectified voltage to a digital voltage; a low dropout linear regulator regulating the rectified voltage to an analog to digital conversion voltage to supply noise sensitive circuits in the implantable device; a microprocessor control unit coordinating functions of the implantable device in response to receiving the digital voltage from the stepdown converter and the low dropout linear regulator; a microprocessor control unit that notifies the NFC Poller through the NFC Listener that Power will increase primarily due to signal packets to the NFC Poller can preemptively increase power to avoid brownout conditions in the implant; and a near field wireless communication receiver front end comprising a near field communication tag providing serial communications with a communication controller, through a RF antenna.

[0222] Example 47. The system of the proceeding example, wherein the resonant coil receives a radio frequency power from the external power coil through impedance matching.

[0223] Example 48. The system of any of the proceeding examples, wherein the capacitor provides current peaks for the Wi-Fi module.

[0224] Example 49. The system of any of the proceeding examples, wherein the stepdown converter comprises a current filter or normalizes an impedance of the receiver coil.

[0225] Example 50. The system of any of the proceeding examples, wherein the low dropout linear regulator supplies voltage for noise sensitive systems.

[0226] Example 51. The system of any of the proceeding examples, wherein the microprocessor control unit controls the RF antenna, an initial measurement unit and communicates to the near field wireless communication receiver.

[0227] Example 52. The system of any of the proceeding examples, wherein the microprocessor control unit activates near field communication tag functions in response to determining that a near field communication poller is a near field wireless communication.

[0228] Example 53. The system of any of the proceeding examples, further comprising an analog front end reading voltages on 16 monopolar electrodes against one of a pair of available references.

[0229] Example 54. The system of any of the proceeding examples, wherein the microprocessor control unit converts the voltages to digital values and transmits the digital voltage to the communication controller over the network.

[0230] Example 55. The system of any of the proceeding examples, wherein the network is active during transmission of a packet of data.

[0231] Example 56. The system of any of the proceeding examples, wherein the network has a bandwidth of approximately 80Mbps and is configured for transmitting signals from 0.5 - 2Mbps.

[0232] Example 57. A computer implemented method comprising: receiving, by one or more processors, a first set of electromyography (EMG) signals and positional data recorded by a one or more of implant devices; processing, by the one or more processors, the first set of EMG signals and the positional data to generate a first actuating signal for a prosthetic device; receiving, by the one or more processors, a second set of EMG signals recorded by a one or more of implant devices; determining, by the one or more processors, that the second set of EMG signals corresponds to a chaotic behavior of the prosthetic device; and generating, by the one or more processors, a second actuating signal for the prosthetic device to prevent the chaotic behavior of the prosthetic device.

[0233] Example 58. The computer-implemented method of the proceeding example, wherein the chaotic behavior of the prosthetic device comprises a repetitive actuation of one or more actuatable joints of the prosthetic device.

[0234] Example 59. The computer-implemented method of any of the proceeding examples, wherein processing, by the one or more processors, the first set of EMG signals comprises applying a machine learning algorithm trained in a learning mode.

[0235] Example 60. The computer-implemented method of any of the proceeding examples, wherein the machine learning algorithm is trained using prosthetic control scenarios.

[0236] Example 61. The computer-implemented method of any of the proceeding examples, wherein each of the prosthetic control scenarios comprises an actuation of one or more actuatable joints of the prosthetic device in response to an EMG signal pattern.

[0237] Example 62. The computer-implemented method of any of the proceeding examples, wherein determining, by one or more processors, that the second set of EMG signals corresponds to the chaotic behavior of the prosthetic device comprises determining a similarity between a first prosthetic control scenario and a second prosthetic control scenario of the prosthetic control scenarios.

[0238] Example 63. The computer-implemented method of any of the proceeding examples, wherein preventing the chaotic behavior of the prosthetic device comprises stopping an actuation of one or more actuatable joints of the prosthetic device.

[0239] Example 64. A computer implemented method comprising: determining, by one or more processors, a connection status between a communication controller, a one or more of implant devices, and a prosthetic device, the one or more of implant devices being configured to record electromyography (EMG) signals; transmitting, by the one or more processors, the connection status to initiate a training of the communication controller; transmitting, by the one or more processors, the EMG signals received in response to movements displayed by a user device; receiving, by the one or more processors, a machine learning algorithm trained using the EMG signals associated to the movements displayed by a user interface; and initiating, by the one or more processors, a normal operation of the communication controller.

[0240] Example 65. The computer-implemented method of the proceeding example, wherein the normal operation of the communication controller comprises: receiving, by the one or more processors, a first set of the EMG signals recorded by the one or more of implant devices; and generating, by the one or more processors, by processing the first set of the EMG signals using the machine learning algorithm, a first actuating signal for the prosthetic device.

[0241] Example 66. The computer-implemented method of any of the proceeding examples, further comprising: receiving, by the one or more processors, a second set of EMG signals recorded by a one or more of implant devices; determining, by the one or more processors, that the second set of EMG signals corresponds to a chaotic behavior of the prosthetic device; and generating, by the one or more processors, a second actuating signal for the prosthetic device to prevent the chaotic behavior of the prosthetic device.

[0242] Example 67. The computer-implemented method of any of the proceeding examples, wherein the chaotic behavior of the prosthetic device comprises a repetitive actuation of one or more actuatable joints of the prosthetic device.

[0243] Example 68. The computer-implemented method of any of the proceeding examples, wherein processing, by the one or more processors, the first set of EMG signals comprises applying a machine learning algorithm trained in a learning mode.

[0244] Example 69. The computer-implemented method of any of the proceeding examples, wherein the machine learning algorithm is trained using prosthetic control scenarios.

[0245] Example 70. The computer-implemented method of any of the proceeding examples, wherein each of the prosthetic control scenarios comprises an actuation of one or more actuatable joints of the prosthetic device in response to an EMG signal pattern.

[0246] Example 71. The computer-implemented method of any of the proceeding examples, wherein determining, by one or more processors, that the second set of EMG signals corresponds to the chaotic behavior of the prosthetic device comprises determining a similarity between a first prosthetic control scenario and a second prosthetic control scenario of the prosthetic control scenarios.

[0247] Example 72. The computer-implemented method of any of the proceeding examples, wherein preventing the chaotic behavior of the prosthetic device comprises stopping an actuation of one or more actuatable joints of the prosthetic device.

[0248] Example 73. The computer-implemented method of any of the proceeding examples, wherein the training comprises adapting a machine learning model for gesture recognition in the prosthetic device, wherein adapting the machine learning model comprises an adjustment of coefficients by adding additional training based on a real-time variability of user-specific signal patterns and environmental conditions.

[0249] Example 74. The computer-implemented method of any of the proceeding examples, wherein adapting the machine learning model comprises an adjustment based on user inputs flagging operation errors of the prosthetic device, the adjustment prioritizing gestures according to an accuracy and a reliability of an operation of the prosthetic device.

[0250] Example 75. The computer-implemented method of any of the proceeding examples, further comprising: in response to determining that a first signal breaches an integrity threshold, activating a fallback control mode for uninterrupted device operation.

[0251] Example 76. The computer-implemented method of any of the proceeding examples, further comprising: cross-validating the first signal of a first type with a second signal of a second type.

[0252] Example 77. The computer-implemented method of any of the proceeding examples, further comprising: performing adaptive noise filtering using Fast Fourier transform-based notch filters.

[0253] Example 78. A system comprising: memory storing application programming interface (API) information; and a server performing operations comprising the computer- implemented method of any of proceeding examples.

[0254] Example 79. A non-transitory computer-readable media encoded with a computer program, the computer program comprising instructions that when executed by one or more computers cause the one or more computers to perform operations comprising the computer-implemented method of any of proceeding examples.

Claims

AMENDED CLAIMS received by the International Bureau on 05 June 2025 (05.06.2025)1. A system comprising: a prosthetic device comprising an actuatable joint, one or more implants, each of the one or more implants comprising an electrode configured to detect electromyography (EMG) signals and positional data from an Inertial Measurement Unit (IMU); and a communication controller comprising a hardware processor, a memory for storing instructions, and a wireless communications device, the communication controller being configured to: receive wireless signals from each of the one or more implants, the wireless signals carrying EMG signals, process, using a machine learning algorithm, the EMG signals and positional data received from each of the one or more implants, wherein the machine learning algorithm is trained in a learning mode according to prosthetic control scenarios, generate a prosthetic control signal based on the received EMG signals, and send the prosthetic control signal to the actuatable joint.

2. The system of claim 1, wherein the communication controller connects to the one or more of implants using a RE link.

3. The system of claim 1, further comprising a user device configured display prosthetic control scenarios.

4. The system of claim 3, wherein the communication controller translates the EMG signals into machine learning coefficients that the machine learning algorithms translates to the prosthetic control signals.

5. The system of claim 4, wherein the prosthetic control scenarios comprise movement videos displayed by the user device, the movement videos indicating movement trajectories performable by at least a portion of the prosthetic device, during the learning mode.

6. The system of claim 5, wherein the communication controller records and processes the EMG signals for training the machine learning algorithm to generate a trained machine learning algorithm.

7. The system of claim 5, wherein the communication controller is configured to execute the trained machine learning algorithm for use during a normal control mode.

8. The system of claim 3, wherein the user device is used as a communication bridge to connect the communication controller to a remote server system.

9. The system of any of claim 3, wherein the communication controller is configured to transmit to the user device, for display, a monitored characteristic comprising a battery level of the communication controller during any of the learning mode and the normal control mode.

10. The system of claim 1, wherein the communication controller is configured to scan each of the one or more of implants using a near field communication enabling detection of identifiers of the one or more of implants.

11. The system of claim 1, wherein the communication controller is configured to provide radio frequency identifiers, secure keys, and channel addresses for a wireless connection to each of the implants.

12. The system of claim 1, wherein the communication controller is configured to be connected to a battery pack.

13. The system of claim 1, wherein the communication controller is configured to supply power to one or more of the one or more of implants through a power coil and is configured to comprise a power link to supply power to the prosthetic device.

14. The system of claim 13, wherein at least one of the one or more of implants is powered by the power coil.

15. The system of claim 13, wherein the power coil is integrated into a portion of a socket, proximal to a respective implant.

16. The system of any of the preceding claims, further comprising an exoskeleton module configured to process the EMG signals and to actuate motorized braces to execute a movement synchronized with an actuation of the prosthetic device.

17. A system comprising: a prosthetic device attached to a subject, the prosthetic device comprising a plurality of actuatable joints, a plurality of set of implants, each set of implants being coupled to the prosthetic device, each implant of the plurality of set of implants comprising an electrode configured to detect electromyography (EMG) and positional signals; and a communication controller comprising a hardware processor, a memory for storing instructions, and a wireless communications device, the communication controller being configured to: receive wireless signals from respective one or more implants, the wireless signals carrying EMG signals, process, using a machine learning algorithm, the EMG signals received from the respective one or more implants, wherein the machine learning algorithm is trained in a learning mode according to prosthetic control scenarios, generate a prosthetic control signal based on the received EMG signals for one or more actuatable joints of the plurality of actuatable joints, and send the prosthetic control signal to the one or more actuatable joints.

18. The system of claim 17, wherein the communication controller is configured to transmit to a user device wirelessly coupled, data associated with a status of the respective communication controller.

19. The system of claim 18, wherein the respective communication controller is configured to process the EMG and positional signals using a machine learning algorithm trained in a learning mode, while executing prosthetic control scenarios displayed by the userdevice, the machine learning algorithm being trained to translate the EMG and positional signals into the prosthetic control signals.

20. The system of claim 19, wherein the communication controller connects to the one or more of implants using a near field communication network comprising a RF link.

21. The system of claim 20, wherein the communication controller connects to each of the one or more of implants using separate RF links, by assigning a pair of communication channels per implant .

22. The system of claim 21, wherein the prosthetic control scenarios comprise movement videos displayed by the user device and recording of the EMG signals that are processed by the communication controller for training the machine learning algorithm to generate a trained machine learning algorithm.

23. The system of claim 22, wherein the communication controller stores the trained machine learning algorithm for use during a normal control mode.

24. The system of claim 21, wherein the user device is used as a communication bridge, to connect the communication controller to a remote server .

25. The system of claim 18, wherein the communication controller is configured to transmit for display, to the user device a monitored characteristic comprising a battery level of the communication controller during any of the learning mode and the normal control mode.

26. The system of claim 18, wherein the communication controller configured to scan each of the one or more of implants using a near field communication enabling detection of identifiers of the plurality of sets of implants.

27. The system of claim 26, wherein the communication controller generates and stores the identifiers of the one or more of implants for a wireless connection.

28. The system of claim 17, comprising an exoskeleton module configured to process the EMG signals and to actuate motorized braces to execute a movement synchronized with an actuation of the prosthetic device.

29. The system of claim 17, the communication controller is configured to connect to an external battery pack.

30. The system of claim 17, wherein the communication controller is configured to supply power to one or more of the plurality of sets of implants through a power coil and comprises a power line to supply power to the prosthetic device.

31. The system of claim 17, wherein at least one of the plurality of sets of implants is powered by a power module comprising a battery and a power coil.

32. The system of claim 31, wherein the power coil is integrated into a portion of a socket, located proximal to a respective implant.

33. The system of claim 31, wherein the power coil comprises an integrated ferrite to provide contact with a conductive surface without disrupting an operation of the power coil.

34. A communication controller comprising: a communication module configured to establish a wireless connection with a one or more of implants, to communicate with a prosthetic device, wherein the prosthetic device comprises a inertial measurement unit to determine a position and a movement of the prosthetic device; a battery; and one or more processors powered by the battery, configured to receive electromyography (EMG) and positional signals recorded by the one or more of implants and configured to process the EMG and positional signals to generate prosthetic control signals that actuate the prosthetic device, wherein the one or more processors comprise microcontroller unit for controlling operations of the communication controller, a digital signal processor for processing the EMG signals, and a neural core for executing machine learning algorithms.

35. A communication controller comprising one or more processors, a memory for storing instructions, and a wireless communications device, the communication controller configured to: receive wireless signals from each of a one or more of implants; process the wireless signals received from each of the one or more of implants; generate a prosthetic control signal based on the received wireless signals; and wirelessly send the prosthetic control to an actuatable joint of a prosthetic, wherein the one or more processors comprise microcontroller unit for controlling operations of the communication controller, a digital signal processor for processing the EMG signals, and a neural core for executing machine learning algorithms.

36. The communication controller of claim 34 or 35, further comprising one or more converters for conditioning a first power value from the battery into a second power value for charging power coils.

37. The communication controller of claim 36, further comprising a near field positioned over one of the power coils configured for near field wireless charging.

38. The communication controller of claim 36, wherein the power coils are implemented from a printed circuit on a polyamide substrate.

39. The communication controller of claim 36, wherein the power coils connect to the communication controller through a micro coax cable or are integrated into a power module.

40. The communication controller of claim 39, wherein the power module comprises a power coil comprising a ferrite shield separating the power coil from a printed circuit board.

41. The communication controller of claim 34 or 35, wherein the one or more processors are configured to route power from the battery of the communication controller to the prosthetic device.

42. The communication controller of claim 34 or 35, wherein the wireless connection comprises a secure private network implemented on the communication module.

43. The communication controller of claim 34 or 35, wherein the one or more processors are configured to prevent other devices from accessing the wireless connection.

44. A system configured for voltage stabilization within an implantable device, the system comprising: a receiver coil included in the implantable device comprising a resonant coil that receives power from an external power coil; a voltage rectifier configured to generate a rectified voltage by converting an alternating current to a direct current; a capacitor charging at a rate limited by a receiver coil impedance; a step-down converter reducing the rectified voltage to a digital voltage; a low dropout linear regulator regulating the rectified voltage to an analog to digital conversion voltage to supply noise sensitive circuits in the implantable device; a microprocessor control unit coordinating functions of the implantable device in response to receiving the digital voltage from the step-down converter and the low dropout linear regulator; a microprocessor control unit that notifies the NFC Poller through the NFC Listener that power increases primarily due to signal packets to the NFC Poller to avoid brownout conditions in the implant; and a near field wireless communication receiver front end comprising a near field communication tag providing serial communications with a communication controller, through a RF antenna.

45. The system of claim 44, wherein the resonant coil receives a radio frequency power from the external power coil through impedance matching.

46. The system of claim 44, wherein the capacitor provides current peaks for the Wi-Fi module.

47. The system of claim 44, wherein the step-down converter comprises a current filter or normalizes an impedance of the receiver coil.

48. The system of claim 44, wherein the low dropout linear regulator supplies voltage for noise sensitive systems.

49. The system of claim 44, wherein the microprocessor control unit controls the RE antenna, an initial measurement unit and communicates to the near field wireless communication receiver.

50. (Original) The system of claim 44, wherein the microprocessor control unit activates near field communication tag functions in response to determining that a near field communication poller is a near field wireless communication.

51. The system of claim 44, further comprising an analog front end reading voltages on 16 monopolar electrodes against one of a pair of available references.

52. The system of claim 51, wherein the microprocessor control unit converts the voltages to digital values and transmits the digital voltage to the communication controller over a network.

53. The system of claim 52, wherein the network is active during transmission of a packet of data.

54. The system of claim 52, wherein the network has a bandwidth of approximately 80Mbps and is configured for transmitting signals from 0.5 - 2Mbps.

55. A computer-implemented method comprising: receiving, by one or more processors, a first set of electromyography (EMG) signals and positional data recorded by a one or more of implant devices; processing, by the one or more processors, the first set of EMG signals and the positional data to generate a first actuating signal for a prosthetic device; receiving, by the one or more processors, a second set of EMG signals recorded by a one or more of implant devices; determining, by the one or more processors, that the second set of EMG signals corresponds to a chaotic behavior of the prosthetic device; and generating, by the one or more processors, a second actuating signal for the prosthetic device to prevent the chaotic behavior of the prosthetic device.

56. The computer-implemented method of claim 55, wherein the chaotic behavior of the prosthetic device comprises a repetitive actuation of one or more actuatable joints of the prosthetic device.

57. The computer-implemented method of claim 55, wherein processing, by the one or more processors, the first set of EMG signals comprises applying a machine learning algorithm trained in a learning mode.

58. The computer-implemented method of claim 57, wherein the machine learning algorithm is trained using prosthetic control scenarios.

59. The computer-implemented method of claim 58, wherein each of the prosthetic control scenarios comprises an actuation of one or more actuatable joints of the prosthetic device in response to an EMG signal pattern.

60. The computer-implemented method of claim 58, wherein determining, by one or more processors, that the second set of EMG signals corresponds to the chaotic behavior of the prosthetic device comprises determining a similarity between a first prosthetic control scenario and a second prosthetic control scenario of the prosthetic control scenarios.

61. The computer-implemented method of claim 55, wherein preventing the chaotic behavior of the prosthetic device comprises stopping an actuation of one or more actuatable joints of the prosthetic device.

62. A computer implemented method comprising: determining, by one or more processors, a connection status between a communication controller, a one or more of implant devices, and a prosthetic device, the one or more of implant devices being configured to record electromyography (EMG) signals; transmitting, by the one or more processors, the connection status to initiate a training of the communication controller, wherein the training comprises adapting a machine learning model for gesture recognition in the prosthetic device, wherein adapting the machine learning model comprises an adjustment of coefficients by adding additional training based on a real-time variability of user-specific signal patterns and environmental conditions and based on user inputs flagging operation errors of the prosthetic device, the adjustment prioritizing gestures according to an accuracy and a reliability of an operation of the prosthetic device; transmitting, by the one or more processors, the EMG signals received in response to movements displayed by a user device; receiving, by the one or more processors, a machine learning algorithm trained using the EMG signals associated to the movements displayed by a user interface; and initiating, by the one or more processors, a normal operation of the communication controller.

63. The computer-implemented method of claim 62, wherein the normal operation of the communication controller comprises: receiving, by the one or more processors, a first set of the EMG signals recorded by the one or more of implant devices; and generating, by the one or more processors, by processing the first set of the EMG signals using the machine learning algorithm, a first actuating signal for the prosthetic device.

64. The computer-implemented method of claim 63, further comprising: receiving, by the one or more processors, a second set of EMG signals recorded by a one or more of implant devices; determining, by the one or more processors, that the second set of EMG signals corresponds to a chaotic behavior of the prosthetic device; and generating, by the one or more processors, a second actuating signal for the prosthetic device to prevent the chaotic behavior of the prosthetic device.

65. The computer-implemented method of claim 64, wherein the chaotic behavior of the prosthetic device comprises a repetitive actuation of one or more actuatable joints of the prosthetic device.

66. The computer-implemented method of claim 63, wherein processing, by the one or more processors, the first set of EMG signals comprises applying a machine learning algorithm trained in a learning mode.

67. The computer-implemented method of claim 65, wherein the machine learning algorithm is trained using prosthetic control scenarios.

68. The computer-implemented method of claim 67, wherein each of the prosthetic control scenarios comprises an actuation of one or more actuatable joints of the prosthetic device in response to an EMG signal pattern.

69. The computer-implemented method of claim 67, wherein determining, by one or more processors, that the second set of EMG signals corresponds to the chaotic behavior of the prosthetic device comprises determining a similarity between a first prosthetic control scenario and a second prosthetic control scenario of the prosthetic control scenarios.

70. The computer-implemented method of claim 64, wherein preventing the chaotic behavior of the prosthetic device comprises stopping an actuation of one or more actuatable joints of the prosthetic device.

71. The computer-implemented method of claim 62, further comprising: in response to determining that a first signal breaches an integrity threshold, activating a fallback control mode for uninterrupted device operation.

72. The computer-implemented method of claim 71, further comprising: cross-validating the first signal of a first type with a second signal of a second type.

73. The computer-implemented method of claim 62, further comprising: performing adaptive noise filtering using Fast Fourier transform-based notch filters.

74. A system comprising: memory storing application programming interface (API) information; and a server performing operations comprising the computer-implemented method of any of claims 55-73.

75. A non-transitory computer-readable media encoded with a computer program, the computer program comprising instructions that when executed by one or more computers cause the one or more computers to perform operations comprising the computer- implemented method of any of claims 55-73.

Images