GB2605351A

GB2605351A - System and method for monitoring muscle performance and providing real-time dynamic advice

Info

Publication number: GB2605351A
Application number: GB2100681.2A
Authority: GB
Inventors: Ertan Erhan
Original assignee: Neurocess Ltd
Current assignee: Neurocess Ltd
Priority date: 2021-01-19
Filing date: 2021-01-19
Publication date: 2022-10-05
Also published as: GB202100681D0

Abstract

A system and method for monitoring muscle performance of athletes and providing real-time dynamic advice therefrom in a wireless network, comprises a WiFi- based sensor device for receiving and transmitting muscle data. The sensor device has one or more surface electromyography sensors (sMEG) 100 to be positioned on a target muscle or muscle groups, each having a pair of active dry electrodes. At least one amplifier, direct current offset adder, analogue band pass filter, analogue to digital converter, Wi-Fi communication protocol 200 and a battery are also provided. The positioning of the sMEG sensors is recorded via a monitoring unit 400 having a user interface and is transmitted to a data processing unit 300. Data is received from activities that occur during resting and contraction of the muscle via the sensor device and is transmitted wirelessly. The data is processed in terms of mathematical values and computing algorithms for muscle contraction, balance between local muscle groups, muscle fatigue and/or muscle growth tracking by the data processing unit. The resulting processed data is compared with previous data of an athlete, current data among the muscle groups of the athlete, current data of other athletes and/or predetermined entries for target values or reference values. Feedback is generated based on the projected data to the users.

Description

SYSTEM AND METHOD FOR MONITORING MUSCLE PERFORMANCE AND

PROVIDING REAL-TIME DYNAMIC ADVICE

Technical Field

The invention relates to systems and methods for monitoring effectively muscle performance of athletes and providing real-time dynamic advice therefrom in order to maximize their efficiency and to minimize injury risks in sports and training activities.

Background of the Invention

Improving and reaching optimal and even peak performance during training and competition is one of the main goal for coaches and athletes. Today, psychophysiological approaches are mostly used in sport activities in order to monitor and attain a better understanding of the processes underlying athletic performance and thus to apply interventions in sport and physical exercise by athletes, coaches and sport medical experts. The most frequently used biometric techniques include electromyography (EMG), electrocardiography (ECG), electroencephalography (EEG), and the assessment of electrodermal activity and breathing rhythm.

Among others, EMG is a technique to analyse muscular signals generated by physiological changes in the state of the muscle membranes. Nevertheless, high-quality EMG, inertial measurement unit (IMU) sensors along with isokinetic dynamometers, force decks and lactate test strips are constrained in terms of mobility. Therefore, surface electrodes are mainly used for the study of neuromuscular activation in postural tasks, functional movements, working conditions or workout.

Despite EMG-based sports garments, which typically use conductive thread for EMG sensing and Bluetooth-Low-Energy (BLE) for data transmission are preferred for lifestyle sports activities, they cannot be safely used for clinically accurate physiological analysis due to noisy and low-quality data measurements, and restrictions of BLE technology to support on-field applications requiring secure, high speed and long-distance multichannel data transmission.

Most instruments that measure EMG send the signals to a computer or other hardware tool to filter the data, so it can be displayed and analysed later by a trained professional. The analysis of these types of data provides useful information to improve performance in motor and sport activities. A valid interpretation depends upon a strong knowledge of both movement science and muscle physiology, and other simultaneous measurements, and even comparative analysis among muscle groups of an athlete and those among other athletes of the same team or of a same or similar profession, should be taken to cross-validate and ensure confidence in the findings.

Musculoskeletal pain and injuries are prevailing in sports where athletes are required to perform at high-intensity activities for more extended periods. In particular, overuse injuries are common and affect badly one third of the adults. A study for the American National Football League, published in FORBES magazine on February 5, 2020, reported that a loss of $ 500 million in just one year was due only to injuries. One of the major factor responsible for this loss is the selection of inadequate or wrong exercise methods for muscle development, i.e. traditional time based or on-site (feeling) approaches by the sports people.

The document US20150182841A1 discloses a communication module for personal physical performance monitoring and/or facilitation for example during sports acts, comprising means for mounting the communication module suitable for communicating sensor signals, such as [MG signals, to a mounting zone on a sports item, the means for mounting comprising two or more electronic contact terminals for making an electronic contact with the sports item while being mounted thereon.

From EP1531726B1 there is known a method and an outfit for measuring of action of muscles of body, in which method electrical signals created by active muscles are measured with a measuring device and a response from action of muscles is given with a response device with a signal perceptible to a user.

Reliable, easy to use, and even smaller compact sensor instruments without requiring a wearable garment or outfit are needed in the art. Further, there is a need in the art for systems and methods for monitoring effectively muscle performance with improved comparative analysis among different muscle groups of an athlete/athletes and providing real-time dynamic accurate advice therefrom in order to maximize their efficiency in sports activities, to minimize injury risks and to pave the way for efficient muscle development.

Summary of the Invention

The invention is broadly defined by the features of the independent claim. Some specific and preferred embodiments are further defined in the dependent claims.

The invention is in general directed to a compact athlete tracking sensor system and method of operation thereof in elite sports advantageously offering clinical level [MG derived muscle data analysis with valuable insights and advice related to the athletic performance, injury risks and training efficiency.

According to one aspect of the present invention, there is provided a system for monitoring effectively muscle performance of athletes and providing real-time dynamic advice therefrom in a wireless network, comprising: - a Wi-Fi-based wireless sensor device for receiving and transmitting muscle electrical signals (i.e. input data); - a central processing unit for keeping and processing the muscle data (i.e. contractions, muscle fatigue and muscle development data) obtained from the sensor device; -a communication unit having a wireless protocol that enables data obtained from the sensor device to be transferred to the central processing unit; and - a monitoring unit having an user interface; wherein the said sensor device is ready and easy to use for reliable and quality muscle performance monitoring, without requiring a wearable garment or outfit thereon and comprises -one or more surface electromyography (sEMG) sensors to be positioned on the target muscle or muscle groups, each having a pair of active dry branched (bipolar) electrodes; - at least one amplifier for amplification of the input signals from the electrodes; - a direct current (DC) offset adder for eliminating negative voltage value of electrical signals due to the said amplification; -an analogue band pass filter for filtering noise and interference; - an analogue to digital converter; - a Wi-Fi communication protocol; and - a battery.

Unlike conventional applications, the sensor device according to the invention is in compact form and does include neither an external microprocessor for monitoring and processing muscle behaviours, which may increase the energy consumption as the core component in an embedded system, nor means for Bluetooth connections, which may limit the EMG data quality or high-range transmission. Instead, an efficient Wi-Fi wireless network communication is achieved and the microcontroller unit (MCU) of Wi-Fi module on the sensors is used to embed a code sampling only the data and transmitting it to the specified address but not processing the raw signal for power efficiency.

Likewise, the sensor device is provided with a pair of electrodes which are connected to the reference of the system with a high value of the resistor (i.e. greater than Mega Ohm value) by achieving to eliminate the need for a reference electrode which is significantly common in the prior art in order to cancel out the power line noise on the human body.

Gold plated copper electrodes are employed in the sensor device, the form of which is preferably spherical and base of which in contact with a muscle may be flat or slightly convex, in order to provide a better positioning and adhesion with the skin even if the shape of the muscle changes.

In a preferred use for an athlete, there is provided eight dry WiFi-based sensor devices in a charging case -including active sEMG sensors-each having for instance a dimension of 31mm x 31mm x 13mm, which may advantageously and artily be accepted as a mini compact configuration with wireless transmission (Wi-Fi) capability. In another preferred use, a flexible compact WiFi-based sensor device is provided with higher surface area but lower thickness, i.e. having ultra-thin sensors of maximum 6mm with a firm grip, to include high-level motion activities.

The system enables multichannel and long-distance data transfer and web-based user interface for real-time plots. No button is provided on the said sensor device to turn it on and off, so that it is user friendly and has much smaller in volume. When the sensors are taken out of the charging case, they are automatically connected to the Wi-Fi wireless network to transmit data. When they are put back in the charging cases, the charging begins and sensors stop data transmission.

In a preferred system according to the invention, the said sensor device further comprises real-time warning means for athletes of undesired physiological level, i.e. via a vibration alarm unit or a visual fatigue indicator upon reaching a predetermined level of high risk of injury, activated by an authorized user manually (i.e. manual entry via the user interface) or by the computing algorithm automatically.

The system may further comprises additional sensing tools such as a thermocouple sensor, an IMU, and pressure sensors to be built into the sensor circuitry to promote onboard EMG data analytics.

According to another aspect of the present invention, there is provided a method for monitoring effectively muscle performance of athletes and providing real-time dynamic advice therefrom, comprising the steps of: - positioning one or more EMG based compact sensor device on the target muscle or muscle groups of an athlete or athletes; - recording the said positioning via a monitoring unit having an user interface and transmitting the same to a data processing unit; - receiving data as electrical signals from activities that occur during the resting and contraction of the muscles via the sensor device and transmitting wirelessly the same to the data processing unit; - processing data in terms of mathematical values and computing algorithms for muscle contraction, balance between local muscle groups, muscle fatigue and/or muscle growth tracking instantly by the data processing unit; - comparing instantly the resulting processed data with - previous data of an athlete, if any, - current data among the muscle groups of the athlete, -current data of other athlete/s; and, - predetermined entries for target values or reference values; - calculating future projections of the collected and compared data as such i.e. via deep learning technology and data mining; - transmitting both the raw data obtained from the sensor module and the processed data with projections wirelessly to the monitoring unit and/or to the cloud; and - generating feedback based on projected data to the users.

The method further comprises the step of warning users upon an undesired physiological level or threat, i.e. predetermined level of high risk of injury, via generating feedback on the monitoring unit for the trainer or other users, and/or on a vibration alarm unit or on a visual fatigue indicator for the athlete.

The method still further comprises entering data for goals, warnings or training plans in advance by athletes, coaches and sport medical experts and taking advantage of cutting-edge data-science analytics in use according to the invention.

Thus, the invention advantageously provides a system and method for receiving, collecting, transmitting data through improved Wi-Fi based compact EMG sensor device easily positioned on the body of an athlete, and then analysing them via computing algorithms and making recommendations for the relevant people, providing scientifically accurate analysis and information to coaches and sport medical experts and developing training plans and discipline according to the predetermined goals.

The following numbered clauses show further illustrative examples only.

1. A system for monitoring effectively muscle performance of athletes and providing real-time dynamic advice therefrom in a wireless network, comprising: - a WiFi-based sensor device (100) for receiving and transmitting muscle electrical signals/data; - a central processing unit (300) for keeping and processing the muscle data obtained and transmitted from the sensor device (100); -a communication unit (200) with a Wi-H communication protocol that enables data obtained from the sensor device (100, 108) to be transferred to the central processing unit; and - a monitoring unit (400) having an user interface; wherein the said sensor device is ready and easy to use for reliable and quality muscle performance monitoring, without requiring a wearable garment or outfit thereon and comprises - one or more surface electromyography (sEMG) sensors (100) to be positioned on the target muscle or muscle groups, each having a pair of active dry electrodes (101); - at least one amplifier (102, 104) for amplification of the input signals from the electrodes (101); - a direct current offset adder (103) for eliminating negative voltage value of electrical signals due to the said amplification; - an analogue band pass filter (105) for filtering noise and interference; - an analogue to digital converter (107); -a Wi-Fi communication protocol (108); and - a battery (110).

2. A system according to clause 1, characterised in that the sensor device further comprises an integrated sensor selected from the group consisting of accelerometer sensor (106), pulse sensor or ECG sensor.

3. A system according to clause 1 or 2, characterised in that the sensor device further comprises real-time warning means for athletes of undesired physiological level via a vibration alarm unit or a visual fatigue indicator upon reaching a predetermined level of high risk of injury, to be activated by the computing algorithm at the central processing unit (300) automatically or by an authorized user manually thereto through the user interface (400).

4. A system according to any one of the preceding clauses, characterised in that the electrodes (101) of the EMG sensor device (100) are spherical gold plated copper electrodes.

5. A system according to any one of the preceding clauses, characterised in that the sensor device (100) has a printed circuit board with a compact mechanical casing.

6. A system according to any one of the preceding clauses, characterised in that the electrodes (101) of the sensor device (100) are connected to the reference of the system in the sensor circuitry with a high value of the resistor that is greater than Mega Ohm value.

7. A system according to any one of the preceding clauses, characterised in that the sensor device (100) further comprises one or more sensing tools in the sensor circuitry to be selected from group consisting of a thermocouple sensor, an IMU and a pressure sensor.

8. A method for monitoring effectively muscle performance of athletes and providing real-time dynamic advice therefrom in a Wi-Fi based wireless network, preferably connected with cloud services (500), comprising the steps of: - positioning one or more EMG based compact sensor device (100) on the target muscle or muscle groups of an athlete or athletes; - recording the said positioning via a monitoring unit (400) having an user interface and transmitting the same to a data processing unit (300); receiving data as electrical signals (108) from activities that occur during the resting and contraction of the muscles via the sensor device (100) and transmitting wirelessly the same to the data processing unit (300); - processing data in terms of mathematical values and computing algorithms for muscle contraction, balance between local muscle groups, muscle fatigue and/or muscle growth tracking instantly by the data processing unit; - comparing instantly the resulting processed data with - previous data of an athlete, if any, and/or; - current data among the muscle groups of the athlete, and/or; -current data of other athlete/s and/or; - predetermined entries for target values or reference values - transmitting both the raw data obtained from the sensor module (100) and the processed data with projections wirelessly to the monitoring unit (400) and/or to the cloud (500); and -generating feedback based on projected data to the users.

9. A method according to clause 8, further comprising the step of warning users upon an undesired physiological level or threat of high risk of injury, via generating feedback on the monitoring unit for the trainer or other users, and on a vibration alarm unit or on a visual fatigue indicator for the athlete.

10. A method according to clauses 8 or 9, further comprising the step of entering data for goals, warnings or training plans in advance by athletes, coaches and sport medical experts.

11. A method according to any one of the preceding clauses 8 to 10, further comprising the step of calculating future projections of the collected and compared data as such i.e. via deep learning technology and data mining.

12. A method according to any one of the preceding clauses 8 to 11, comprising monitoring effectively muscle performance of athletes and providing real-time dynamic advice therefrom in a wireless network, wherein the physiological and biomechanical estimation, analysis and comparison are directed to lactate level, peak-torque, muscular strength, coordination, dehydration, joint angular velocity, orientation, motion classification, training efficiency, injury risk, Quadriceps/Hamstring ratio, bilateral asymmetry and athletic form.

13. A method according to any one of the preceding clauses 8 to 12, characterised in that the operation of the compact sensor device (100) comprises the following steps: placing conductive electrodes (101) of an active EMG sensor (100) onto target muscle or muscle groups to receive muscle signals therefrom; - amplifying the electrical signals received by the active EMG sensor (100) through the electrodes (101) at an instrumentation amplifier (102) and preferably at a second stage amplifier (104); eliminating negative voltage value of the electrical signals amplified by the instrumentation amplifier (102) at a DC offset adder (103); -filtering the resulting amplified electrical signals at a band-pass filter (105); converting samples of an analogue signal into digital values of the amplified and filtered electrical signals at a digitizer (107); - transmitting the signals received from the digitizer (107) to the central processing unit (300) via Wi-H communication protocol (108) through a router (200) located in the sports facility, - processing the output signals transmitted from the sensor device (100) at a central processing unit (300), -transmitting the processed data from the central processing unit (300) to a user interface of a monitoring unit (400) (i.e. an external tablet computer or smart phone) via the wireless communication (200) and cloud services (500), and vice versa.

14. A method according to clause 13, characterised in that the operation of the computing algorithm for processing data further comprises the step of receiving independently the signals related to acceleration of muscles or muscle groups upon movement from an integrated accelerometer sensor (106) of the sensor device (100) and transmitting them simultaneously to the digitizer (107), 15. A method according to any one of the preceding clauses 8 to 14, characterised in that the operation of the computing algorithm for processing data comprises a deterministic temporal classification approach using many-to-many bidirectional long-short term memory recurrent neural network, which learns the contextual information of a given sequence in real-time and provides onset-detection accordingly in order to distinguish EMG-onset in a motion-sourced noise environment, and which automatically classifies the onset and offset parts of the signal and analyse only the signals' active regions successfully.

16. A method according to clause 15, characterised in that the operation of the computing algorithm for processing data follows a sample-by-sample classification, which allows the model to capture fast, dynamic motion-sourced noise contamination, which frame-by-frame metrics fail to detect.

Brief Description of Drawings

The invention will now be explained in more detail with reference to the accompanying drawings, in which, FIG. 1 shows a schematic block diagram of a sensor device according to one embodiment of the invention.

FIG. 2A-B shows a schematic rear and front perspective views of a sensor device with surface electrodes according to one embodiment of the invention, respectively.

FIG. 3 illustrates a system extended from the sensor device to external devices or cloud services.

Detailed Description of the Invention

According to the invention, there is provided a system and method for monitoring effectively muscle performance with improved sensing capability and comparative analysis among different muscle groups of an athlete/athletes and providing real-time dynamic accurate advice therefrom in order to maximize their efficiency in sports activities and to minimize injury risks, having a main focus on maximum mobility, endurance, and signal quality that can achieve clinical accuracy for a better experience and use for sports people including athletes, coaches and sport medical experts.

FIG. 1 shows the main internal components of an improved sensor device (100) according to one embodiment of the invention, which is ready and easy to use for reliable and quality muscle performance monitoring, without requiring a wearable garment or outfit thereon. The sensor device (100) having a printed circuit board (PCB) with a compact mechanical casing (as best seen in FIG. 2) comprises one or more active EMG sensors, each having a pair of highly conductive electrodes (101), necessary circuits for amplification of the input signals from the electrodes (101) and independently transmitting the collected data from the muscles wirelessly to a central processing unit (i.e. to a mini PC) and then to a user interface via Wi-Fi communication protocol/module (108) through a router.

Gold-plated active copper electrodes (101) are preferred for the EMG sensors for more precise sensing capabilities. The sensor device (100) with dry feature, has a filtering capability to increase the voltage range and minimize the interferences that may occur in operation. For this purpose, electrical signals (input data) received from the muscles (having i.e. high input impedance) through the electrodes (101) are preferably amplified twice, first at instrumentation amplifier (102) (i.e. 30-100 times) and then at second stage amplifier (104) (i.e. 5-20 times) in order to eliminate various undesired interference due to movement, impacts and vibrations and to be processed later accurately. Then, a direct current (DC) offset adder (103) is employed to eliminate negative voltage value of electrical signals (undesired direct current) due to the said amplification.

Possible remaining noise and interference are cleaned by filtering at an analogue band-pass filter (105) and then a digitizer (107) is employed to convert samples of an analogue signal into digital values using analogue to digital converter (ADC) with a resolution of 16 byte. The EMG output signals obtained in this way is transmitted via Wi-Fi communication protocol/module (108) through a router to the central processing unit (i.e. mini PC).

The sensor device (100) further comprises a battery (110) and preferably an accelerometer sensor (106) which measures the acceleration of muscles or muscle groups upon movement.

The sensor devices (100) are placed on the specific muscle locations or the target muscle group with a double sided elastic adhesive tapes or kinesio tapes. Each sensor has a Wi-Fi module with internal microcontroller unit (MCU) in which a code is embedded only for sampling the data and transmitting it to the specified address. No external microprocessor is used in the sensor device (100), as there will be no data processing on the sensors.

Frequency spectrum is significant for signal processing algorithms to calculate the injury risks. Unlike the common use of a rectifier to eliminate the negative voltages of the similar embodiments in the prior art, which leads a change in the frequency spectrum of the raw signal, the said direct current offset adder circuit (103) acts as a DC adder of 1.5V to the reference of the instrumentation amplifier (102) to shift the raw muscle signals on electrodes up, and thus to eliminate negative signals without disturbing the frequency of the signal.

FIG. 3 shows one example of a system that can take advantage of the invention in general. The system comprises a sensor device (100) which collects and transmits the required data to a central processing unit (300) through the router (200) installed in the sports facility by using Wi-Fi communication protocol. The processing unit (300) controls and processes the input and output of data between the sensor device (100) and a monitoring unit (400), i.e. an external tablet computer or smart phone, which can be reachable by any authorized user by means of cloud services (500) over the internet and/or connected wirelessly in the facility. Data is processed by the processing unit (300) comprising typically a microcontroller operated by firmware and an amount of memory. A mini PC can be used for this purpose.

The system allows a user to examine and analyse the physiological signals of an athlete in detail thanks to the data processing method, and allows the amount of muscle contraction, balance values between local muscle groups, muscle fatigue and muscle development to be monitored instantly.

The method for the operation of the compact sensor device (100) according to the invention comprises the following steps: placing conductive electrodes (101) of an active EMG sensor (100) onto target muscle or muscle groups to receive muscle signals therefrom; - amplifying the electrical signals received by the active EMG sensor (100) through the electrodes (101) at an instrumentation amplifier (102) and preferably at a second stage amplifier (104); eliminating negative voltage value of the electrical signals amplified by the instrumentation amplifier (102) at a DC offset adder (103); - filtering the resulting amplified electrical signals at a band-pass filter (105); - converting samples of an analogue signal into digital values of the amplified and filtered electrical signals at a digitizer (107); -receiving independently the signals related to acceleration of muscles or muscle groups upon movement from an integrated accelerometer sensor (106) of the sensor device (100) and transmitting them simultaneously to the digitizer (107), - transmitting the signals received from the digitizer (107) to the central processing unit (300) via Wi-H communication protocol/module (108) through a router (200) located in the sports facility, - processing the output signals transmitted from the sensor device (100) at a central processing unit (300), -transmitting the processed data from the central processing unit (300) to a user interface of a monitoring unit (400) (i.e. an external tablet computer or smart phone) via the wireless communication (200) and cloud services (500), and vice versa.

The processed data at the central processing unit provides information and analysis related to local muscle contraction, the balance values between local muscle groups, muscle dehydration level, lactic acid accumulation in the muscle and muscle coordination together with possible injury risks, all of which can be calculated instantly and/or within a desired time interval. Finally, the results obtained from such calculations are available for authorized users through Wi-Fi wireless communication (108, 200) and user interface (400, 500).

Although the voltage emitted by the muscles in the human body varies from person to person, the automatic calibration feature and the data processing method according to the invention provide instant identification and analysis from the moment a muscle or muscle group of an athlete or athletes start operating. Accordingly, contraction percentages are calculated and analysed and corresponding advice is produced at the user interface on whether the athlete performs the movement correctly or not and how much the injured muscle is activated, if any, together with alarming signals in high risk incidence to the users, if required. Therefore, the system also allows for instructions and information of the authorized users to be transmitted and mutually exchanged wirelessly to and from the sensor device (100) via the user interface and/or the central processing unit (300), manually or automatically.

Each electrode (101) of an EMG sensor according to one embodiment is preferred to be a 10 mm diameter copper core, 100 nm gold plated, and a design in accordance with ISEK and SENIAM standards. The distance between the centres of the electrodes is preferred as 20 mm. Low noise biomedical amp with 130 dB CM RR may be used in amplifications. The signal amplification is divided into two parts to keep noise and interference as low as possible. Both amplifier parts have a self-adjustable rheostat system. The total amplifier gain may be changed between 100-1000. The direct current value that may occur when the electrodes slides in different directions can be eliminated by using a high-pass filter. Besides, the high-pass filter sharply filters undesired occurrence that can cause motion, impact and similar interference (i.e. with 20 Hz cut-off frequency). By this way, any noise and interference that can disrupt the electric signal of active EMG sensor due to muscle group movement, electrode (101) motion and impacts are successfully minimized.

Active EMG sensor (100) signals also inform the user about the fatigue level of the muscles. Thanks to signal processing algorithms, it can be calculated locally how fatigue the muscles are, whether they are fatigued in a balanced way, and the change in the starting time of fatigue compared to previous training. At the same time, the active EMG sensor (100) data of the same muscle movements for a certain period of time are recorded and analysed so that the an athlete's local and general muscle development can be calculated during this process. In addition, the agility of the athlete can be monitored instantly with the accelerometer sensor (106). On the other hand, the recorded accelerometer data are analysed simultaneously and the progress in the agility of individuals throughout the process can be observed. Optionally integrated sensors, as well as accelerometer sensor, can also be included in wearable or similar products. Integrated sensors can be used as pulse or ECG sensors.

sEMG measurements are sensitive to several artefacts caused by motion and impact, which are frequently encountered in competitive sports. The contemporary methods used for onset detection are EMG Envelope, Sample Entropy (Sampen), Modified Adaptive Linear Energy Detector (M-ALED) and Adaptive Contraction Detection (ACD). Each method has a distinct advantage; however, they fail to thoroughly differentiate the EMG-onset and motion artefacts. To this end, the invention proposes a deterministic temporal classification approach using many-to-many Bidirectional Long-Short Term Memory Recurrent Neural Network (BLSTM-RNN), which learns the contextual information of a given sequence in real-time and provides onset-detection accordingly in order to distinguish EMG-onset in a motion-sourced noise environment. Accordingly, the algorithm of the approach automatically classifies the onset and offset parts of the signal and analyse only the signals' active (i.e., the part that contains muscular contraction) regions successfully.

The results on the dataset collected from the football players are shown in the Table 1 below. The invention's proposed method reaches a very distinguished accuracy of 96.73% and an El-score of 95.51%, which is the highest among the other state of the art metrics.

Table 1: Comparison of results obtained from various methods Metric (%) EMG-envelope Sampen M-ALED ACD Invention Accuracy 61.41 84.95 58.86 65.54 96.73 Fl-Score 58.10 78.06 41.24 64.53 95.51 The invention can be recognised as clearing up the onset and offset detection problem for the real-life framework. The approach of invention, i.e., BLSTM-RNN, resulted in the highest accuracy of 96.73 %, that is 11.78% more than the second-highest, i.e. sample entropy. The said approach follows a sampleby-sample classification, which allows the model to capture fast, dynamic motion-sourced noise contamination, which frame-by-frame metrics of the conventional approach fail to detect. Moreover, the bidirectionality and memory advantages of the BLTSM-RNNs enable the model to use and include the contextual information of the raw sEMG sequence.

Physiological estimations carried out according to the invention by only using sEMG data include but not limited to lactate level (g/mmol), muscular conductivity, hamstring to quadriceps ratios, bilateral eccentric/concentric strength ratios and biomechanical estimations to joint angular velocity, peak-torque, and extracted force. These estimations of muscular states are used to infer and predict critical information, including motion class (i.e., prediction of motion type realised by the subject), training efficiency, injury risk, and athletic form. Estimations of muscular states and motion types are then used to extract injury risk probabilities such as anterior cruciate ligament (ACL) and hamstring strain injury (HSI). As sEMG data depends on sensor placement, subject type (i.e., variations of body type and age), and noise contamination, ML models' estimation accuracy drops correspondingly. To this end, we proposed a model-agnostic meta-learning based weight optimisation to our models along with a one-shot learning deep neural networks for overcoming data dependency, i.e., estimation over tasks with new sensor placements, athletes and noise sequence. Instead of generalising over tasks using the provided data, the ML models proposed in the invention uses a meta-learner scheme to "learn-to-learn". All of the mentioned information is displayed in a web-based user interface.

The device's software interface offers flexibility for remote training which is particularly important in the context of the current Covid-19 outbreak, in which prior art athlete monitoring devices are ineffective due to their incapability of transmitting live data between multiple users in various locations, lack of a cloud-based platform to deliver data of numerous athletes from separate areas in real-time to provide a complete experience of distant exercising which can be monitored by training staff.

Likewise, athlete muscle activities can be securely and adequately followed in real-time online during the sports competition or training process by utilizing improved sensor technology, deep learning technology, cloud computing technology, data mining technology and wireless communication technology according to the present invention.

Claims

CLAIMS1. A system for monitoring muscle performance of athletes and providing real-time dynamic advice therefrom in a wireless network, comprising: - a WiFi-based sensor device (100) for receiving and transmitting muscle data; -a central processing unit (300) for storing and processing the muscle data obtained and transmitted from the sensor device (100); - a communication unit (200) with a Wi-H communication protocol that enables the data obtained from the sensor device (100) to be transferred to the central processing unit; and a monitoring unit (400) having a user interface; wherein the sensor device comprises: - one or more surface electromyography, sEMG, sensors (100) to be positioned on a target muscle or muscle groups, each having a pair of active dry electrodes (101); - at least one amplifier (102, 104) for amplification of input signals from the electrodes (101); - a direct current offset adder (103) for eliminating negative voltage value of electrical signals due to the amplification; - an analogue band pass filter (105) for filtering noise and interference; - an analogue to digital converter (107); -a Wi-Fi communication protocol (108); and - a battery (110).
2. The system according to claim 1, characterised in that the sensor device further comprises an integrated sensor selected from a group consisting of accelerometer sensor (106), pulse sensor or ECG sensor.
3. The system according to claim 1 or 2, characterised in that the sensor device further comprises a real-time warning means for athletes of undesired physiological level via a vibration alarm unit or a visual fatigue indicator upon reaching a predetermined level of high risk of injury, to be activated by a computing algorithm at the central processing unit (300) automatically or by an authorized user manually through the user interface (400).
4. The system according to any one of the preceding claims, characterised in that the electrodes (101) of the EMG sensor device (100) are spherical gold-plated copper electrodes.
5. The system according to any one of the preceding claims, characterised in that the sensor device (100) has a printed circuit board with a compact mechanical casing.
6. The system according to any one of the preceding claims, characterised in that the electrodes (101) of the sensor device (100) are connected to a reference of the system in the sensor circuitry with a high value resistor.
7. The system according to claim 6, wherein the high value resistor is greater than or equal to 1 MO.
8. The system according to any one of the preceding claims, characterised in that the sensor device (100) further comprises one or more sensing tools in the sensor circuitry to be selected from group consisting of a thermocouple sensor, an IMU and a pressure sensor.
9. A method for monitoring muscle performance of athletes and providing real-time dynamic advice therefrom in a Wi-Fi based wireless network, comprising the steps of: - positioning one or more electromyography, EMG, based compact sensor device (100) on a target muscle or muscle groups of an athlete or athletes; -recording the positioning via a monitoring unit (400) having an user interface and transmitting the positioning to a data processing unit (300); - receiving data as electrical signals (108) from activities that occur during resting and contraction of the target muscle or muscle groups via the sensor device (100) and transmitting wirelessly the data to the data processing unit (300); -processing the data in terms of mathematical values and computing algorithms for muscle contraction, balance between local muscle groups, muscle fatigue and/or muscle growth tracking by the data processing unit; - comparing the resulting processed data with - previous data of an athlete, if any, and/or; -current data among the muscle groups of the athlete, and/or; - current data of other athlete's and/or; - predetermined entries for target values or reference values; - wirelessly transmitting both the raw data obtained from the sensor module (100) and the processed data with projections to the monitoring unit (400); and -generating feedback based on projected data to users.
10. The method according to claim 9, wherein the WiFi based wireless network is connected with cloud services (500).
11. The method according to claim 10, wherein the raw data and processed data with projections are wirelessly transmitted to the cloud (500).
12. A method for monitoring muscle performance of athletes and providing real-time dynamic advice therefrom in a Wi-Fi based wireless network connected with cloud services (500), comprising the steps of: - positioning one or more electromyography, EMG, based compact sensor device (100) on a target muscle or muscle groups of an athlete or athletes; - recording the positioning via a monitoring unit (400) having an user interface and transmitting the positioning to a data processing unit (300); -receiving data as electrical signals (108) from activities that occur during resting and contraction of the target muscle or muscle groups via the sensor device (100) and transmitting wirelessly the data to the data processing unit (300); - processing the data in terms of mathematical values and computing algorithms for muscle contraction, balance between local muscle groups, muscle fatigue and/or muscle growth tracking by the data processing unit; comparing the resulting processed data with - previous data of an athlete, if any, and/or; -current data among the muscle groups of the athlete, and/or; -current data of other athlete/s and/or; predetermined entries for target values or reference values; - wirelessly transmitting both the raw data obtained from the sensor module (100) and the processed data with projections to the cloud services (500); and - generating feedback based on projected data to users.
13. The method of claims 9 to 12, wherein the data is processed instantly; and/or wherein the resulting processed data is compared instantly.
14.The method according to claim 9 to 13, further comprising a step of warning users upon an undesired physiological level or threat of high risk of injury, via generating feedback on the monitoring unit for a trainer or other users, and on a vibration alarm unit or on a visual fatigue indicator for the athlete.
15. The method according to any one of claims 9 to 14, further comprising a step of entering data for goals, warnings or training plans in advance by athletes, coaches and sport medical experts.
16. The method according to any one of the preceding claims 9 to 15, further comprising a step of calculating future projections of the collected and compared data as such via deep learning technology and data mining.
17. The method according to any one of the preceding claims 9 to 16, comprising monitoring muscle performance of athletes and providing real-time dynamic advice therefrom in a wireless network, wherein a physiological and biomechanical estimation, analysis and comparison are directed to lactate level, peak-torque, muscular strength, coordination, dehydration, joint angular velocity, orientation, motion classification, training efficiency, injury risk, Quadriceps/Hamstring ratio, bilateral asymmetry and/or athletic form.
18. The method according to any one of the preceding claims 9 to 17, characterised in that the operation of the compact sensor device (100) comprises the following steps: placing conductive electrodes (101) of an active EMG sensor (100) onto a target muscle or muscle groups to receive muscle signals therefrom; - amplifying the electrical signals received by the active EMG sensor (100) through the electrodes (101) at an instrumentation amplifier (102); eliminating negative voltage value of the electrical signals amplified by the instrumentation amplifier (102) at a DC offset adder (103); filtering the resulting amplified electrical signals at a band-pass filter (105); - converting samples of an analogue signal into digital values of the amplified and filtered electrical signals at a digitizer (107); - transmitting the signals received from the digitizer (107) to the central processing unit (300) via Wi-Fi communication protocol (108) through a router (200), - processing the output signals transmitted from the sensor device (100) at a central processing unit (300), -transmitting the processed data from the central processing unit (300) to a user interface of a monitoring unit (400) via the wireless communication (200) and cloud services (500), and vice versa.
19. The method of claim 18, wherein the router (200) is located in a sports facility.
20. The method according to claims 18 or 19, wherein the electrical signals are amplified at a second stage amplifier (104)
21. The method according to claims 18 to 20, wherein the monitoring unit (400) is an external tablet computer or smart phone.
22. The method according to claims 18 to 21, characterised in that the operation of the computing algorithm for processing data further comprises a step of independently receiving signals related to acceleration of muscles or muscle groups upon movement from an integrated accelerometer sensor (106) of the sensor device (100) and transmitting them simultaneously to the digitizer (107).
23. The method according to any one of the preceding claims 8 to 22, characterised in that the operation of the computing algorithm for processing data comprises a deterministic temporal classification approach using a many-to-many bidirectional long-short term memory recurrent neural network, which learns contextual information of a given sequence in real-time and provides onset-detection accordingly in order to distinguish EMG-onset in a motion-sourced noise environment, and which automatically classifies the onset and offset parts of the signal and analyse only the signals' active regions successfully.
24. The method according to claim 23, characterised in that the operation of the computing algorithm for processing data follows a sample-by-sample classification.