WO2025219728A1

WO2025219728A1 - Neurorehabilitation device

Info

Publication number: WO2025219728A1
Application number: PCT/GB2025/050839
Authority: WO
Inventors: Patrick Gabriel Sagastegui ALVA; Jumpei Kashiwakura; Dario Farina
Original assignee: Imperial College Innovations Ltd
Current assignee: Ip2ipo Innovations Ltd
Priority date: 2024-04-19
Filing date: 2025-04-17
Publication date: 2025-10-23
Anticipated expiration: 2026-10-19
Also published as: GB2640472A; GB202405575D0

Abstract

There is described a neurorehabilitation device comprising: one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined signals.

Description

Neurorehabilitation device

Field of the Disclosure

The present disclosure relates to a neurorehabilitation device as well as methods of manufacturing, configuring, and using a neurorehabilitation device, and systems including a neurorehabilitation device.

Background to the Disclosure

After central nervous system (CNS) damage, e.g. as may occur after a person has had a stroke or a spinal cord injury, a key aspect of rehabilitation is neuroplasticity. That is, it is important to modify the neural networks in the brain and the spinal cord to repair any neuronal pathways that have been damaged or to promote the establishment of new pathways so as to improve the person’s motor skills. In practice, a healthcare professional, such as a doctor or nurse, may conduct limb exercises and stimulation of peripheral (chemo)receptors while proactively involving a patient.

Such a process that assists in repairing neuronal pathways may be termed neurorehabilitation.

To improve the chances of recovery, it is of great importance that a target intensity and dose of activity is achieved during an initial period that follows the CNS damage, e.g. in an initial three-month recovery period following an accident (acute and semi-acute period following the damage). In practice, training occurs at a much lower dosage than needed. Reasons for this include recent increases in patient numbers (due to ageing populations) that have led to an unavailability of rehabilitation professionals as well as also the timeconsuming and labour-intensive nature of existing rehabilitation techniques. Moreover, patients may not be able to perform any movement during the first months after CNS damage and this makes active training not possible.

Summary of the Disclosure

According to an aspect of the present disclosure, there is described a neurorehabilitation device comprising: one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined (e.g. amplified) signals.

Preferably, the device comprises a multiplexer, wherein the multiplexer is arranged to receive signals from (e.g. the amplifiers of) a plurality of modular sensor units and to selectively transmit these signals to a further computer device.

Preferably, the further computer device comprises a processor of the neurorehabilitation device.

Preferably, the further computer device comprises a processor separate to (e.g. external to) the neurorehabilitation device.

Preferably, the multiplexer is arranged to selectively transmit the signals based on a communication from the further computer device. Preferably, the communication indicates one or more modular sensor units for which signals should be transmitted.

Preferably, the communication indicates an area, a muscle, and/or a muscle group for which signals should be transmitted.

Preferably, the communication indicates one or more sensor units forwhich signals should be transmitted.

Preferably, the multiplexer is arranged to selectively transmit the signals based on a feature of the signals, preferably an intensity of the signals.

Preferably, the amplifier is provided in a back-to-back arrangement with the sensors (e.g. without an intervening circuit board located between the electrode and the amplifier). Preferably, a length of electrical track that connects the electrodes and the amplifier is less than 10mm, less than 5mm, less than 1 mm, less than 0.1 mm, and/or less than 0.01 mm.

Preferably, each sensor unit comprises an actuator for providing stimulation to a muscle of the wearer.

Preferably, the device comprises a stiffener. Preferably, the sensor units and/or the sensors are located on the stiffener (e.g. so as to reduce artifacts introduced into the determined signals by the actuators).

Preferably, a connection between a sensor and an amplifier of a modular sensor unit is arranged to pass through the stiffener.

Preferably, the stimulation comprises a haptic stimulation and/or a vibration.

Preferably, the actuators are arranged to provide stimulation at a frequency of between 80 and 100 Hz. Preferably, the actuators are arranged to provide stimulation at a frequency of between 100 and 140 Hz.

Preferably, the device comprises a plurality of modular sensor units. Preferably, each modular sensor unit is connected to the multiplexer by a bus.

Preferably, the amplifier comprises a first stage and a second stage, wherein: the first amplifier stage comprises an active amplifier stage, preferably a low-noise instrumentation amplifier stage; and/or the second amplifier stage comprises an operational amplifier stage.

Preferably, the first amplifier stage and the second amplifier stage are powered by a shared power supply.

Preferably, the amplifier comprises an analogue to digital converter (ADC). Preferably, the amplifier is arranged to bias a signal entering the ADC to a mid-point of a voltage range of the ADC.

Preferably, the one or more sensors comprise one or more electrodes arranged in a monopolar configuration.

Preferably, the one or more sensors comprise one or more electrodes arranged in a bipolar configuration.

Preferably, the sensor unit comprises an actuating electrode that is arranged to both identify a muscle activity and provide a stimulation. Preferably, the actuating electrode is a part of a bipolar arrangement of electrodes that is used to reduce a crosstalk of the neurorehabilitation device.

Preferably, the actuating electrode comprises an electrode formed on a (e.g. metallic) protrusion, the protrusion being arranged to vibrate so as to provide the stimulation.

Preferably, the device comprises a plurality of sensors. Preferably, the sensors comprise electrodes.

Preferably, the device comprises a bracelet and/or a sleeve.

Preferably, the device comprises an expansion port for connecting hardware to the neurorehabilitation device.

Preferably, the sensors are arranged to record electromyography, EMG, signals and/or surface electromyography, sEMG, signals.

Preferably, the sensors are arranged to detect muscle activity over a plurality of points of a wearer so as to detect the muscle activity of a muscle.

Preferably, the device comprises one or more actuators for providing a stimulation to a muscle of the wearer. Preferably, the actuators comprise haptic feedback units and/or vibrotactile motors.

Preferably, the device comprises one or more of: a processor unit for processing sensor readings and/or stimulation signals; a communication interface for communicating with a further computer device; an amplification unit for amplifying signals from the sensors; and a housing. Preferably, the device comprises at least 4, at least 8, at least 12, and/or at least 16 recording channels. Preferably, the device comprises at least 25, at least 50, at least 70, and/or at least 100 recording channels.

Preferably, the one or more actuators are arranged to provide stimulation to the muscle in dependence on the recorded muscle activity. Preferably, the actuators are arranged to provide stimulation so as assist the muscle activity and/or the actuators are arranged to provide stimulation so as to resist the muscle activity.

Preferably, the device comprises a plurality of sensing units, wherein each sensing unit comprises: an electrode, a pre-amplifier, a band-pass filter, and a post-amplifier.

Preferably, the device comprises a plurality of electrode units. Preferably, the electrode units are formed of a rigid material, preferably polylactide, PLA.

Preferably, the device comprises a flexible material, preferably thermoplastic polyurethan, TPU. Preferably, the electrode units are mounted on the flexible material.

Preferably, the device comprises a flexible printed circuit board, PCB.

Preferably, the device comprises a plurality of modular units, wherein each unit comprises at least one sensor and at least one actuator.

Preferably, the device comprises a crosstalk reduction structure. Preferably, the crosstalk reduction structure comprises an arrangement of electrodes, preferably an arrangement of electrodes with a selective interelectrode distance between each pair of electrodes. Preferably, the device comprises an arrangement of at least four selective electrodes with different interelectrode distances between the electrodes.

Preferably, the processor is arranged to: determine a first set of signals (e.g. narrow cross talk signals) based on a first pair of electrodes; and determine a second set of signals (e.g. wide cross talk signals) based on a second pair of electrodes; wherein the distance between the second pair of electrodes is greater than the distance between the first pair of electrodes.

Preferably, a first electrode is part of both of the first pair of electrodes and the second pair of electrodes. Preferably, the processor is arranged to determine a muscle activation based on each of (e.g. a comparison between) the first set of signals and the second set of signals.

Preferably, the first pair of electrodes and the second pair of electrodes are selected from the at least four selective electrodes.

Preferably, the device comprises a processor for comparing the crosstalk between a plurality of pairs of electrodes with different interelectrode distances.

Preferably, the crosstalk reduction structure is arrange to monitor a narrow crosstalk, NCT, associated with a muscle and/or a wide cross talk, WCT, associated with an area surrounding a muscle.

According to another aspect of the present disclosure, there is described a method of determining a stimulation to provide to a wearer of a neurorehabilitation device, the method comprising: recording, using one or more sensors, a muscle activity of a wearer of the neurorehabilitation device; and determining a stimulation to provide to the user in dependence on the recorded muscle activity.

Preferably, the method comprises providing, using one or more actuators, the stimulation to the wearer.

Preferably, the method comprises recording muscle activity relating to a muscle and/or a muscle group of the wearer and providing the stimulation to said muscle and/or muscle group.

Preferably, the method comprises determining a muscle associated with the muscle activity and/or the neurorehabilitation device.

Preferably, the method comprises determining one or more features of the muscle activity. Preferably, the features comprise one or more of: an intensity, a duration, a frequency, and a length, of muscle activity. Preferably, the method comprises determining one or more of: a muscle activation, a desired muscle activation, a desired action of the wearer, and/or a gesture being performed by the wearer based on the recorded muscle activity.

Preferably, the method comprises determining a baseline muscle activity associated with the wearer.

Preferably, the method comprises determining the muscle activity and/or the stimulation using a machine learning, ML, model and/or an artificial intelligence, Al, algorithm.

Preferably, the method comprises providing feedback to the user in dependence on the muscle activity. Preferably, the feedback comprises haptic feedback.

Preferably, the method comprises determining a target muscle activity for a user. Preferably, the stimulation is determined in dependence on the target muscle activity. Preferably, the stimulation is determined in dependence on a difference between the recorded muscle activity and the target muscle activity.

Preferably, the method comprises determining a stimulation dose in dependence on the target muscle activity. Preferably, the method comprises outputting this stimulation dose. Preferably, the method comprises outputting this stimulation dose to a further computer device.

Preferably, the method comprises determining the target muscle activity in dependence on an activity history and/or a user profile of the user. Preferably, the activity history is associated with previous recorded muscle activity. Preferably, the method comprises increasing a difficulty of the target muscle activity during a neurorehabilitation process.

Preferably, the method comprises determining an amount of the muscle activity that is attributable to a voluntary movement of the user and/or determining an amount of the muscle activity that is attributable to a provided stimulation.

Preferably, the method comprises recording an updated muscle activity following the providing of the stimulation and determining an updated stimulation based on this updated muscle activity. Preferably, the method comprises continuously updating a provided stimulation based on a continuous monitoring of muscle activity.

Preferably, the method comprises initially recording a muscle activity prior to the providing of stimulation.

Preferably, the method comprises determining an association between an amount of provided stimulation and an amount of induced muscle activity.

Preferably, the method comprises determining a target voluntary muscle activity for a user. Preferably, the stimulation is determined in dependence on the target voluntary muscle activity. Preferably, the stimulation is determined in dependence on a difference between a voluntary muscle activity determined from the recorded muscle activity and the target voluntary muscle activity.

According to an aspect of the present disclosure, there is described a method of manufacturing a neurorehabilitation device, the method comprising: providing one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and providing one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined (e.g. amplified) signals.

Preferably, the method comprises selecting a number of modular sensor units in dependence on a desired size and/or function of the neurorehabilitation device.

Preferably, the method comprises connecting each of the modular sensor units to a multiplexer. According to an aspect of the present disclosure, there is described a kit of parts comprising: one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined (e.g. amplified) signals.

Preferably, the kit of parts comprises a housing for containing the sensor units and/or the actuators. Preferably, the kit of parts comprises one or more linkages for connecting and supporting the sensor units.

Any feature in one aspect of the disclosure may be applied to other aspects of the invention, in any appropriate combination. In particular, method aspects may be applied to apparatus aspects, and vice versa.

Furthermore, features implemented in hardware may be implemented in software, and vice versa. Any reference to software and hardware features herein should be construed accordingly.

Any apparatus feature as described herein may also be provided as a method feature, and vice versa. As used herein, means plus function features may be expressed alternatively in terms of their corresponding structure, such as a suitably programmed processor and associated memory.

It should also be appreciated that particular combinations of the various features described and defined in any aspects of the disclosure can be implemented and/or supplied and/or used independently.

The disclosure also provides a computer program and a computer program product comprising software code adapted, when executed on a data processing apparatus, to perform any of the methods described herein, including any or all of their component steps.

The disclosure also provides a computer program and a computer program product comprising software code which, when executed on a data processing apparatus, comprises any of the apparatus features described herein.

The disclosure also provides a computer program and a computer program product having an operating system which supports a computer program for carrying out any of the methods described herein and/or for embodying any of the apparatus features described herein.

The disclosure also provides a computer readable medium having stored thereon the computer program as aforesaid.

The disclosure also provides a signal carrying the computer program as aforesaid, and a method of transmitting such a signal.

The disclosure extends to methods and/or apparatus substantially as herein described with reference to the accompanying drawings.

The project leading to this application has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement number 810346).

Furthermore, this work was supported by the UK Medical Research Council [grant number EP/T020970/1].

The disclosure will now be described, by way of example, with reference to the accompanying drawings.

Description of the Drawings

Figures 1 a and 1 b show neurorehabilitation devices according to the present disclosure.

Figures 2a and 2b describe a method of providing sensory feedback to a wearer of a neurorehabilitation device based on muscle activity detected for that wearer. Figures 3a - 3d show modules of a system comprising a neurorehabilitation device.

Figures 4a - 4f show componentry of a system comprising a neurorehabilitation device.

Figure 5a illustrates the difference between a conventional arrangement of components of a neurorehabilitation device and an arrangement according to the present disclosure.

Figure 5b shows an arrangement of sensor elements in an embodiment of the neurorehabilitation device.

Figures 6a, 6b, 6c, and 6d show different arrangements of electrodes of a neurorehabilitation device.

Figures 7a, 7b, and 7c show embodiments of multiplexers for use with a neurorehabilitation device.

Figures 8a and 8b show forms of neurorehabilitation devices.

Description of the Preferred Embodiments

Referring to Figure 1 a, there is shown a neurorehabilitation device 1 a, 1 b, 1 c, 1d. This neurorehabilitation device is typically provided in the form of a bracelet 1 a or a sleeve 1 b. Equally, the device may be provided as a patch 1 c, a leg-worn sleeve, a cuff, a strap, a sock, an integrated part of an item of clothing, etc. In general, the device may be provided in any form that places the device adjacent to a muscle of a wearer.

Referring to Figure 1 b, the neurorehabilitation device 1 comprises one or more sensors 2, which sensors are typically arranged to record electromyography (EMG) signals and/or surface EMG (sEMG) signals of muscles adjacent the neurorehabilitation device.

The neurorehabilitation device also comprises a plurality of actuators (and/or stimulators and/or haptic mechanisms), which actuators are arranged to provide stimulation to these muscles. Therefore, the neurorehabilitation device is able to detect EMG signals of muscles, to determine a desired action of a wearer and/or a present muscle activity of a user, and to provide stimulation in dependence on this desired action and/or muscle activity. This enables the rehabilitation device to be used by a wearer to support a rehabilitation process. The actuators typically provide stimulation in the form of a vibration (e.g. a haptic stimulation), where the use of a vibration instead of an electrical stimulation reduces the amount of interference caused by the vibration; specifically, the use of a vibration reduces interference with any electrical signals being picked up by the sensors 2. This enables continual monitoring of the signals even while the stimulation is being provided.

The actuators are typically used to activate two key neural pathways in the body. Firstly, these actuators may target cortical areas, establishing a brain-muscle closed-loop system for initiating a neurorehabilitation process. Secondly, the actuators may engage spinal pathways, where stimulation of the spinal pathways activates afferent fibres within muscle spindles. These fibres synapse directly with alpha motoneurons of the muscles to which the fibres belong. Such activation of a spinal pathway creates a spinal-muscle closed- loop, leading to increased muscle tone in a muscle being stimulated by an actuator. This provision of closed loop provides augmentation to a neurorehabilitation process that can assist in muscle force development and aid in maintaining muscle activation stability. Moreover, when the spinal pathway is activated, the spindle of the stimulated muscle has a negative synaptic connection with the antagonist muscle. This results in relaxation of the antagonist muscle whenever vibration occurs, simultaneously stretching the antagonist muscle and releasing muscle tone. This is particularly beneficial for patients with spasticity to release the spastic muscle.

Typically, the neurorehabilitation device 1 comprises a plurality of sensors 2, which sensors are arranged to record EMG signals on multiple points of a limb of a user. By including multiple sensors located at different points, precise localisation of the sensors is not required (where this enables the device to be attached by a non-specialist).

The neurorehabilitation device 1 comprises one or more of: One or more sensors or electrodes, e.g. one or more sensing electrodes arranged to detect and record EMG signals.

One or more amplification units, e.g. one or more amplifiers arranged to amplify signals recorded using the electrodes.

One or more actuators, e.g. one or more haptic feedback units or vibrotactile motors and/or one or more actuators arranged to stimulate a muscle of a wearer of the neurorehabilitation device.

A processor unit, e.g. a processor arranged to analyse the signals received from the electrodes/amplification units and/or to provide instructions to the actuators.

An internal memory unit, e.g. to store data collected from the electrodes/amplification units for extended periods of time.

One or more expansion ports to enable additional hardware or sensors to be connected to the neurorehabilitation device.

A communication interface that enables the neurorehabilitation device to communicate with other computer devices, e.g. a universal serial bus (USB) interface and/or a wireless interface, such as a Bluetooth® interface or a local area network (LAN) interface that enables the neurorehabilitation device to communicate via the Internet.

A housing and/or a casing to protect the other components.

Typically, the electrodes are designed in a circular shape. In various embodiments, the electrodes have a diameter in the range of 2mm to 10mm. In some embodiments, the electrodes comprise rounded edges (e.g. edges rounded with a radius of 0.5mm) to avoid scratching the skin of a wearer of the neurorehabilitation device.

The sensors and/or electrodes are typically formed of materials with good conductivity, such as copper or brass. These materials may then be coated with a layer (e.g. a 1 -micrometer-thick layer) of gold to avoid any oxidation or skin irritation. The height of the electrode may be in the range of 2-3mm, with this height being offset from the housing (e.g. by 0.5 to 1 .5mm) to ensure good contactwith the skin surface. A suitable distance between the electrodes is 0.2-0.6mm to prevent interference while minimizing the overall size.] It will be appreciated that each of these exemplary dimensions are exemplary - other sizes and shapes of electrodes may be used.

The amplification unit typically comprises (and/or is arranged to perform) a first amplifier stage of active amplification (e.g. pre-amplification), which first stage typically comprises a low-noise instrumentation amplifier (INA), followed by a second stage of operational amplification (e.g. using an op-amp), which second stage provides signal conditioning and subsequent amplification.

Typically, both of these amplification stages are powered by a single power supply. To eliminate the need for a dual supply and to streamline component count, the amplification unit may comprise an analogue to digital converter (ADC) alongside a virtual ground that biases a detected signal (e.g. an EMG signal) entering the amplification unit at a mid-point of a voltage range of the ADC. This biasing ensures that the alternating current (AC) components of the detected signal are accurately captured within the ADC's dynamic range, effectively mapping the detected signal to the full range of the ADC while preventing signal clipping or distortion.

In some embodiments, the neurorehabilitation device 1 incorporates at least 4, at least, 8, at least 12, and/or at least 16 EMG recording channels. Such embodiments are of particular use where the neurorehabilitation device comprises the bracelet 1 a. In some embodiments, the neurorehabilitation device 1 incorporates at least 25, at least, 50, at least 70, and/or at least 100 EMG recording channels. Such embodiments are of particular use where the neurorehabilitation device comprises the sleeve 1 b. Such embodiments may be used where the neurorehabilitation device is arranged to cover a plurality of muscles or muscle groups, e.g. an entire limb’s muscle groups.

By using a plurality of electrodes and/or recording channels, the activity of a muscle can be determined even where the neurorehabilitation device is not positioned directly above the muscle. Therefore, an embodiment with a plurality of electrodes and/or channels provides a more versatile neurorehabilitation device that does not require a precise fit. Furthermore, users who have CNS damage often have weaker muscle signals than healthy individuals. Therefore, determining the area of muscle activity can be particularly challenging with these users. By using a plurality of electrodes and/or channels to determine muscle activation, a precise location of a muscle activation can be determined even for users that have CNS damage.

In some embodiments, the electrodes and the actuators may be combined into simultaneous sensing and stimulating units, which are arranged to both sense muscle activity in a certain area and also to stimulate the muscle in this area. This enables the provision of stimulation to a suitable area based on a muscle activity reading.

The neurorehabilitation device 1 may be incorporated into an existing system, e.g. an existing rehabilitation robot. In these embodiments, the neurorehabilitation device may be arranged to use components (e.g. hardware of software modules) of this existing system. For example, the neurorehabilitation device may use a processor and/or an electrode of this existing system.

Equally, the neurorehabilitation device 1 may be used as a standalone device. Such an implementation is particularly useful where the neurorehabilitation device is provided for home and/or personal use. In this regard, conventional muscle stimulation systems require expensive and/or specialist equipment that might only be available to hospitals. The neurorehabilitation device of the present disclosure can be provided relatively cheaply and used at home so as to simplify the provision of neurorehabilitation therapy.

Furthermore, the neurorehabilitation device could be used as a part of therapy provided by a healthcare professional (e.g. within a physio session), in these situations, the neurorehabilitation device 1 may be used to help patients establish a strong sensory-motor loop, so as to enhance the effectiveness of conventional physical therapy.

In some embodiments, the neurorehabilitation device 1 is arranged to connect to a further computer device, such as a local personal computer (PC) or a mobile phone, where this enables the neurorehabilitation device to use a capability of the PC, such as a powerful processor of the PC. This can enable a user to receive feedback on their rehabilitation using the neurorehabilitation device and a PC (e.g. without using expensive or specialist equipment).

The present disclosure describes a neurorehabilitation device that provides stimulation to a wearer of the device based on a closed-loop control system. Specifically, the device detects a muscle activity of the wearer, such as the wearer contracting a muscle, and the device provides stimulation to the user based on this muscle activity. For example, the device may resist the contraction of the muscle so as to encourage the user to contract the muscle more strongly (such operation may be particularly beneficial in the later stages of a neurorehabilitation process after a neural pathway and/or a muscle of a user has already been strengthened to some extent in the earlier stages of this process). Equally, the device may assist the contraction of the muscle to provide encouragement to the user (such operation may be particularly beneficial in the earlier stages of a neurorehabilitation process where a neural pathway and/or a muscle of a user is relatively weak). Referring to Figure 2a, there is described a method of providing stimulation to a muscle of a wearer of a rehabilitation device 1 . This method is typically carried out by the neurorehabilitation device. Certain steps of the method may be carried out by other devices with the method then being implemented by a plurality of devices, the plurality of devices including the neurorehabilitation device.

In a first step 11 , the neurorehabilitation device 1 records EMG signals associated with muscle activity. Typically, this step comprises recording the signals over a period of time, though equally the signals may be detected instantaneously and recorded for a single point in time. The first step may involve recording a signal for a specific muscle. Equally, this step may involve recording signals (and activity) for a plurality of muscles, such as a muscle group.

In a second step 12, the neurorehabilitation device 1 or a connected computer device processes the recorded signals. Processing the signals typically comprises one or more of: amplifying the signals, filtering and/or band-pass filtering the signals (e.g. to remove noise orto remove outlying or anomalous data points), and extracting features from the signals. Regarding the extraction of features, the neurorehabilitation device or the connected device is typically arranged to process the recorded signals so as to extract one or more features from these signals. The features may, for example, comprise peaks in activity, statistical features (e.g. mean signal readings), or patterns in the signal data that can be used to determine muscle activity.

In various embodiments, the extraction of the features, or the computation of muscle activity based on these features, may comprise using a machine learning (ML) model, a neural network, and/or an artificial intelligence (Al) algorithm. In particular, extracted features may be fed into a machine learning model that is arranged to determine and output a muscle activity that is associated with these features.

In some embodiments, the neurorehabilitation device 1 or the connected computer device is arranged to form a baseline or benchmark reading for a wearer at a first time and then to process recorded signals in order to detect a change from this baseline reading (and thereby determine a muscle activity). More specifically, the neurorehabilitation device may be arranged to perform a baseline measurement process during which a user activates their muscles and the resultant signals are measured. These signals may then be used as baseline signals, where the extraction of the features occurs based on a difference of a measured signal from a baseline signal.

In particular, the processing 12 of the signals may comprise determining one or more muscle activations based on the signals (e.g. to detect that a muscle is being flexed). The processing may further comprise determining a feature of this activation, e.g. an intensity, a duration, a length (e.g. a length of a movement associated with the muscle activation), and/or a frequency.

In a third step 13, the neurorehabilitation device 1 provides stimulation and/or feedback to the muscle(s) of the wearer of the neurorehabilitation device in dependence on the processing 12 of the signals (e.g. via actuators and/or haptic motors). Typically, this comprises providing stimulation to the muscle(s) for which muscle activity has been recorded (in the first step 11).

The stimulation may comprise providing feedback to a user; for example, to indicate to a user that they should flex a certain muscle and/or alter a feature of a muscle activation (e.g. an actuator of the neurorehabilitation device may be vibrated in order to indicate a location of a muscle that a user should flex). Equally, the stimulation may comprise providing stimulation that affects (e.g. assists or counteracts) a muscle activity of the user. Therefore, the stimulation may act to help the user to activate a muscle (to complete an activity) or may act to resist an activity of a muscle (to provide resistance to a user and to make the completion of an activity more difficult).

The actuators may be arranged to provide proprioceptional or force feedback of an activated muscle so as to increase muscle engagement during a rehabilitation process. The stimulation delivered by the actuators can be fine-tunned to elicit a desired cognitive cortical response, where this improves the inducement of neuroplasticity for a wearer of the neurorehabilitation device 1. Moreover, the device can promote associative plasticity by having cortical inputs synchronised with the peripheral inputs from the stimulator at the spinal level.

Typically, the stimulation is aimed at a frequency band of 80 - 100 Hz, where this elicits the firing of muscle spindles. As described herein, this firing of the muscle spindles may then be analysed to update an amount of stimulation (e.g. an intensity associated with the stimulation).

Typically, the device is arranged so as to synchronise ‘cortical’ inputs that relate to the motor output commands generated from the motor cortex of a user with ‘peripheral’ inputs that relate to the sensory stimulation elicited from the device. Since device is arranged to provide stimulation at the same location as the initial muscle activation (e.g. the stimulation provided by the device is determined so as to be at the location at which the muscle activation is detected), the cortical and peripheral signals will be ‘synchronised’ and establish a sensory-motor closed loop.

The device may be arranged to fine-tune the stimulation by adjusting the frequency of the stimulation to elicit responses at the spinal level. In particular, the stimulation is typically arranged to activate a spinal pathway in the spinal cord, where the result of this pathway activation is the contraction of the stimulated muscle. Typically this requires the use of a specific stimulation, e.g. a stimulation within a frequency range of 100 - 140Hz. In some embodiments, a stimulation may be provided that has a frequency within the aforementioned range and/or an amplitude between 0.5mm and 1 mm (where the amplitude of the stimulation is related to a displacement generated by the actuator (and so an amount of displacement of the skin adjacent the actuator).

The frequency range of the stimulation is typically in the range of 80 - 140 Hz and/or in the range 100 - 140 Hz.

In some embodiments, the method comprises providing a target muscle activation to a user, determining an actual muscle activation based on the recorded and/or processed signals, and then providing feedback to the user based on a difference between the target muscle activation and the actual muscle activation. This difference may also be used to determine a suitable stimulation dose for the user, where this stimulation is then provided to the user via the actuators. The stimulation dose may be saved and then used for future performances of the method, where this enables a wearer of the neurorehabilitation device 1 to track a change in the stimulation dose over time to assess their progress.

By providing this feedback, the neurorehabilitation device 1 is able to induce the formation of desirable neural pathways in order to improve the rehabilitation of the wearer of the device.

The recording of the EMG data allows the neurorehabilitation device 1 to track the motor intention of a wearer even where the limb movement ofthat wearer is limited. This enables the neurorehabilitation device to be used by a wide range of wearers regardless of their motor function level. The proposed device also requires minimal assistance or training and so enables neurorehabilitation to be performed more cheaply and easily than existing products. In particular, many acute or severely impaired patients are not able to use conventional rehabilitation techniques as they do not have active joints. However, even where a user cannot move a muscle noticeably, there is still a weak muscle activation that occurs when that user attempts to move that muscle. The present device is able to detect such an activation and so the device can be used to stimulate a muscle even for severely impaired patients.

Various embodiments of the device also enable: monitoring of a wearer’s recovery, the gamification of neurorehabilitation, and auto-advising of an intensity and dose of exercise training.

Typically, the neurorehabilitation device is arranged to record sEMG readings using a multi-channel approach, where this eliminates potential drawbacks of EMG recording such as spatial localization and the electrode lift-off. In addition, the present disclosure considers a novel approach of sEMG acquisition allows a higher selectivity of the target muscle by modulating the cross talk effect between the muscles. This modulation is accomplished by strategically positioning electrodes within the electrode array, thereby controlling an interelectrode distance in this electrode array (as will be described in more detail below). Low-distance electrode spacing reduces cross-talk effects and enhances sensitivity to the target muscle. Furthermore, by providing adjustable gain settings in the amplification system, the neurorehabilitation device 1 described herein enables optimization of the amplification level, ensuring that weak EMG signals coming from low- distance electrodes are accurately detected and recorded without saturation or excessive noise. This approach optimizes signal fidelity and enhances the precision of muscle monitoring.

While the above method includes a step of providing stimulation to a muscle of a wearer, it will be appreciated that the neurorehabilitation device 1 may provide benefits without providing any stimulation. For example, by recording the EMG data, the device may enable analysis of a user’s rehabilitation even if the device does not (or is not arranged to) provide stimulation.

In particular, with or without the stimulation, the neurorehabilitation device 1 may be arranged to determine a suitable exercise dose and/or intensity based on the recorded and/or processed signals. In some embodiments, the neurorehabilitation device may determine a suitable exercise dose and then output this suitable dose to a further device, where this dose may then be applied using a separate device. In a practical example, the neurorehabilitation device may determine a suitable dose based on exercises performed by a user at home, with this dose then being provided to a rehabilitation professional that can devise a suitable rehabilitation plan based on the dose.

Similarly, while the above method includes a step of detecting signals using a neurorehabilitation device, it will be appreciated that the neurorehabilitation device 1 may provide benefits without performing this detecting step. For example, by providing stimulation to a muscle, the device may contribute to a neurorehabilitation process even if the device does not have knowledge of muscle signals and/or if the device receives muscle signals from a further device.

That being said, the neurorehabilitation device 1 is typically arranged to both sense muscle signals for a muscle and to provide stimulation to that muscle, where this enables the provision of a compact device that can provide appropriate stimulation to the muscle using a closed-loop control method.

The neurorehabilitation device 1 is typically arranged to provide stimulation based on a signal indicating muscle activity, where the stimulation may be provided within 100ms of detecting the muscle activity. Typically, the stimulation comprises a vibration that does not interfere with the performance of the sensors (e.g. of the sEMG sensors). This enables the stimulation to be provided without needing to stop or alter the sensor monitoring. This also enables the sensor monitoring to continue as the stimulation is provided.

In some embodiments, the neurorehabilitation device 1 is arranged to record signals relating to muscle activity, with the signals being transmitted to an external computer device following the recording. This external computer device may then be used to process the signals before transmitting instructions to the neurorehabilitation device, the instructions causing the neurorehabilitation device to provide stimulation to a muscle. Such embodiments provide a neurorehabilitation device that is able to stimulate a muscle based on signals recorded by the device, while also enabling these signals to be processed using a more powerful and capable external computer device. For example, the signals may be transmitted to a server with a powerful processor, processed at this server, and then appropriate instructions may be sent back to the neurorehabilitation device.

In some embodiments, the neurorehabilitation device 1 comprises a processor that is arranged to determine stimulation parameters on the device, where this may enable real-time control of the stimulation in response to the sensed EMG activity (e.g. without the delay related to communicating with a further computer device). Referring to Figure 2b, there is shown a more detailed method of providing stimulation to a muscle of a wearer of the neurorehabilitation device 1 . This method may be performed by the neurorehabilitation device 1 , by another computer device (e.g. a connected device), or by a combination of computer devices.

In a first step 21 , the neurorehabilitation device 1 (or another computer device) determines a muscle activation of the wearer. Typically, this comprises determining the muscle activation from the EMG data.

In a second step 22, the computer device determines whether a parameter of the muscle activation exceeds a threshold. For example, a percent maximal voluntary contraction (%MVC) of a muscle may be determined from the EMG data and compared to a threshold x, where this indicates a percentage of a maximal voluntary contraction (MVC) that is being achieved by a user. As an example, this threshold may be 5%, 10%, or 20%. It will be appreciated that the use of a %MVC is optional; numerous parameters may be used to assess the muscle activation, where typically the parameter comprises a measure of intensity of the muscle activation.

If the parameter is determined not to exceed this threshold, then the method may return to a state prior to the first step 21 (e.g. where the computer device waits until another muscle activation is detected).

If the parameter exceeds the threshold, then, in a third step 23, the neurorehabilitation device 1 provides stimulation to the wearer of the device. This stimulation may depend on the parameter of the muscle activation.

In a fourth step 24, a duration of the muscle activation is determined and this duration is compared to a threshold value (e.g., to a target duration). If this target duration is not exceeded, then the method returns to before the second step 22, the parameter of the muscle activation is re-determined, and then the second step is repeated. Therefore, throughout the target duration, the parameter(s) of the muscle activation are repeatedly determined and compared to target parameters with the provided stimulation being altered based on a difference between the measured parameters and the target parameters. The target duration is typically in the order of seconds, e.g. the target duration may be no more than 30 seconds, no more than 20 seconds, and/or no more than 10 seconds.

Once the target duration is exceeded, in a fifth step 25 the method proceeds to a pause (e.g. where the pause duration may be similar to, or the same as, the target duration). This gives the user a chance to rest. The method may then be repeated, e.g. for a target number of muscle activations. Following a target number of activations, the method may terminate. Equally, the termination of the method may depend on a user input orthe meeting of a termination condition (e.g. this condition may be associated with exceeding a threshold muscle activation).

The threshold parameters (e.g. the target muscle contraction and the target duration) may change during the repetitions of the method. For example, the method may be repeated for a plurality of muscle activations, with the target duration for a second activation being lower than a target duration for a first activation and/or with a target contraction for a second activation being lower than a target contraction for a first activation.

Referring to Figures 3a - 3d, a detailed embodiment of a neurorehabilitation device and a system comprising the neurorehabilitation device is described. It will be appreciated that the combination of componentry described below is an example and that, in practice, the neurorehabilitation device may be provided with any combination of the componentry described below.

Referring to Figure 3a, the neurorehabilitation device 1 is located adjacent a patient 110 so that the neurorehabilitation device can receive EMG data (e.g. EMG signals) from N recording EMG points located on this patient. This EMG data is transferred to a multichannel EMG monitoring module 120, which processes the EMG data and feeds this processed data into a calculation module 130. The calculation module is arranged to communicate with an analysis application 150 - which may be implemented on the neurorehabilitation device or may be implemented on a separate computer device - in order to determine feedback signals for stimulating the patient. The calculation module 130 then provides these signals to a haptic stimulation module 140, which module is arranged to operate K stimulation points in order to provide feedback and/or stimulation to the patient.

Referring to Figure 3b, the patient 110 can be modelled as a cortex 112 which is subject to voluntary control of the patient (e.g. an intentional muscle contraction) and afferent firing spikes, which are triggered by mechanoreceptors 116 of that patient (which mechanoreceptors may be stimulated by the actuators providing feedback to the patient).

Based on a combination of the voluntary control input and the involuntary afferent firing strikes, the cortex 112 causes motoneuron firing spikes, which cause an activation of a muscle 114 of the patient 110. Based on this modelling of the patient, it is apparent that the muscle activation can be controlled by using the actuators of the neurorehabilitation device 1 to cause a desired intensity of afferent firing spikes (with this intensity being determined based on the effect of the voluntary control input at a given time.

The neurorehabilitation device 1 is arranged to received EMG signals from N recording EMG points that are located adjacent a muscle of the patient. These signals are received by the multichannel EMG monitoring module 120. As shown in Figure 3c, this multichannel EMG monitoring module comprises a plurality of EMG strips 122-1 , 122-2, 122-M, wherein each strip comprises N EMG units. Each EMG unit comprises: an electrode, a pre-amplifier, a band-pass filter, and a post-amplifier. Using this componentry, the units are thus able to receive and process a signal relating to one of the EMG points so as to provide a processed output to the calculation module 130.

Referring to Figure 3d, the calculation module 130 comprises an EMG acquisition module that is arranged to receive the processed EMG signals from the multichannel EMG monitoring module 130 and to transfer these signals to an internal gesture model analysis module 134. This module is arranged to determine, from the processed signal, a muscle activity that is associated with the EMG signals. For example, the internal gesture model analysis module may identify a gesture that has been performed by a wearer of the neurorehabilitation device 1 and/or may determine an intensity, duration, or frequency of a muscle activation that has been detected by the neurorehabilitation device.

The calculation module 130 communicates with the analysis application 150, and more specifically with a controller 152 of the analysis location, to compare the determined gesture information to target parameters. The controller 152 may, for example, comprise a controller of a feedback application or a game application, where the analysis application may be arranged to output a target forthe user. This target may, for example, indicate a desired movement or a desired intensity of movement. Equally, this target may indirectly prompt a desired movement, e.g. the target may be a part of a game, where a user may need to swing a virtual racquet, pull a virtual lever, or complete a virtual puzzle. These tasks may be associated with a certain desirable movement; for example, the user may be required to pull a virtual lever in a certain direction and/or with a certain strength. By indirectly prompting the movement, the analysis application may provide a more pleasant user experience. This is especially relevant for neurorehabilitation applications, since a user may need to repeat similar movement regularly for an extended period of time. By connecting these applications to a feedback application or a game application, the boredom of the user is reduced as they repeat these movements. Furthermore, the controller 152 may increase a difficulty of the movements over time based on a progress of the user, e.g. to increase an amount of feree required to operate a lever or by providing a more difficulty to hit target.

The controller 152 communicates with a performance evaluation module 154 of the analysis application 150, where the performance evaluation module is arranged to determine a difference between an actual muscle activation and a desired muscle activation. For example, the performance evaluation module may receive parameters relating to an actual muscle activation from the calculation module 130 and then compare these parameters to target parameters from the controller 152. Based on this difference, the performance evaluation module 154 is arranged to provide an output to a stimulation feedback calculation module 136 of the calculation module 130.

The stimulation feedback calculation module 136 is arranged to determine feedback signals for stimulating the patient. The feedback signals may be associated with a prompting indication (e.g. vibration that indicates that a user should tense more without actually assisting in this tensing). Equally, the feedback signals may cause a muscle activation forthe user (e.g. by providing stimulation to a muscle of the user) in order to bridge a gap between a desired activation and an actual activation.

The determined feedback signals are provided to the haptic stimulation module 140, which haptic stimulation module comprises a plurality of actuators (e.g. haptic elements that each comprise a haptic driver 142-1 and a haptic motor 144-1). These actuators operate in dependence on the provided feedback signals to stimulate the mechanoreceptors 116 of the patient 110 and thereby affect the motoneuron firing spikes provided by the cortex 112.

By continually monitoring the muscle activity via the N recording EMG points, the neurorehabilitation device 1 is able to continuously identify a muscle activity of a muscle of a user. Since the neurorehabilitation device is providing the feedback signals to the haptic stimulation module 140, the neurorehabilitation device is also able to determine an amount of this activity that is attributable to the voluntary control of a user and an amount of activity that is attributable to the stimulation provided by the haptic stimulation module. Furthermore, typically the method commences with zero feedback being provided by the haptic stimulation module, where this enables a baseline to be established for the muscle activity caused by the voluntary control signals (and by ramping up the stimulation from this zero feedback starting point, the neurorehabilitation device is also enables able to determine a relationship between stimulation and muscle activity) .

The amount of feedback and/or stimulation provided by the haptic stimulation module 140 may then be continuously modified in dependence on one or more of: an amount of muscle activity detected by the neurorehabilitation device 1 ; an amount of detected muscle activity that is attributable to voluntary control of a wearer of the neurorehabilitation device; a difference between a target (or desired) muscle activity and a determined muscle activity (either a determined overall muscle activity or a determined muscle activity attributable to voluntary control). Altering the stimulation provide by the haptic stimulation module 140 will alter the muscle activity detected by the multichannel EMG monitoring module 120 so that the disclosed neurorehabilitation device forms a closed loop control circuit that is updated continuously so as to achieve a target muscle activation.

In a practical example of the use of the neurorehabilitation device 1 , the device is worn on the limb (or adjacent the muscle group) that requires rehabilitation. The device then is synchronised with a computed device on which the analysis application 150 is installed (e.g. a smartphone or PC). The application enables the wearer of the device to check their rehabilitation history and progress and to identify and set a training suggestion. When the patient selects or confirms a training process, the application launches a corresponding set of exercises. To complete these exercises, the wearer may need to contract the muscle group corresponding to the location of the device being worn. This muscle activation serves as a training exercise for the wearer, while also providing EMG data that is recorded, stored and analysed to evaluate the performance of the wearer. During this exercise, the muscle activation is also used to trigger sensory feedback that is delivered to the patient. This feedback enables the wearer to learn which muscle to activate or move and improves the formation of neuroplasticity pathways during the exercises. The analysis application may then indicate when the wearer should rest and/or when the user should finish the exercise programme. In another practical example of the use of the neurorehabilitation device 1 , the device (or an associated computer device) may be worn during the course of a normal day of the user. The device may then identify muscle activations relating to activities of that user (e.g. making tea, climbing stairs, or carrying objects) and provide stimulation based on these identified muscle activations. In such a scenario, the neurorehabilitation device is able to provide stimulation and feedback while a user goes about their daily lives so as to provide non-invasive therapy to that user.

Referring to Figure 4a, there is shown an implementation of the neurorehabilitation device 1 which comprises a sensory strip that has, on a top portion, a reference ground 202, one or more sensor (e.g. or sEMG) units 204, a multiplexer 206 for processing the signals from these sensor units, and an inter-strip bus 208. Referring to Figure b, there is shown a bottom portion of this device.

The sensor units 204 are arranged to detect signals, e.g. EMG signals, associated with a muscle activation and to provide these signals to the multiplexer 206. The multiplexer is arranged to process the signals and to select one or more of signals (from one or more respective sensor units) to pass to a central processor. The selection of the signals may, for example, be determined based on a user input and/or based on a parameter of the signals (such as a signal intensity).

The inter-strip bus is arranged to facilitate the communication of these signals between components of the sensory strip.

Referring to Figure 4c, an aspect of the present disclosure relates to a modular sensor unit that comprises an amplifier and a sensor (e.g. an electrode).

The use of modular sensor units provides scalability in that it allows a manufacturer to include a desired number of sensor units that is suitable for a particular context. For example, the modular sensor units may be ‘snap-in’ units that can be pressed together and/or pulled apart (e.g. using a snap fit) to enable a user to readily connect ordisconnect a plurality of modular sensor units. Typically, each sensor unit is associated with, and/or connected to, an amplification board so that these components can be provided in a single module. The sensor units are connected via an inter-EMG bus 210 that enables signals to pass between the sensor units.

More specifically, each modular sensor unit typically comprises one or more sensors (e.g. electrodes 216) and an amplifier 212 that is arranged to amplify the signals from the sensors before transmitting the signals to a further computer device (e.g. the multiplexer 206 and/or a central processor of the neurorehabilitation device).

By including the amplifier 212 within each modular sensor unit, it is possible to amplify these signals shortly after their recording so as to minimize artifacts introduced into the signals after their recording and to minimize the amplification of any artifacts in the signal (e.g. artifacts caused by the vibration of actuators).

In other words, by locating the amplifier 212 within the sensor unit and adjacent the electrodes 216 it is possible to amplify only the signals recorded by the electrodes, whereas if the amplifier were located far from the electrodes (e.g. as might be the case where a single amplifier is used to amplify signals from multiple sensors), then this amplifier would amplify both the recorded signals and any artifacts that have been introduced to the signals on the way to the amplifier.

This is illustrated by Figure 5a, which shows the difference between a conventional arrangement and the arrangement of the present disclosure. With a conventional arrangement, in which each electrode used to record EMG activity is used spaced from an amplifier used to amplify the detected signals, a significant amount of interference is introduced into the signals prior to the amplification. In contrast, with the present arrangement, the detected EMG signals are amplified shortly after recording so that the amount of interference that might enter the signals before this amplification is reduced. In order to minimize a transfer length between the sensors and the amplifiers, the modular sensor unit may comprise an amplifier that is connected directly to an electrode in a back-to-back arrangement (e.g. without any printed circuit board (PCB) layers located between the electrode and the amplifier).

In various embodiments, the length of electrical track (e.g. wires) that connects the electrodes and the amplifier is less than 10mm, less than 5mm, less than 1 mm, less than 0.1 mm, and/or less than 0.01 mm.

In some embodiments, the modular sensor unit may comprise an actuator and/or a vibrator. For example, the electrodes may be formed on a (e.g. metal) elongated portion of the sensor unit, where the electrodes are arranged to sense electrical signals that are associated with muscular activity of a user. The actuator may then be arranged to vibrate this elongated portion so as to provide stimulation to the muscles of the user.

Typically, the neurorehabilitation device 1 comprises a plurality of such modular sensor units, where each of the units is connected to the multiplexer 206 and wherein the multiplexer determines the units for which signals that are transmitted to a further computer device associated with the neurorehabilitation device.

Referring again to Figure 4c, the modular sensor units and/or the neurorehabilitation device 1 may also comprise a stiffener 214 that is arranged to provide structural rigidity to the components of the device.

The stiffener 214 may be arranged to reduce an amount of vibration experienced by the sensor units, where this reduces the amplitude of artifacts introduced into the EMG signals by the vibration of the actuators. The stiffener may be located adjacent the electrodes 216 and/or between the electrodes and the amplifiers 210 (where there is typically an electrical connection that connects the electrodes 216 and the amplifiers by passing through the stiffener. In some embodiments, the stiffener is provided as a part of the sensory strip, where a plurality of modular sensor units may then share a single stiffener.

Referring to Figures 4d and 4e, there is shown, respectively a monopolar and a bipolar arrangement of electrodes, which electrodes are arranged to detect EMG signals of a wearer of the neurorehabilitation device 1 .

Referring first to Figure 4d, the monopolar arrangement comprises a plurality of electrodes Ei, E2 E_n, where each electrode is arranged to compare a detected signal and/or value recorded by that electrode to a reference signal and/or value recorded by an electrode (or channel) R in order to obtain a recorded signal Chi, Ch2 Ch_n. Such an arrangement enables the neurorehabilitation device 1 to determine the (e.g. relative) volume conduction properties of the muscles adjacent the neurorehabilitation device 1 (e.g. the high density array of the sleeve 1 b) based on values recorded by the electrodes, which enables the neurorehabilitation device to calculate complex physiological measurements such as motor unit decomposition. This enables the neurorehabilitation device to observe direct individual outputs from the spinal cord to individual muscles as shown in Figure 5b.

Specifically, as is shown in Figure 5b, each of the sensor elements (e.g. channels) of the neurorehabilitation device 1 may be located at a row index and a column index. By evaluating the signal at each pair of row/column indices (e.g. at each sensor location), the activity of various muscle areas adjacent the neurorehabilitation device can be detected so as to enable analysis of the muscle activity of a user.

Furthermore, these measurements that have been described with reference to Figure 5b, enable the neurorehabilitation device 1 (or a device connected to the neurorehabilitation device) to perform image processing in order to obtain spatial muscle activity for calibration purposes.

Referring to the specific implementation shown in Figure 5b, figure 5b-A shows the decomposition of motor unit readings which indicates the activity of a particular muscle (or muscles) of the user. These readings may be obtained using a monopolar arrangement of electrodes, as shown in Figure 4d, Thereafter, Figure 5b-B shows how these muscle activities are used to determine a type and extent of activation of the muscle. In the implementation of Figure 5b, the muscle activities are processed in order to determine flexion, extension, radial, and ulnar activations for a muscle adjacent the neurorehabilitation device 1 .

Referring then to Figure 4e, in a bipolar configuration the signals are determined using pairs (Ei- and Eu) of spaced electrodes, where a value across these electrodes is evaluated to determine the EMG signals Ch1 . Such an arrangement provides an indication of a difference in activity present in the areas of the electrodes. While Figure 4e shows only a single pair of electrodes, it will be appreciated that more electrodes may be provided where various signals and channels may be obtained by taking readings between different pairs of electrodes. With such an arrangement (with a plurality of electrodes) any crosstalk effect present in the recorded signals may be modulated by selecting different electrode pairs from within the array of electrodes, thereby increasing or decreasing the muscle selectivity (as is described further below with reference to Figure 6).

It will be appreciated that the monopolar and bipolar configurations are not exclusive and that the neurorehabilitation device 1 may comprise electrodes in each of monopolar and bipolar configurations. Indeed, the same electrode may be used (e.g. sequentially or simultaneously) as part of both a monopolar arrangement and a bipolar arrangement.

Referring to Figure 4f, there is shown a stimulator M. The stimulator (e.g. the actuator) is arranged to provide stimulation to a muscle of a wearer of the neurorehabilitation device 1 . Typically, the stimulator is located on, and/or is a part of, the sensory strip of the neurorehabilitation device. The stimulator may be located in a groove of the sensory strip so as to isolate the vibration caused by the stimulator from the sensor units of the neurorehabilitation device and to reduce the artifacts introduced by the vibration into the signals recorded by the sensor units.

One potential issue with the detection of EMG signals is crosstalk interference between muscles. That is, it can be difficult to isolate EMG signals for a given muscle (since a user that is activating one muscle will likely also be activating other muscles, and these other muscle activations will produce EMG data that acts as noise when trying to detect the EMG signals associated with the activation of the given muscle).

Therefore, referring to Figures 6a - 6d, the present disclosure provides an arrangement of electrodes that acts to modulate the crosstalk interference level between muscles.

The arrangement of Figures 6a - 6d comprises an array of electrodes where different pairs of electrodes can be selected from within this array in order to obtain signals associated with these pairs of electrodes. More specifically, the arrangement comprises a plurality of electrodes associated with different interelectrode distances, so that by selecting electrodes with different distances it is possible to modulate a muscular crosstalk effect. In some embodiments, each recording unit of the neurorehabilitation device 1 comprises an arrangement of four selective electrodes with different inter-electrode distances where the selection of specific electrode pairs from within this arrangement enables modulation of the crosstalk effect. in this regard, electrode pairs with shorter interelectrode distances facilitate narrow crosstalk (NCT) monitoring, whereas electrode pairs with wider distances enable wide cross talk (WCT) monitoring. With NCT monitoring, the smaller interdistance between the electrodes results in the electrodes (and the sensor comprising the electrodes) monitoring a smaller area, primarily capturing signals from superficial muscles, and providing measurements of the targeted muscle with large specificity to this targeted muscle. This reduces the likelihood of signal contamination from neighboring muscles, making NCT well-suited for detecting fine motor control. Conversely, WCT, with its higher interelectrode distance, expands the sensor monitoring area, enabling the capture of signals from larger muscle groups and deeper muscles. This leads to a significant overlap of signals from neighboring muscles.

The electrode configuration of Figure 6 selectively enables NCT and WCT monitoring so as to define the spatial sensitivity of the recording unit. NCT enhances spatial resolution between different recording units by highlighting significant differences between muscles across units. Conversely, WCT decreases spatial resolution by covering a wider area, potentially leading to similarities between recording units.

Simultaneously measuring NCT and WCT allows for a comprehensive evaluation of muscle function. This involves assessing activation levels of individual muscles through NCT and understanding coordinated activity of muscle groups via WCT. Discrepancies between NCT and WCT can differentiate between superficial and deeper muscles, with NCT reflecting superficial muscle activity and WCT indicating activity from both deeper and superficial muscles. Subtracting these activities, it is possible to obtain the individual activity of deeper muscles. In particular, an increase in the ratio between NCT and WCT can be used to identify an increase in a difference in muscle activations (e.g. to indicate that a local muscle is undergoing substantial activity).

Moreover, this analysis - and the use of the array of electrodes - can reveal compensation patterns or inefficient movement strategies. For instance, if narrow crosstalk signals suggest low activation of a specific muscle while wide crosstalk signals indicate high overall muscle activation, it may imply compensatory recruitment of other muscles to perform a task.

Therefore, the present disclosure envisages the provision of an electrode arrangement that comprises a plurality of electrodes with differing interelectrode differences. The disclosure envisages the simultaneous monitoring of first signals via a first pair (or pairs) of electrodes with a small separation so as to determine narrow cross talk signals and second signals via a second pair (or pairs) of electrodes with a large separation (e.g. larger than the first pair) so as to determine wide cross talk signals. A shared electrode may be present in each of the first pair and the second pair of electrodes, where this enables the monitoring of wide cross talk signals and narrow cross talk signals for the same muscle area (adjacent this shared electrode). The disclosure further envisages the determination of a muscle activation based on each of a narrow cross talk measurement and a wide cross talk measurement. For example, a muscle activation may be determined on the basis of a wide cross talk measurement taken using electrodes with a comparatively high interelectrode distance and this activation may be confirmed, with more specific measurements taken of the activity, using a narrow cross talk measurement taken using electrodes with a lower interelectrode distance.

Referring to Figures 6a - 6d again, there are shown four different (exemplary) readings that can be taken using the displayed electrode arrangement. The arrangement shown in Figures 6a - 6d also comprises a stimulator S and a reference electrode R to increase the versatility of this arrangement. As shown in these figures, and in particular Figures 6b and 6d, in some embodiments the stimulator S (e.g. a metallic enclosure of the stimulator S) is used as a reference electrode.

As shown in Figure 6a, in a first reading with a detection area ratio of 3, a first pair of electrodes with a distance of d1 is used to obtain narrow cross talk readings and a second pair of electrodes with an interelectrode distance of 3*d1 is used to obtain wide cross talk readings.

As shown in Figure 6b, in a third reading with a detection area ratio of 1 .6 a first pair of electrodes with a distance of 3*d1 is used to obtain narrow cross talk readings and a second pair of electrodes with an interelectrode distance of 5*d1 is used to obtain wide cross talk readings. For this reading, the metallic enclosure of the stimulator (S) is used as reference electrode (R).

As shown in Figure 6c, in a second reading with a detection area ratio of 4, a first pair of electrodes with a distance of d1 is used to obtain narrow cross talk readings and a second pair of electrodes with an interelectrode distance of 4*d1 is used to obtain wide cross talk readings.

As shown in Figure 6d, in a fourth reading with a detection area ratio of 5, a first pair of electrodes with a distance of d1 is used to obtain narrow cross talk readings and a second pair of electrodes with an interelectrode distance of 5*d1 is used to obtain wide cross talk readings. For this reading, the metallic enclosure of the stimulator (S) is used as reference electrode (R).

As shown in Figures 6a - 6d, each of these readings enables an activity over a different area to be analysed so that, by considering a plurality of readings simultaneously it is possible to determine a muscle activation based on both the activity of a muscle group (e.g. the group of muscles 1 , 2, and 3) and also to determine a feature of the activity of a single muscle. Higher ratios, typically around 4 or 5, facilitate enhanced isolation of deeper muscles, particularly in larger muscle groups like the lower limb or back muscles. Conversely, smaller ratios, typically ranging from 1 .66 to 3, are more suitable for smaller muscle groups or those closer to the surface, such as forearm muscles.

In some embodiments, a muscular activity of a user is determined by subtracting NTC activity from WTC activity. This subtraction may be performed in hardware by using an operational amplifier. Equally, it is possible to monitor separately the signals generated from the NTC and WTC activity and to process these signals digitally. Such subtraction enables an evaluation to be made of the activity of a specific muscle relative to that of a surrounding muscle group.

In some embodiments, the neurorehabilitation device is arranged to determine one or more pairs of electrodes to be used for a measurement based on a muscle associated with the measurement (e.g. the neurorehabilitation device may identify that a first set of pairs of electrodes is suitable for measuring forearm activity and that a second set of pairs of electrodes is suitable for measuring quadriceps activity.

Typically, the neurorehabilitation device (and the electrode arrangement) is arranged to provide NCT to WCT ratios in the range of 1 - 10, preferably in the range of 1 - 5, more preferably in the range of 1 .6 - 5.

As shown in Figures 6a - 6d, one of the selective electrodes of the electrode arrangement typically comprises an electrode of a sensor unit, which electrode is also used to determine signals associated with the muscle activity of the wearer of the neurorehabilitation device 1.

In some embodiments, a metallic enclosure of a stimulator (S) is used as part of the electrode arrangement (e.g. as an electrode) to both sense muscle activity as a reference electrode (R) and also to provide stimulation. In other words, the actuator (and/or stimulator) may comprise an electrode that is capable of both sensing muscle activity and providing stimulation to a muscle. This may involve the actuator comprising a portion (e.g. a metallic portion) that is arranged to vibrate so as to provide stimulation to a muscle and a portion (e.g. an electrode on this metallic portion) that is arranged to detect muscle activity. By using this actuator as an electrode within the electrode arrangement, it is possible to provide a compact neurorehabilitation device that still has high accuracy. This design is also able to maintain a favorable NCT to WCT ratio, typically falling within the range of 1 .6 to 5.

Referring to Figures 7a - 7c, the present disclosure also considers a method of bidirectional signal processing, in particular to select (via the multiplexer 206) one or more signals that are passed to a processor of the neurorehabilitation device 1 and/or to a processor of a further computer device.

Referring first to Figure 7a, in some embodiments, all of the signals (e.g. from all of the sensor units) being received from the EMG units S1 , S2, ... Sn on the neurorehabilitation device 1 may be passed through to a central processor (e.g. off-device), where this central processorthen determines which signals to process in depth. For example, the central processor may decide to perform in-depth processing only on signals that exceed an intensity threshold.

However, this can result in the processor receiving a large, and unnecessarily large, number of signals. Therefore, as shown by Figure 7b, the neurorehabilitation device 1 may be arranged to carry out at least some processing on-device. In particular, a processor of the neurorehabilitation device may be arranged to process the signals received from the EMG sensors to determine a subset of the signals to transmit to a further (e.g. external) computer device, where these signals can then be subjected to more in-depth processing. This on-device processing may, for example, comprise only transmitting signals that exceed a threshold intensity to the further device.

Referring then to Figure 7c, in some embodiments the neurorehabilitation device 1 comprises a multiplexer 206, which multiplexer is arranged to enable a further computer device that is external to the computer device to select one or more sensors of the neurorehabilitation device, where the neurorehabilitation device is arranged to transfer to the further computer only the signals from the selected sensors.

This method of selective activation reduces the computation and bandwidth load on the neurorehabilitation device while ensuring the further computer device is receiving all desired signals. This provides a versatile processing method that enables frequent updating of the selected signals and that also enables a user of the further computer device to select for in-depth processing signals relating to a specific muscle activity on which that user wishes to focus.

The multiplexer 206 may determine the signals to transmit to the further computer device in dependence on a communication from this further computer device (or another computer device), which communication indicates one or more of: sensor units for which signals should be transmitted; an area of the neurorehabilitation device for which signals should be transmitted; and a muscle and/or a muscle group of the wearer for which signals should be transmitted.

The communication may also define conditions for the selective transmission of recorded sensor signals. For example, the communication may define a threshold intensity that is used by the multiplexer 206 to determine which signals to transmit to the further computer device.

Regarding the forming of the neurorehabilitation device 1 , the neurorehabilitation device 1 may comprise a flexible printed circuit board (PCB) onto which are mounted a plurality of sensor units and/or electrode units. The flexibility of the PCT enables the neurorehabilitation device to be shaped in a desired arrangement.

Typically, the sensor units each comprise an unit that is capable of both: detecting an EMG signal relating to a muscle activity of a muscle of a wearer; and providing stimulation to that muscle. The electrode units may also comprise a processor for: amplifying or filtering EMG signals; and/or processing feedback signals so as to provide the stimulation.

The sensor units are typically each modular units, where this enables the formation of various arrangements of electrode units and neurorehabilitation devices. Typically, the neurorehabilitation device is associated with a large number of channels, e.g. hundreds of channels, where this involves the placement of a corresponding number of electrode units.

The neurorehabilitation device 1 typically comprises a bus interconnector that is arranged to connect the electrode units so as to enable a master processor (either on the neurorehabilitation device or separate to the neurorehabilitation device) to process the signals detected by the electrode units.

The neurorehabilitation device 1 further comprises one or more of a processor for performing analysis of detected signals and a communication interface for communicating with a further computer device (e.g. so that this further device is able to analyse the signals).

The neurorehabilitation device 1 may be formed using an additive manufacturing process and/or a 3D printing process, where this may comprise additive manufacturing using a plurality of material dispensers that dispense different types of material (e.g. to provide a combination of rigid material and flexible material).

Referring to Figures 8a and 8b, the neurorehabilitation device 1 may comprise a fusion of a rigid material 3, e.g. polylactide (PLA) and a flexible material 4, e.g. thermoplastic polyurethane (TPU), where rigid sections, e.g. that comprise the electrode units, are connected by flexible sections in order to provide a neurorehabilitation device of a desired size/shape. The neurorehabilitation device may be provided in a bracelet or sleeve form as shown in Figure 6a, where the components of the neurorehabilitation device are connected in a loop and arranged to be placed around a wearer’s limb. Equally, the neurorehabilitation device may be provided in an array form as shown in Figure 6b, where the components are in the form of an array and a user is able to connect a first end of the array to a second end of the array to form a bracelet or a sleeve. The combination of the rigid material and the flexible material (e.g. in a single combined part) enables the neurorehabilitation device to be provided without the need for extra spacing for functional features such screws and hinge installations and so enables the provision of a more compact (and sensor-dense) device.

Alternatives and modifications

It will be understood that the present invention has been described above purely by way of example, and modifications of detail can be made within the scope of the invention.

Reference numerals appearing in the claims are by way of illustration only and shall have no limiting effect on the scope of the claims.

Claims

1 . A neurorehabilitation device comprising: one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined signals.

2. The neurorehabilitation device of any preceding claim, comprising a multiplexer, wherein the multiplexer is arranged to receive signals from a plurality of modular sensor units and to selectively transmit these signals to a further computer device.

3. The neurorehabilitation device of claim 2, wherein the further computer device comprises a processor of the neurorehabilitation device.

4. The neurorehabilitation device of claim 2, wherein the further computer device comprises a processor separate to the neurorehabilitation device.

5. The neurorehabilitation device of any of claims 2 to 4, wherein the multiplexer is arranged to selectively transmit the signals based on a communication from the further computer device.

6. The neurorehabilitation device of claim 5, wherein the communication indicates one or more modular sensor units for which signals should be transmitted.

7. The neurorehabilitation device of claim 5 or 6, wherein the communication indicates an area, a muscle, and/or a muscle group for which signals should be transmitted.

8. The neurorehabilitation device of any of claims 2 to 7, wherein the multiplexer is arranged to selectively transmit the signals based on a feature of the signals.

9. The neurorehabilitation device of claim 8, wherein the multiplexer is arranged to selectively transmit the signals based on an intensity of the signals.

10. The neurorehabilitation device of any preceding claim, wherein the amplifier is provided in a back-to- back arrangement with the sensors.

11 . The neurorehabilitation device of any preceding claim, wherein a length of electrical track that connects the electrodes and the amplifier is less than 10mm, less than 5mm, less than 1 mm, less than 0.1 mm, and/or less than 0.01 mm.

12. The neurorehabilitation device of any preceding claim, wherein the device comprises a stiffener.

13. The neurorehabilitation device of claim 12, wherein the modular sensor units are located on the stiffener.

14. The neurorehabilitation device of claim 12 or 13, wherein a connection between a sensor and an amplifier of a modular sensor unit is arranged to pass through the stiffener.

15. The neurorehabilitation device of any preceding claim, wherein: the stimulation comprises a haptic stimulation and/or a vibration; and/or wherein the actuators comprise haptic feedback units and/or vibrotactile motors.

16. The neurorehabilitation device of any preceding claim, wherein each sensor unit comprises an actuator for providing stimulation to a muscle of the wearer.

17. The neurorehabilitation device of any preceding claim, wherein stimulation is provided at a frequency of between 100 and 140 Hz.

18. The neurorehabilitation device of any preceding claim, comprising a plurality of modular sensor units, preferably wherein each modular sensor unit is connected to the multiplexer by a bus.

19. The neurorehabilitation device of any preceding claim, wherein the amplifier comprises a first stage and a second stage, wherein: the first amplifier stage comprises an active amplifier stage, preferably a low-noise instrumentation amplifier stage; and/or the second amplifier stage comprises an operational amplifier stage.

20. The neurorehabilitation device of claim 19, wherein the first amplifier stage and the second amplifier stage are powered by a shared power supply.

21 . The neurorehabilitation device of any preceding claim, wherein the amplifier comprises an analogue to digital converter, ADC.

22. The neurorehabilitation device of claim 21 , wherein the amplifier is arranged to bias a signal entering the ADC to a mid-point of a voltage range of the ADC.

23. The neurorehabilitation device of any preceding claim, wherein: the one or more sensors comprise one or more electrodes arranged in a monopolar configuration; and/or the one or more sensors comprise one or more electrodes arranged in a bipolar configuration.

24. The neurorehabilitation device of any preceding claim wherein the sensor unit comprises an actuating electrode that is arranged to both identify a muscle activity and provide a stimulation.

25. The neurorehabilitation device of claim 24, wherein the actuating electrode is a part of a bipolar arrangement of electrodes that is used to reduce a crosstalk of the neurorehabilitation device.

26. The neurorehabilitation device of claim 24 or 25, wherein the actuating electrode comprises an electrode formed on a protrusion, the protrusion being arranged to vibrate so as to provide the stimulation.

27. The neurorehabilitation device of any preceding claim, wherein the one or more actuators are arranged to provide stimulation to the muscle in dependence on the recorded muscle activity.

28. The neurorehabilitation device of claim 27, wherein the actuators are arranged to provide stimulation so as assist the muscle activity.

29. The neurorehabilitation device of claim 27, wherein the actuators are arranged to provide stimulation so as to resist the muscle activity.

30. The neurorehabilitation device of any preceding claim, comprising an arrangement of at least four selective electrodes with different interelectrode distances between the electrodes.

31 . The neurorehabilitation device of any preceding claim, comprising a bracelet and/or a sleeve.

32. The neurorehabilitation device of any preceding claim, comprising an expansion port for connecting hardware to the neurorehabilitation device.

33. The neurorehabilitation device of any preceding claim, wherein the sensors are arranged to record electromyography, EMG, signals and/or surface electromyography, sEMG, signals.

34. The neurorehabilitation device of any preceding claim, wherein the sensors are arranged to detect muscle activity over a plurality of points of a wearer so as to detect the muscle activity of a muscle.

35. The neurorehabilitation device of any preceding claim, wherein the device comprises one or more of: a processor unit for processing sensor readings and/or stimulation signals; a communication interface for communicating with a further computer device; an amplification unit for amplifying signals from the sensors; and a housing.

36. The neurorehabilitation device of any preceding claim, wherein: the device comprises at least 4, at least 8, at least 12, and/or at least 16 recording channels; and/or the device comprises at least 25, at least 50, at least 70, and/or at least 100 recording channels.

37. The neurorehabilitation device of any preceding claim, comprising a plurality of modular sensor units, wherein each modular sensor unit comprises: an electrode, a pre-amplifier, a band-pass filter, and a post-amplifier.

38. The neurorehabilitation device of any preceding claim, wherein the device comprises a plurality of electrode units, preferably wherein the electrode units are formed of a rigid material, more preferably wherein the electrode units are formed of polylactide, PLA.

39. The neurorehabilitation device of any preceding claim, wherein the neurorehabilitation device is formed of a flexible material, preferably thermoplastic polyurethan, TPU.

40. The neurorehabilitation device of any preceding claim, comprising a flexible printed circuit board, PCB.

41. The neurorehabilitation device of any preceding claim, comprising a crosstalk reduction structure.

42. The neurorehabilitation device of claim 41 , wherein the crosstalk reduction structure comprises an arrangement of electrodes, preferably an arrangement of electrodes with a selective interelectrode distance between each pair of electrodes.

43. The neurorehabilitation device of claim 42, wherein the crosstalk reduction structure comprises an arrangement of at least four selective electrodes with different interelectrode distances between the electrodes.

44. The neurorehabilitation device of any preceding claim, wherein the processor is arranged to: determine a first set of signals based on a first pair of electrodes; and determine a second set of signals based on a second pair of electrodes; wherein the distance between the second pair of electrodes is greater than the distance between the first pair of electrodes.

45. The neurorehabilitation device of claim 44, wherein a first electrode is part of both of the first pair of electrodes and the second pair of electrodes.

46. The neurorehabilitation device of claim 44 or 45, wherein the processor is arranged to determine a muscle activation based on each of the first set of signals and the second set of signals.

47. The neurorehabilitation device of any of claims 44 to 46, wherein the first pair of electrodes and the second pair of electrodes are selected from the at least four selective electrodes.

48. The neurorehabilitation device of any preceding claim, comprising a processor for comparing the crosstalk between a plurality of pairs of electrodes with different interelectrode distances.

49. A method of determining a stimulation to provide to a wearer of the neurorehabilitation device of any preceding claim, the method comprising: recording, using the one or more sensor units, a muscle activity of a wearer of the neurorehabilitation device; and determining a stimulation to provide to the user in dependence on the recorded muscle activity.

50. A method of determining a stimulation to provide to a wearer of the neurorehabilitation device, the method comprising: recording, using one or more sensors, a muscle activity of a wearer of the neurorehabilitation device; and determining a stimulation to provide to the user in dependence on the recorded muscle activity.

51 . The method of claim 49 or 50, comprising providing, using one or more actuators, the stimulation to the wearer.

52. The method of any of claims 49 to 51 , comprising recording muscle activity relating to a muscle and/or a muscle group of the wearer and providing the stimulation to said muscle and/or muscle group.

53. The method of any of claims 49 to 52, comprising determining a muscle associated with the muscle activity and/or the neurorehabilitation device.

54. The method of any of claims 49 to 53, comprising determining one or more features of the muscle activity, preferably wherein the features comprise one or more of: an intensity, a duration, a frequency, and a length, of muscle activity.

55. The method of any of claims 49 to 54, comprising determining one or more of: a muscle activation, a desired muscle activation, a desired action of the wearer, and/or a gesture being performed by the wearer based on the recorded muscle activity.

56. The method of any of claims 49 to 55, comprising determining a baseline muscle activity associated with the wearer.

57. The method of any of claims 49 to 56, comprising determining the muscle activity and/or the stimulation using a machine learning, ML, model and/or an artificial intelligence, Al, algorithm.

58. The method of any of claims 49 to 57, comprising providing feedback to the user in dependence on the muscle activity, preferably, wherein the feedback comprises haptic feedback.

59. The method of any of claims 49 to 58, comprising determining a target muscle activity for a user, preferably, wherein the stimulation is determined in dependence on the target muscle activity, more preferably, wherein the stimulation is determined in dependence on a difference between the recorded muscle activity and the target muscle activity.

60. The method of any of claims 49 to 59, comprising determining a stimulation dose in dependence on the target muscle activity, preferably, comprising outputting this stimulation dose, more preferably comprising outputting this stimulation dose to a further computer device.

61 . The method of any of claims 49 to 60, comprising determining the target muscle activity in dependence on an activity history and/or a user profile of the user. Preferably, the activity history is associated with previous recorded muscle activity.

62. The method of any of claims 49 to 61 , comprising increasing a difficulty of the target muscle activity during a neurorehabilitation process.

63. The method of any of claims 49 to 62, comprising determining an amount of the muscle activity that is attributable to a voluntary movement of the user and/or determining an amount of the muscle activity that is attributable to a provided stimulation.

64. The method of any of claims 49 to 63, comprising recording an updated muscle activity following the providing of the stimulation and determining an updated stimulation based on this updated muscle activity, preferably, comprising continuously updating a provided stimulation based on a continuous monitoring of muscle activity.

65. The method of any of claims 49 to 64, comprising initially recording a muscle activity prior to the providing of stimulation.

66. The method of any of claims 49 to 65, comprising determining an association between an amount of provided stimulation and an amount of induced muscle activity.

67. The method of any of claims 49 to 66, comprising determining a target voluntary muscle activity for a user, preferably, wherein the stimulation is determined in dependence on the target voluntary muscle activity, more preferably, wherein the stimulation is determined in dependence on a difference between a voluntary muscle activity determined from the recorded muscle activity and the target voluntary muscle activity.

68. A computer programme product comprising instructions that, when executed by a processor, cause the processor to perform the method of any preceding claim.

69. A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a processor cause the processor to perform the method of any of claims 1 to 67.

70. A processor for a neurorehabilitation device, the processor being configured to perform the method of any of claims 1 to 67.

71 . A neurorehabilitation device comprising the processor of claim 70.

72. An apparatus for determining a stimulation to provide to a wearer of the neurorehabilitation device, the apparatus comprising: means for (e.g. a processor for) recording, using one or more sensors, a muscle activity of a wearer of the neurorehabilitation device; and means for (e.g. a processor for) determining a stimulation to provide to the user in dependence on the recorded muscle activity.

73. A method of manufacturing a neurorehabilitation device, the method comprising: providing one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and providing one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined signals.

74. The method of claim 73, comprising selecting a number of modular sensor units in dependence on a desired size and/or function of the neurorehabilitation device.

75. The method of claim 73 or 74, comprising connecting each of the modular sensor units to a multiplexer.

76. A kit of parts comprising: one or more modular sensor units, wherein each sensor unit comprises: one or more sensors for determining a signal, the signal identifying a muscle activity of a wearer of the neurorehabilitation device; and an amplifier for amplifying the determined signals; and one or more actuators for providing stimulation to a muscle of a wearer of the neurorehabilitation device based on the determined signals.

77. The kit of parts of claim 76, comprising a housing for containing the sensor units and/or the actuators.

78. The kit of parts of claim 76 or 77, comprising one or more linkages for connecting and supporting the sensor units.