Smart garment and artificial intelligence (AI) system for biomechanics assessment

Abstract

The present invention relates to a biomechanic assessment wearable including: material worn by a wearer; electrodes for use in monitoring the wearer; and electronics devices for receiving information from the electrodes. Each electronics device includes inertial measurement unit (IMU) and electromyography (EMG) sensors.

Description

SMART GARMENT AND ARTIFICIAL INTELLIGENCE (Al) SYSTEM FOR BIOMECHANICS ASSESSMENTTECHNICAL FIELD

[0001] The present invention generally relates to a wearable for assessment of biomechanics.BACKGROUND

[0002] The reference to any prior art in this specification is not and should not be taken as an acknowledgement or any form of suggestion that the prior art forms part of the common general knowledge.

[0003] Biomechanical assessment systems are known.

[0004] Optical motion capture systems, recognised as the gold standard for biomechanical assessment, require extensive setup and calibration, limiting their practicality for clinical and at-home use. Even advanced systems incorporating multisensor inertial measurement units (IMUs) involve significant setup due to sensor placement and synchronisation. Moreover, the existing systems also have limited biofeedback capabilities.

[0005] Current IMU systems focus primarily on estimation of motion (kinematics) and do not provide measurement, estimation, or analysis of important biomarkers of musculoskeletal health / disease / injury risk. IMU systems also fail to provide kinetic analysis (body forces, external forces, joint moments) that is an important intermediary step to computing key biomarkers (applied forces, stresses, strains) of musculoskeletal tissue health (bone, muscle, tendon, ligament, cartilage).

[0006] When analysing musculoskeletal forces / stresses / strains, estimations relying solely on kinematics and kinetics fail to account for neural control of movement that is crucial to muscle force estimation (and thus bone / ligament / tendon / cartilageforces / stresses / strains estimation). Therefore, advanced biomechanical assessment systems that use IMU should be paired with electromyography (EMG) sensors, which further increases setup time and system complexity.

[0007] EMG systems are typically standalone systems that require individual placement of sensors by an experienced biomechanist, sports scientist, or clinician. Moreover, most existing gold-standard EMG systems with multiple sensors rely on single-use gel electrodes, which can be inconvenient, time-consuming to apply, and may cause skin irritation over prolonged use.

[0008] Smart garment systems that incorporate both IMU and EMG sensors often lack sufficient sensor coverage and configuration. Even if these systems record EMG from multiple sites, their IMUs are typically confined to a central unit, which limits their ability to capture specific biomechanical measurements needed to estimate individual muscle forces and strains. Additionally, the quality of EMG data from these systems is often poor compared to the gold standard gel-based systems, resulting in less accurate and reliable measurements of muscular activity.

[0009] Estimation of kinetics and muscle / joint / ligament / bone / cartilage forces / stresses / strains is computationally expensive, and technically challenging, even with accurate estimation or measurement of motion and muscle excitation. Current physics-based models require lengthy data pre-processing steps, careful calibration, and computationally expensive analysis that requires guidance by experienced biomechanists. Additionally, current physics-based approaches are not well suited to wearable systems (IMU) that record sparser and noisier data than their lab-based counterparts (marker-based optical motion capture systems).

[0010] Artificial intelligence (Al) offers a solution that can estimate kinetics and muscle / joint / ligament / bone / cartilage forces / stresses / strains from wearable systems (IMU and EMG) that can be used outside the laboratory and can operate in real-time with little to no user input. However, current Al methods have shown poor generalizability to new people and movements, lack interpretability, and do not consider neurological, biomechanical and physiological processes that explain human movement.

[0011] Physics-informed neural networks (PINN) have been proposed to bridge the gap between physics-based and data-driven models. Unfortunately, PINN are difficult to deploy beyond “toy” problems and have seen little translation to practical applications. Integrating the concepts of PINN (minimising loss functions based on physical laws) in Al models for biomechanical assessment will improve generalizability and allow the development of models that can operate with data from wearable systems while maintaining the neurological, biomechanical and physiological processes that underpin established physics-based models.

[0012] There is a need for portable biomechanics assessment system that can estimate key biomarkers of musculoskeletal tissue health / disease / injury. Combining an IMU and EMG embedded smart garment with a PINN enables accurate and robust estimation of key biomarkers, in real-time, from a low-cost and easy-to-use system.SUMMARY OF THE INVENTION

[0013] According to one aspect of the present invention, there is a biomechanical assessment wearable provided, including: material worn by a wearer; textile electrodes for monitoring the wearer; electronic devices configured to capture signals from the electrodes, including electromyography (EMG) sensors; and additional sensors integrated into the electronic devices for measuring physiological parameters and motion, such as, but not limited to, inertial measurement units (IMU) and other relevant sensors.

[0014] Preferably, electronics devices include both IMU and EMG sensors distributed at multiple sites in the wearable for improved biomechanical assessment. Each electronics device may be interfaced with at least two electrodes to receive information.

[0015] The wearable may include an electronics hub for receiving data from the electronics devices and transmitting data to an external receiver unit. Data from the electronics' devices can be received via wired connections or through wireless communication. Advantageously, the hub may perform on-board processing toattenuate signal noise, reduce data packet size, store data locally and enable transmission of the data from the electronics devices. The electronics hub may include a radiofrequency (RF) transmission module, incorporating an integrated antenna and other necessary components for transmitting data to a monitoring station which may have ICT architecture for enabling cloud storage of the data.

[0016] The electrodes may be embedded in the material via a heat transfer process. Each electrode may include a snap fastener assembly. The electrodes may be washable. Each electronics device may be coupled to more than one electrode. Each electronics device may be positioned above its associated electrodes. Preferably, the electronics devices are in communication with a hub device, which may occur via wired or wireless communication.

[0017] Each electronics device may include: a housing for housing the IMU, EMG, and any additional sensors, and a dock to interface with the sensor housing, preferably for a snap-fit assembly and easy removal when desired. The dock may be embedded in the material via a heat transfer process.

[0018] Each electronics device may include one or more additional physiological sensors for sensing physiological parameters (i.e., heart rate, temperature, respiration).

[0019] The electrodes may be used to stimulate the wearer. The stimulation may activate the muscles for an improved exercise routine (e.g., EMS), or to suppress pain (e.g., TENS). These stimulation functions can also be part of a biofeedback system, providing real-time feedback to the wearer. In addition to TENS, the garment may incorporate vibration motors to deliver vibration-based biofeedback. To facilitate these functionalities, the wearable may include an integrated electronics device or utilise the electronics hub to control and manage the stimulation and biofeedback processes. This device or hub ensures proper signal generation, modulation, and delivery for effective EMS, TENS, and vibration-based feedback.

[0020] Preferably, the wearable includes athletics wear. The material may be tight fitting. Preferably, the tight, formfitting material hugs the wearer. Examples of such wearables include, but are not limited to, smart garments for the lower body (e.g., leggings, pants, shorts), smart garments for the upper body (e.g., shirts, t-shirts) andbespoke applications (e. g., specifically designed sleeves for monitoring knee and / or elbow biomechanics).

[0021] According to another aspect of the present invention, there is provided an artificial intelligence (Al) system including: the wearable; and an artificial intelligence (Al) processor that uses data from the wearable.A database of gold-standard laboratory data, which has been analysed by expert biomechanists, is used to train Al models for different applications.

[0022] The processor may utilise one or more Al models when processing data. The Al models may compute digital biomarkers, in real-time, from data acquired from the wearable. The biomarkers may be reported to the wearer in real-time and / or processed to provide training instruction to the wearer. The Al models may determine kinematic kinetic, muscle, joint, bone, tendon, and / or ligament variables.

[0023] The processor may further process personalised information about the wearer. The processor may utilise a database of laboratory (optical motion capture, force plates, gel-electrode EMG) and wearable (IMU portable gel or dry-electrode EMG) measurements.

[0024] The Al processor may be trained using conventional loss functions and / or physics-based loss functions. The Al processor may be trained with, and incorporate, muscle-tendon unit (MTU) kinematics solvers.

[0025] The Al processor may provide feedback to optimise digital biomarkers to promote tissue healing and reduce injury risk for the wearer. The Al processor may provide personalised recommendations to achieve optimal modifications for the wearer.

[0026] The Al processor may provide, to the wearer, real-time feedback via audio prompts, haptic vibration in the wearable, visual feedback, or low-intensity electrical stimulation in the wearable.

[0027] According to another aspect of the present invention, there is provided a biomechanic assessment wearable including: material worn by a wearer;electrodes for use in monitoring the wearer; and electronics devices for receiving information from the electrodes, each electronics device including inertial measurement unit (IMU) and electromyography (EMG) sensors.

[0028] Any of the features described herein can be combined in any combination with any one or more of the other features described herein within the scope of the invention.BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Preferred features, embodiments and variations of the invention may be discerned from the following Detailed Description which provides sufficient information for those skilled in the art to perform the invention. The Detailed Description is not to be regarded as limiting the scope of the preceding Summary of the Invention in any way. The Detailed Description will make reference to a number of drawings as follows:

[0030] Figure 1 is a schematic view showing artificial intelligence (Al) system including a biomechanical monitoring lower body smart garment in accordance with an embodiment of the present invention;

[0031] Figure 2 shows various views including an electrode for interfacing with an electronics device of the lower body smart garment of Figure 1 ;

[0032] Figure 3 shows various views including the electronics device of Figure 2;

[0033] Figure 4 shows fastening of the electronics device of Figure 3;

[0034] Figure 5 is a block diagram of an artificial intelligence (Al) processor of the system of Figure 1 ;

[0035] Figure 6 shows various biomechanical wearable types and electrode placement for use in the system of Figure 1 ; and

[0036] Figure 7 shows further various biomechanical wearables and electrode placement for use in the system of Figure 1 ;

[0037] Figure 8 shows an alternative snap-fit electronics device; and

[0038] Figure 9 shows an alternative electronics device with two electrodes.DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

[0039] According to an embodiment of the present invention, there is provided an artificial intelligence (Al) system 100 as shown in Figure 1. The Al system 100 includes a biomechanical assessment lower limb smart garment 102 (i.e. a wearable). The Al system 100 further includes an Al processor 104 for processing biomechanical data 106 received from the lower limb smart garment 102.

[0040] The lower limb smart garment 102 includes tight form-fitting material 108 worn by and hugging a wearer. Concealed inner electrodes are borne by the material 108 and used in monitoring the wearer during activity. The lower limb smart garment 102 includes distributed electronics devices 110 borne by the material 108 and for receiving sensed electrical-signal information from the electrodes. Each electronics device 110 includes inertial measurement unit (IMU) and electromyography (EMG) sensors.

[0041] Advantageously, the electronics devices 110 include both IMU and EMG sensors distributed at multiple sites on the lower limb smart garment 102 for improved biomechanical assessment. The number of electronics devices 110 of the lower limb smart garment 102 is scalable, and records IMU and EMG quality and quantity comparable to dedicated, standalone gold-standard systems.

[0042] The lower limb smart garment 102 includes an electronics hub 116 that controls and synchronises the collection of biomechanical data from multiple electronic devices 110 distributed across the garment. The hub 116 acts as a controller for the wired and wireless communication, managing data transmission between the sensors and coordinating real-time data collection. It performs on-board processing to optimise data handling, which includes reducing packet size and enabling efficient transmission of data. Additionally, the hub 116 may oversee the delivery of biofeedback through any additional sensors integrated into the system. The processed biomechanical data 106 is transmitted from the hub 116 to the Al processor 104 at a monitoring station.Additionally, the hub electronics device may act as the Al processor itself enabling onboard data processing and biofeedback delivery.

[0043] The architecture of the system 100 incorporates both wired and wireless connections for data transmission between the hub 116 and Al processor 104, allowing integration of multiple sensors, including additional wireless IMU modules that can be placed on any anatomical landmarks.

[0044] Advantageously, the lower limb smart garment 102 is washable, reusable, and comfortable. The slim-line design allows the lower limb smart garment 102 to be worn alone or beneath clothes, enabling biomechanical data acquisition during day-to- day activities.

[0045] Turning to Figure 2, each electronics device 100 is positioned directly above or in close proximity to, and is coupled to receive information from its associated electrodes 200. Each electronics device may be associated with two bipolar electrodes whilst a ground electrode for the entire system is situated at the hip.

[0046] Rubber-like padded material 202 and conductive fabric 204 are laser cut and heat pressed onto the garment material 108. A snap fastener assembly 206 is fixed to the garment material 108 contacting the conductive textile fabric 204 in contact with the wearer’s skin. A wire 208 fixed to the snap fastener assembly 206 carries electricalsignal information to the EMG sensor of the electronics device 100. Additional non- conductive fabric 210 isolates the snap fitting base 212 from the skin.

[0047] The electrodes 200 include conductive textiles 204 and are laser cut and embedded in the garment material 108 via a heat transfer process for measuring electromyography with the EMG sensor of the electronics device 100.

[0048] The electrodes 200 are embedded in the garment material 108 via a heat transfer process using an iron / heat press. First a stretchable rubber-like material 202 is heat pressed onto the compression fabric substrate material 108. The conductive textile electrode 204 is then ironed on top. The primary stretchable, rubber-like heat transfer material 202 acts to isolate the electrode conductive fabric 204, increase friction to limit slip on the skin and provides a thick padded underlay to improve electrode-skin contact impedance and improve signal quality.

[0049] The electrodes 200 are washable and comparable in functionality with traditional Ag / AgCI gel electrodes during dynamic motor tasks. The electrodes 200 can be made for any shape / size depending on the application, i.e., circular bipolar electrodes, rectangular bar-type electrodes, or multi-electrode arrays. The textile electrodes 200 are interfaced with the electronics device 100 via a snap fastener / crimping technique including a wire 208 soldered to the snap fastener 206 to carry to signal from the conductive textile material 204 to the EMG sensor through a receptacle board which incorporates pogo pins 216.

[0050] Electrodes 200 may contain silver, graphene, carbon or PEDOT:PSS, and may be manufactured via a screen-printing process which perform comparably yet have longer life spans after washing.

[0051] Turning to Figure 3, the wearable electronic devices 110 (including EMG andIMU sensors) are embedded in the lower limb smart garment 102 via an electronics housing 112 designed to enable the internal sensors to be removed for washing and also used in other applications (i.e., wired, wireless and multi-use applications).

[0052] The EMG and IMU sensors are housed inside a 3D printed, or injected moulded, nylon-like resin / plastic housing 112. The removeable electronics housing 112 interfaces with a 3D printed dock 114 which is embedded in the lower limb smart garment 102 via a heat transfer process.

[0053] The dock 114, embedded in the lower limb smart garment 102, housees the receptacle board 216 which acts as an interconnect between other receptacle boards and interfaces with several sensors throughout the lower limb smart garment 102. Wired connections 300 run through the lower limb smart garment 102 that are sewn in place on the material 108 making connections to the receptacle boards 216. The receptacle boards 216 have pogo pins allowing them to make connection to the peripheral devices in their housings 112 when they snap into place.

[0054] When plugged in, the EMG and IMU sensors 302 of the electronic devices 110 make electrical connections (power, ground, data, etc.) and the sensors 302 stream data to the main hub 116 of the lower limb smart garment 102. The electronics device housing 112, includes a base 304 and a lid 306, removable with a tool.

[0055] The communication between the sensors 302 embedded in the lower limb smart garment 102 is enabled using wired data transmission, linking the docks 114 via a daisy-chain of wires 300. The hub electronics device 116 acts as the central communication point, retrieving data from all peripheral electronic devices, including the EMG and IMU sensors 302.

[0056] The conductors 300 may be washable traces containing silver, graphene, carbon or PEDOTPSS. The sensors 302 will be removed for washing. Also, the conductor traces 300, docks 114, receptacle boards 216 and connections can be washable so the smart garment 102 can be washed upon removal of the electronic devices.

[0057] The design of the housing 112 can enable the use of wearable sensors 302 as individual wireless nodes distributed across the body of the wearer. To enable this functionality, additional features are incorporated, such as a dock 114 which contains a standalone power supply (management of the power distribution and a rechargeable battery). A radiofrequency (RF) transmission module is part of the electrical design of the sensors 302, which can be used when they serve as individual wireless nodes.

[0058] This functionality is also enabled within the design of the housing 112 of the hub 116 similarly integrated in the lower limb smart garment 102. The acquired biomechanical data from the sensors 302 is aggregated at the hub electronics device 116 which transmits it wirelessly to a remote receiver of the processor 104. Another smart device (e.g., a smart phone, a tablet, or a dedicated receiver unit) or a personal computer can be used as the processor 104. Additionally, the hub electronics device may act as the processor. An ICT architecture enabling cloud storage of the data may be provided.

[0059] The system 100 supports data transmission using existing wireless technologies including Bluetooth, Wi-Fi, and proprietary RF protocols. These technologies comply with accepted frequencies for industrial and medical applications and adhere to the standards of wireless body area networks communication.

[0060] The system architecture allows for the integration of various wireless sensors apart from the IMU and EMG sensors 302. For instance, for enhanced analysis of the physiological state of the user, additional sensors such as heart rate, temperature, etc.,can be included. The functionality of the system 100 can be extended through sensory biofeedback, for example haptic (using actuators) or transcutaneous electrical nerve stimulation (TENS), using existing electrodes 110 in the lower limb smart garment 102 via modulated electric current pulses. The electrodes 200 can be used to stimulate the wearer. The stimulation may activate the muscles for an improved exercise routine (e.g., EMS), or to suppress pain (e.g., TENS). To facilitate these functionalities, the lower limb smart garment may include an integrated electronics device or utilise the electronics hub to control and manage the stimulation and biofeedback processes. This device or hub ensures proper signal generation, modulation, and delivery for effective EMS, TENS, and vibration-based feedback.

[0061] Turning to Figure 4, the dock 114 is embedded in the material 108 of the lower limb smart garment 102 via a heat transfer process using an iron 400. An ironed collar 402 embeds the dock 114 in the garment material 108. The housing 112 can be removed from the dock 114, and the removed housing 112 can then be docked for use in another like dock 114 on a strap 404 or shoe 406. In this manner, the sensors 302 can be removed from the lower limb smart garment 102 and reused in wireless applications with straps 404 to easily fix to a person (i.e., in clinics with health professionals or outside of typical laboratory environments for data collection in the field).

[0062] Turning to Figure 5, biomechanical data is streamed from the central hub device 116 of the lower limb smart garment 102 to the Al processor 104 for analysis.

[0063] A database 500 of high-fidelity biomechanical data is analysed by experts in computational biomechanics. The analysis uses neuromusculoskeletal modelling 502 to compute digital biomarkers of musculoskeletal health and performance. The digital biomarkers 504 provide reference for the development and training of physics-informed Al models 506 that include physics-based loss functions that constrain and regularise the algorithms, and incorporate physical and physiological models of neuromuscular function. The database includes marker-based optical motion capture, ground reaction forces, EMG, and IMU data recorded in a laboratory environment. The database includes data from healthy and pathological populations performing different tasks (incl. walking, running, squatting, stair climbing, jumping, lowering and rising from a chair, side-step cutting, lunging, split-squatting, and heel-raising). The database includesneuromusculoskeletal models used to analyse kinematics, kinetics, and muscle / bone / joint / ligament / tendon / cartilage forces / stresses / strains for various joints, and for various conditions (osteoarthritis / ligament injury / tendinopathies). The Al models 506 are developed to compute digital biomarkers, in real-time, from the biomechanic data acquired from the multi-sensor lower limb smart garment 102. The Al models 506 can be deployed and computed via garment-embedded processors at the central hub electronics device 116, on smart devices (e.g., phones tablets), or via cloud-based processing. The outputted biomarkers 508 are reported to the wearer in real-time via multiple feedback modalities and / or are processed to provide training instruction to the wearer, and / or to provide a summary of digital biomarkers to the user.

[0064] The collected IMU and EMG biomechanical data is combined with wearer information 510 (e.g., height, weight) as input to Al models 506. Model optimization and compression 512 may be performed.

[0065] The Al algorithms support input of additional user information including, but not limited to, body segment measurements, body segment inertial properties, 3- dimensional body scans, medical imaging (e.g., MRI, CT, ultrasound), muscle strength tests. The Al algorithms use additional information to improve output accuracy (e.g., estimates of joint loading) or add functionality (e.g., predict cartilage strain).

[0066] Al models provide real-time estimation of digital biomarkers (e.g., joint forces / tissue strains) in addition to kinematic and kinetic variables (e.g., joint motion and torque).

[0067] Al algorithms are developed from a database of laboratory quality data that has been processed and analyzed by experts in computational biomechanics. The database is non-public and included 3-dimensional optical motion capture, ground reaction forces, EMG, IMU, and medical imaging data. Data analysis includes rigid-body physics models, muscle driven-simulation and finite element modelling to solve the linear and angular kinematics, linear and angular dynamics, joint powers, muscle forces, joint contact forces, ligament forces / stresses / strains, tendon forces / stresses / strains, cartilage forces / stresses / strains and bone forces / stresses / strains.

[0068] Al algorithms are developed, trained, and validated in TensorFlow and / or PyTorch, both free-to-use open-source platforms and are made robust and reliable via the combination of physics-based models and data-driven approaches.

[0069] The Al algorithms consist of, but are not limited to, artificial neural networks of multiple types, including feed-forward neural networks, convolutional neural networks, long-short memory recurrent neural networks, gated recurrent unit networks, transformers, auto-encoders, generative adversarial networks, Kolmogorov-Arnold networks.

[0070] The Al training process includes conventional loss functions (e.g., minimise error terms such as mean-squared-error between predicted and ground-truth variables, and physics-based loss functions (e.g., derived from Newtonian equations of motion, muscle contraction dynamics, neural activation).

[0071] The Al algorithm includes pre-trained muscle-tendon unit (MTU) kinematics solvers. The pre-trained (MTU) kinematics solvers are developed from musculoskeletal simulation and provide rapid estimation of muscle-tendon lengths, moment arms, lines of action.

[0072] The system 100 provides feedback to optimise digital biomarkers to promote tissue healing, and reduce injury risk (e.g., modify joint loading during walking to promote cartilage health in individuals with hip OA).

[0073] The system 100 provides personalised recommendations to achieve optimal modifications. To achieve this, the system 100 tracks how a user’s changes to kinematics, kinetics, and muscle excitations affect digital biomarkers.

[0074] Data is logged and analysed for long-term changes to digital biomarkers and provides additional metrics such as cumulative loading and tissue strains.

[0075] Real-time feedback can be provided via audio prompts (from smart phone), haptic vibration (in the lower limb smart garment 102), visual plots / chart / numbers (smart phone / tablet / computer), or low-intensity electrical stimulation (in the lower limb smart garment 102).

[0076] In summary, for accurate tracking of motion using wearable sensors 302, the system 100 preferably includes at least one IMU for each segment of the body that the lower limb smart garment 102 covers (e.g., pelvis, thigh, shank) and covers enough segments for a given application (e.g., knee biomechanics analysis includes a minimum of two sensors 302 placed on a thigh and a shank respectively).

[0077] The system 100 includes adequate EMG measures to record the unique muscle excitation solutions necessary for accurate estimation of internal forces.

[0078] The garment 102 is easy to don and doff, require little to no setup or skin preparation, and is reusable with washable fabrics. The garment 102 incorporates dry textile electrodes, which can be embedded into clothing and provide comfortable and convenient means of measuring muscle activity.

[0079] The lower limb smart garment 102 is comfortable, wearable under regular clothing, and does not impede typical daily activities. The lower limb smart garment 102 is suitable for exercise (i.e., lightweight, thin, disperse heat).

[0080] To inform clinical practice, rehabilitation, or injury prevention, the system 100 provides clinically relevant analysis and feedback. To achieve this, the system 100 for biomechanical analysis can:• Compute tissue-level forces, stresses, and / or strains that directly impact tissue health, and drive tissue adaptation (e.g., hip contact forces in people with hip osteoarthritis).• Compute external biomechanics (e.g., kinematics and kinetics) that can be used to describe motion and aid in feedback and modification of movement / exercise to change tissue level variables.• Operate in real-time to enable biofeedback assisted training and modification of movements / exercises.• Be user friendly and require little to no training.

[0081] Known wearable systems do not provide adequate analysis, ease-of-use, and feedback to inform clinical practice. Known systems have the following limitations:• When multiple IMU are used, are limited to kinematics analysis, which has limited clinical significance and should not be used to infer or predict internal tissue biomechanics.• When EMG are used, provides analysis of only muscle excitation (i.e., how much neural activity is being sent to the muscle). This metric, and the many metrics built on this (e.g., muscle work) do not account for the complex relationship between kinematics, muscle excitation, and force output. Thus, how musculoskeletal tissues are affected is not analysed.• When laboratory data is collected (e.g., optical motion capture, ground reaction forces), current data analysis methods require lengthy data cleaning / processing and digital modelling to compute relevant digital biomarkers. This process requires expertise in computational biomechanics and can take several days to process data for a single subject.

[0082] Instead, the present system 100 addresses limitations of known systems and provides the following benefits:• Utilises a smart garment which can be easily donned without supervision of an expert, uses washable textile electrodes and contains EMG sensors and IMU sensors which can record muscle excitations and joint kinematics comparable to research grade systems.• Combines Al methods with physics-based computations to compute relevant biomarkers (e.g., tissue forces / stresses / strains). This approach is a first for biomechanics and addresses issues with Al reliability, allowing its implementation in ecologically valid environments.• Is developed using thousands of hours of data from a world-class biomechanics research laboratory. The high-fidelity data combined with expertise of the inventors in computational modelling has enabled the development of algorithms to compute internal biomechanics, exceeding the limited capabilities of competitors.• Data analysis is completely automated and does not require any expertise or manual data cleaning, curating, or processing.• Algorithms can be deployed on smart devices and enable real-time computation of digital biomarkers for biofeedback-informed training or movement modification that take into account internal biomechanics.• Data analysis is logged and computes simple metrics, providing the basis for personalised guidance that can be easily accessed and applied by the end-users (e.g., gait modification recommendations to optimise joint loading).

[0083] The system 100 provides a new means to measure digital biomarkers during activities of daily living. The measured biomarkers are associated to the health of musculoskeletal tissues, including but not limited to cartilage, joints, bones, muscles, tendons, and ligaments. The system 100 integrates a wearable such as lower limb smart garment 102, artificial intelligence, and biofeedback.

[0084] The lower limb smart garment 102 comprises of tight-fitting apparel material 108 with embedded electronics 110 and fabric electrodes 200 to measure key physiological parameters (motion and muscle activity) during human movement.

[0085] The artificial intelligence (Al) processor 104 processes the measured data to calculate all relevant digital biomarkers, including but not limited to tendon and ligament forces / stresses / strains, muscle forces / lengths / velocity / stiffness / power, cartilage forces / stresses / strains, bone forces / stresses / strains, joints angle / velocity / moments / power / stiffness.

[0086] The biofeedback provides information to the wearer regarding the measured biomarkers. The biofeedback can be delivered in real-time. Delivery modalities include, but are not limited to, visual, plots / charts / numbers, audio prompts, haptic vibration, and low intensity electrical stimulation.

[0087] A person skilled in the art will appreciate that many embodiments and variations can be made without departing from the ambit of the present invention.

[0088] The present invention has application to a range of wearables in Figure 6 including:• Pants (Training and rehabilitation for tissue regeneration of lower limb musculoskeletal injuries, muscle, tendon ligament, etc. - i.e., Achilles tendon rupture, ACL rupture, hamstring tear, etc.). Pants also offer general rehabilitation metrics (e.g., joint ROM / power, muscle force / power) and tests (e.g., timed-up- and-go, 6-minute walk test, stair climb) for elder care.• Shorts (Hip osteoarthritis)• Shirt (Upper body / limb musculoskeletal injuries)• Sleeve (Tennis elbow injuries, upper limb injuries).

[0089] The electronics devices 110 can be used in not only wired applications of Figure 6, but also wireless applications in Figure 7 enabling the use of smart garment for improved usability, personalisation and data acquisition. This includes outdoor data collection, and use in clinics for patient rehabilitation / retraining / feedback.

[0090] Turning to Figure 8, the snap-fit housing 112 can be removed from the dock 114.

[0091] Turning to Figure 9, each electronics device 100 can include two electrodes 200.

[0092] The IMU sensors can be comprised of a 3-axis accelerometer, 3-axis gyroscope, and 3-axis magnetometer, providing 9DOF metrics when tracking the movement of the user in 3-D space.

[0093] The EMG sensors capture both positive and negative portions of EMG signal using a single-power supply. The EMG sensors include cascaded high pass and low pass filters to reduce presence of motion artefact and prevent aliasing when digitised by a microcontroller.

[0094] The EMG sensors include a microcontroller which has a high-precision deltasigma analog-to-digital-converter (ADC) capable of digitising analog signals at resolutions (8-20 bits) and sample rates (8 - 384000 Hz).

[0095] The EMG sensors include a microcontroller which has a 256-input hardware multiplexer (MUX) and sequencing successive approximation ADC capable of switching between input channels with low latency.

[0096] The EMG sensors include a microcontroller which has on-board digital filtering blocks which can be configured to remove signal artefacts (e.g. motion artefact, stimulation artefact) improve the signal-to-noise (SNR) ratio of the recorded EMG signal and output EMG linear envelopes using edge-computing.

[0097] The microcontroller used can be configured to use multiple different communication protocols including but not limited to: UART, I2C and SPI.

[0098] Electrodes 200 are located on the soft tissues of the muscles, while the shape and the size of the electrodes 200 may vary depending on the activity, region of interest or application. A minimum of two electrodes is included for each EMG sensor. Arrays of electrodes 200 can be included in the garment 102 for collecting EMG data and personalising electrode placement for best performance with different users.

[0099] The IMU sensors obtain raw motion sensors data and perform on-board processing to provide information on the orientation and position of the device 110. The on-board processing algorithm accounts for the drifting errors, and / or magnetic interference. The hardware design of the IMU sensors includes a microcontroller and an RF-module for enabling seamless integration with the wired and wireless networks.[000100] The wireless transmission from the devices 110 (see Fig. 7) is enabled by the electrical design of the sensors 302. This design may incorporate various types of transmitting antennas, including but not limited to a flat printed circuit board (PCB) antenna, a 3D PCB antenna, an external antenna connected through a UFL connector, or an off-board PCB antenna with matching circuitry. The transmission protocol support high-speed (with configurable data-rate) wireless transmission with low latency and low data loss. The system 100 supports data transmission using existing wireless technologies including but not limited to Bluetooth, Bluetooth Low Energy (BLE), Wi-Fi, near field communication and proprietary RF protocols that comply with accepted frequencies for medical applications, the ISM band, and standards of wireless body area networks communication. The central hub 116 collects the biomechanical data acquired by the sensors 302 embedded in the garment 102. It then synchronously transmits this data to the receiver 104, which is connected to the individual wireless units distributed across the body.[000101] Tissue-level biomechanics analysis is conducted with the use of a network of wearable sensors 302 embedded in the wearables 102. The analysis makes use of muscle-tendon unit properties derived from individual characteristics to inform the Al model. The analysis involves inclusion of pre-developed muscle-tendon unit kinematics solver in the Al framework. The analysis involves the combination of IMU, EMG, and Al / neural networks for real-time estimation of tendon and ligament forces / stress / strains,muscle forces / lengths / velocity / stiffness / power, cartilage forces / stress / strains, bone forces / stress / strains. The analysis makes use of physics / neuromuscular dynamicsbased loss functions for training Al / neural networks to predict digital musculoskeletal biomarkers. The analysis makes use of a combination of IMU and EMG embedded smart garment wearables 102 with Al / neural networks models for real-time estimation of tendon and ligament forces / stress / strains, muscle forces / lengths / velocity / stiffness / power, cartilage forces / stress / strains, bone f o rces / stress / strai n s .[000102] In compliance with the statute, the invention has been described in language more or less specific to structural or methodical features. It is to be understood that the invention is not limited to specific features shown or described since the means herein described comprises preferred forms of putting the invention into effect.[000103] Reference throughout this specification to ‘one embodiment’ or ‘an embodiment’ means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, the appearance of the phrases ‘in one embodiment’ or ‘in an embodiment’ in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more combinations.

Claims

The claims defining the invention are as follows:1 . A biomechanic assessment wearable including: material worn by a wearer; electrodes for use in monitoring the wearer; and electronics devices for receiving information from the electrodes, each electronics device including inertial measurement unit (IMU) and electromyography (EMG) sensors.

2. A biomechanic assessment wearable as claimed in claim 1 , wherein the electronics devices include both IMU and EMG sensors distributed at multiple sites in the wearable for improved biomechanical assessment.

3. A biomechanic assessment wearable as claimed in claim 1 , wherein each electronics device is interfaced with at least two electrodes to receive information.

4. A biomechanic assessment wearable as claimed in claim 1 , including an electronics hub for receiving data from the electronics devices and transmitting data to an external receiver unit.

5. A biomechanic assessment wearable as claimed in claim 4, wherein: data from the electronics devices can be received via wired connections or through wireless communication; or the hub performs on-board processing to attenuate signal noise, reduce data packet size, store data locally and / or enable transmission of the data from the electronics devices.

6. A biomechanic assessment wearable as claimed in claim 1 , wherein the electrodes are embedded in the material via a heat transfer process and are washable, with each electrode including a snap fastener assembly.

7. A biomechanic assessment wearable as claimed in claim 1 , wherein each electronics device includes: a housing for housing the IMU and EMG, and a dock to interface with the housing, preferably for a snap-fit assembly and easy removal when desired.

8. A biomechanic assessment wearable as claimed in claim 1 , wherein each electronics device includes one or more additional physiological sensors for sensing physiological parameters.

9. A biomechanic assessment wearable as claimed in claim 1 , wherein the electrodes are used to stimulate the wearer, the stimulation activating the muscles for an improved exercise routine or to suppress pain, the stimulation functions being part of a biofeedback system providing real-time feedback to the wearer.

10. A biomechanic assessment wearable as claimed in claim 1 , incorporating vibration motors to deliver vibration-based biofeedback.

11. A biomechanic assessment wearable as claimed in claim 1 , wherein the wearable includes athletics wear and the material is tight fitting to hug the wearer.

12. An artificial intelligence (Al) system including: the wearable of claim 1 ; and an artificial intelligence (Al) processor that uses data from the wearable.

13. An artificial intelligence (Al) system as claimed in claim 12, wherein the processor utilises one or more Al models when processing data, the Al models computing digital biomarkers, in real-time, from data acquired from the wearable.

14. An artificial intelligence (Al) system as claimed in claim 13, wherein the biomarkers are reported to the wearer in real-time and / or processed to provide training instruction to the wearer.

15. An artificial intelligence (Al) system as claimed in claim 12, wherein the processor processes personalised information about the wearer and utilises a database of laboratory and / or wearable measurements.

16. An artificial intelligence (Al) system as claimed in claim 12, wherein the Al processor is trained using conventional loss functions and / or physics-based loss functions, the Al processor preferably being trained with, and incorporating, muscletendon unit (MTU) kinematics solvers.

17. An artificial intelligence (Al) system as claimed in claim 12, wherein the Al processor provides: feedback to optimise digital biomarkers to promote tissue healing and reduce injury risk for the wearer; personalised recommendations to achieve optimal modifications for the wearer; and / or to the wearer, real-time feedback via audio prompts, haptic vibration in the wearable, visual feedback, or low-intensity electrical stimulation in the wearable.

18. A biomechanic assessment wearable including: material worn by a wearer; dry electrodes for monitoring the wearer; electronic devices configured to capture signals from the electrodes, including electromyography (EMG) sensors; and additional sensors integrated into the electronic devices for measuring physiological parameters and motion, such as, but not limited to, inertial measurement units (IMU) and other relevant sensors.

19. A system comprising of wearable sensors embedded in the smart garment and at least one wireless sensor placed on the body of the user: a. The sensors embedded in the smart garment must include but not limited to surface electromyography (sEMG) sensors for detecting and measuring muscle activity signals, and motion tracking sensors represented by the inertial measurement units (IMUs). b. The sEMG signals are recorded using the electrodes made of conductive textile materials embedded in the garment. c. The garment may include one or multiple individual clothing items, such as leggings, shorts, sleeves, shirts etc., made of elastic materials. d. The IMU sensors comprise of a 3-axis accelerometer, 3-axis gyroscope, and 3-axis magnetometer, providing 9DOF metrics when tracking the movement of the user in 3-D space. e. The architecture of the network of sensors embedded in the garment includes central and peripheral devices communicating via wired protocol. f. Each of the sensors is enclosed in a custom-designed 3D printed housing (may be injection molded) designed in such a way that enables electricalconnectivity between multiple different devices, allowing for sensors to be easily detachable and reusable across several applications, both wired and wireless. g. Housing docks are embedded in the garment and trace connections are made between receptacle boards in the docks linking the network of sensors. h. The biomechanical data acquired by the garment is wirelessly transmitted to a dedicated receiver using configurable transmission protocol. i. Wireless sensors are enclosed in housings with alternative dock design which includes an individual power-supply, such as a rechargeable battery. j. The sensors are seamlessly interchangeable as separate wireless nodes or part of a wired system embedded in a garment. k. One or more actuators may be included in the garment to provide sensory feedback to the user, such as haptic response. Existing electrodes and microcontrollers incorporated in the sensors can be programmed to provide transcutaneous electrical nerve stimulation.

20. The system in Claim 19 with the sEMG sensors including an amplifier and filtering components for enhancing the quality of acquired muscle potential information: a. The sEMG sensors circuitry has been designed to capture both positive and negative portions of sEMG signal using a single-power supply. The sEMG sensors circuitry includes cascaded high pass and low pass filters to reduce presence of motion artefact and prevent aliasing when digitised by the microcontroller. b. The sEMG sensors circuitry design includes a microcontroller which has a high-precision delta-sigma analog-to-digital-converter (ADC) capable of digitising analog signals at resolutions (8-20 bits) and sample rates (8 - 384000 Hz). c. The sEMG sensors circuitry design includes a microcontroller which has a 256-input hardware multiplexer (MUX) and sequencing successive approximation ADC capable of switching between input channels with low latency.d. The sEMG sensors circuitry design includes a microcontroller which has on-board digital filtering blocks which can be configured to improve the signal-to-noise (SNR) ratio of the recorded sEMG signal and also output sEMG linear envelopes using edge-computing. e. The microcontroller used can be configured to use multiple different communication protocols including but not limited to: UART, I2C and SPL21. The system in in Claim 19 wherein the sEMG sensors are manufactured via a laser cutting process and embedded in the garments via a heat transfer method yet may also be screen printed using conductive inks. Electrodes are located on the soft tissues of the muscles, while the shape and the size of the electrodes may vary depending on the activity, region of interest or application: a. A minimum of two electrodes must be included for each EMG sensor. b. Arrays of electrodes may be included in the garment for collecting sEMG data and personalising electrode placement for best performance with different users.

22. The system in in Claim 19 wherein the IMU sensors allow to obtain raw motion sensors data and perform on-board processing to provide information on the orientation and position of the device: a. The on-board processing algorithm accounts for the drifting errors, and / or magnetic interference: b. The hardware design of the IMU sensors includes a microcontroller and an RF-module for enabling seamless integration with the wired and wireless networks;23. The system in in Claim 19 wherein the wireless transmission is enabled by the electrical design of the sensors, which incorporates on-board printed circuit board antennas: a. The transmission protocol support high-speed (with configurable data- rate) wireless transmission with low latency and low data loss; b. The system supports data transmission using existing wireless technologies including but not limited to Bluetooth, Bluetooth Low Energy (BLE), Wi-Fi, near field communication and proprietary RF protocols thatcomply with accepted frequencies for medical applications and standards of wireless body area networks communication. c. The central node collects the biomechanical data acquired by the sensors embedded in the garment. It then synchronously transmits this data to the receiver, which is connected to the individual wireless units distributed across the body.

24. A method for providing tissue-level biomechanics analysis with the use of a network of wearable sensors embedded in smart garment and system in claim 19: a. Use of a database of laboratory-based data (marker-based optical motion capture, ground reaction forces, IMU, EMG) to develop neuromusculoskeletal models for calculation of digital biomarkers that is used to develop Al models that ca be deployed with IMU / EMG embedded smart garments. b. Use of muscle-tendon unit properties derived from individual characteristics to inform Al model. c. Inclusion of pre-developed muscle-tendon unit kinematics solver in Al framework. d. Combination of IMU, EMG, and Al / neural networks for real-time estimation of tendon and ligament forces / stress / strains, muscle forces / lengths / velocity / stiffness / power, cartilage forces / stress / strains, bone forces / stress / strains. e. Use of physics / neuromuscular dynamics-based loss functions for training Al / neural networks to predict digital musculoskeletal biomarkers. f. Combination of IMU and EMG embedded smart garments with Al / neural networks models for real-time estimation of tendon and ligament forces / stress / strains, muscle forces / lengths / velocity / stiffness / power, cartilage forces / stress / strains, bone forces / stress / strains. g. Options to include subject-specific measurements (height / weight / anthropometrics) and scans (MRI / CT / ultrasound) to improve model precision.

Images