Method for determining internal muscle load, and corresponding device and program

Abstract

The invention relates to a method for determining items of data representative of an internal muscle load (CM) applied by at least one muscle belonging to a muscle group. The method comprises the steps of:-obtaining (A10) items of kinematic data (i (t)) and electromyogram data (emg (t)) collected during the contraction of the at least one muscle; -determining (A20) a muscle-tendon length (lMT (t)); determining (A30) a muscle activation level (a (t)); -determining (A40) the muscle length (lM (t)) and the contraction rate (vM (t)); -determining (A50) an estimate of the muscle deformation (dM (t)); calculating (A60) a muscle stress index (sM (t)); and-calculating (A70) the internal muscle load (CM) by multiplying the muscle stress index (sM (t)) and the duration of force (t).

Description

Technical Field

[0001] This invention relates to the field of physiological measurement, and more particularly to non-invasive physiological measurements of muscle exertion associated with performing exercise tasks. Physiological measurements include measuring the physical quantities resulting from the force exerted by a user's muscles, particularly the internal muscle load of one or more of the user's muscles.

[0002] This invention can be applied to various fields to evaluate the relationship between muscle exertion by an individual and the motor task performed. This invention can be used in clinical settings, ergonomics, and sports. Background Technology

[0003] The rapid proliferation of portable connectivity items (such as watches or straps that measure heart rate) is no longer a real point of contention. Most people use connected watches or equivalent devices in their daily lives and during their physical activities. Connected watches measure external training loads (such as walking speed and distance). Duration of exertion) and internal cardiovascular load (the exertion of training on the trainee; for example, by measuring heart rate). (Duration of exertion). However, it is well known that internal cardiovascular load estimated using heart rate is not correlated with internal muscle load. In other words, the intensity of muscle exertion, especially the intensity of muscle exertion in different muscles / muscle groups, cannot be estimated simply by measuring heart rate.

[0004] To date, internal muscle load has not been estimated, either in the laboratory or in relevant environments (sports, medical, workplace, especially industrial environments, daily activities). However, quantifying internal muscle load would be extremely useful, for example, in identifying acute and / or chronic changes in an individual's muscle capacity (e.g., aging, acute / chronic fatigue associated with repetitive movement tasks performed in occupational environments / sports), and in quantifying whether physical therapy treatments proposed by medical professionals (e.g., physical therapists, rehabilitation physicians, etc.) are effective against target muscles or muscle groups. These analyses are particularly relevant because a range of studies have shown that each individual uses their muscles in unique ways during various movement tasks (walking, cycling, strength training) (i.e., muscle coordination). Depending on the muscle coordination used by an individual, the distribution and extent of adaptations (e.g., hypertrophy) or alterations (e.g., MSD, sarcopenia) may therefore vary significantly from person to person.

[0005] Therefore, tools are needed to assess internal muscle load in order to better address the problems posed by existing technologies. Summary of the Invention

[0006] The described technology was conceived after considering these problems with the prior art. More specifically, the present invention relates to a method for determining data representing the internal muscle load generated by at least one muscle belonging to a muscle group, the method being implemented by means of an electronic device including a processor and a memory. Such a method includes the steps of: - Obtain kinematic and electromyographic data of the at least one muscle during contraction; - Based on the kinematic data, using a musculoskeletal model, determine the representative muscle-tendon length data for the at least one muscle; - Based on the electromyography data, determine the representative muscle activation level of the at least one muscle; - Based on the muscle-tendon length and muscle activation level, and two neuromuscular and muscle-tendon models, determine the muscle length and data representing the contraction velocity of the at least one muscle; - Based on the muscle length, determine an estimate of the muscle deformation of the at least one muscle; - Calculate the muscle stress index based on the muscle activation data, the muscle deformation, and muscle force indices dependent on the contraction speed; and - The internal muscle load is calculated by multiplying the muscle stress index by the duration of the force exerted.

[0007] This allows for the quantification of the internal muscle load on muscles or muscle groups relevant to a motor task. This internal muscle load characterizes the level of force exerted by the muscle for a given external load.

[0008] By using muscle activation data, muscle deformation, and muscle force indices dependent on contraction speed to calculate internal muscle load, the method according to the invention allows for more accurate prediction than existing techniques of muscle changes and adaptations that occur when muscles contract at lengths greater than the optimal length.

[0009] The step of determining the muscle-tendon length of the at least one muscle based on specific characteristics includes the following sub-steps: - Simulate joint angles during exertion based on kinematic data and musculoskeletal geometry models; and -Calculate the temporal evolution of the muscle-tendon length based on the joint angles and musculoskeletal geometry described by the musculoskeletal model.

[0010] The step of determining data representing the muscle activation level of the at least one muscle based on specific characteristics includes the following sub-steps: - The electromyography data is filtered in the frequency domain to provide a signal envelope; - Rectify the signal envelope; - Normalize the signal envelope to provide a normalized envelope describing the neural activation level of the muscle; and -Based on the neural activation level of the muscle, data representing muscle activation are calculated.

[0011] According to another aspect, the invention also relates to an electronic device comprising a processor and a memory, arranged to determine data representing internal muscle load generated by at least one muscle belonging to a muscle group. Such a device includes: - A device for acquiring kinematic and electromyographic data of the at least one muscle during static or dynamic exertion; - A determining device for determining, based on kinematic data and electromyographic data, the muscle-tendon length of the at least one muscle, data representing the level of muscle activation, data representing the muscle length, an estimate of muscle deformation, and data representing the contraction speed. - A computing device configured to calculate a muscle stress index based on the muscle activation data, muscle deformation, and muscle force indices depending on the contraction speed, and configured to calculate data representing the internal muscle load.

[0012] According to another aspect of the invention, the invention also relates to an electronic system comprising at least one processor and at least one memory, arranged to determine data representing internal muscle load generated by at least one muscle belonging to a muscle group, the system comprising smart clothing paired with a communication device and connected to a processing server via a communication network. Such a system further comprises: - A device for acquiring kinematic and electromyographic data of the at least one muscle in contact with the smart garment during static or dynamic exertion; - A determining device for determining, based on the kinematic data and electromyographic data, the muscle-tendon length of the at least one muscle, data representing the activation level, data representing the muscle length, an estimate of the muscle deformation, and data representing the contraction speed. - A computing device configured to base its calculations on the muscle activation data, the muscle deformation, and a force index f that depends on the contraction velocity. FV (t) Calculate the muscle stress index, and be configured to calculate data representing the internal muscle load.

[0013] The obtaining device, the determining device, and the computing device are distributed among the various components of the system.

[0014] According to a preferred implementation, different steps of the method according to this disclosure are implemented by one or more software programs or computer programs, which include software instructions intended to be executed by a data processor of an execution terminal according to the technology. The data processor is designed to control the execution of different steps of the method, which are implemented at the level of a communication terminal, remote server, and / or blockchain within a distributed framework of processing operations to be performed as determined by script source code or compiled code.

[0015] Therefore, this technology also relates to programs executable by a computer or a data processor, wherein such programs include instructions for controlling the steps of performing the methods described above.

[0016] The program can be in any programming language, and can be in the form of source code, object code, or intermediate code between source code and object code (such as partially compiled code), or any other desired form.

[0017] This technology also relates to an information medium that can be read by a data processor, the information medium containing instructions for the program described above.

[0018] Information media can be any entity or terminal capable of storing programs. For example, media can include storage media such as ROM (Read-Only Memory), such as CD-ROM or microelectronic circuit ROM, or magnetic recording media such as removable media (memory card) or hard disk or SSD.

[0019] The information medium can also be a transmissible medium, such as electrical or optical signals that can be transmitted via cables or optical fibers, radio, or other means. Programs according to this technology can be downloaded, in particular, from an internet-type network.

[0020] Alternatively, the information medium may be an integrated circuit in which a program is incorporated, wherein the circuit is adapted to execute or be used to execute the method in question.

[0021] According to one implementation, this technology is implemented by means of software and / or hardware components. In this regard, the term "module" in this document may refer to a software component, a hardware component, or a group of hardware and software components.

[0022] Based on the module-related descriptions discussed below, a software component corresponds to one or more computer programs, one or more subroutines of a computer program, or more generally, any element of a program or software capable of implementing a function or group of functions. Such software components are executed by the data processor of a physical entity (terminal, server, gateway, set-top box, router, etc.) and have access to the hardware resources of that physical entity (memory, storage media, communication bus, electronic input / output cards, user interface, etc.).

[0023] Similarly, according to the module-related descriptions discussed below, a hardware component corresponds to any element of a hardware assembly capable of implementing a function or group of functions. It can be a programmable hardware component or a component with an integrated processor for executing software, such as an integrated circuit, smart card, memory card, electronic card for executing firmware, etc.

[0024] Each component of the system described above naturally implements its own software module.

[0025] The different implementations described above can be combined with each other to implement this technology.

[0026] Other objects, features, and advantages of the invention will become clearer from the following simple and non-limiting description, which is provided in conjunction with the accompanying drawings, which illustrate: - Figure 1 A schematic flowchart illustrating the steps of the method according to the present invention is shown; - Figure 2 A second schematic flowchart showing the steps of the method according to the invention is shown; - Figure 3 The graph shows the muscle activity of the semimembranosus, semitendinosus, and biceps femoris during the running cycle. MT Examples of quantitative time evolution (unit: meter); - Figure 4a It shows an example of the evolution of the force-velocity relationship; - Figure 4b An example illustrating the evolution of the force-length relationship is shown; - Figure 5 A simplified physical architecture of the device used to determine internal muscle load is shown; - Figure 6 A simplified physical architecture of the system used to determine internal muscle load is shown; - Figure 7 The relationship between muscle damage distribution and (a) the internal muscle load distribution measured by the method according to the present invention and (b) the muscle activation distribution is explained. Detailed Implementation

[0027] Principles Review As previously stated, the object of this invention is to determine, for a given exercise, kinematics, or exercise group, data (indices) representing the internal muscle load generated during that exercise or exercise group. This internal muscle load contrasts with external training load, which represents a set of mechanical outputs (e.g., distance traveled at a given speed, load lifted, etc.) associated with the exercise task performed by the athlete. Therefore, internal muscle load depends on the individual performing the proposed training activity, while external training load, conversely, is individual-independent. Thus, to understand this load, muscle exertion must be determined, thereby providing usable data that allows for the quantification of that muscle exertion.

[0028] Therefore, this disclosure relates to a methodology for obtaining internal muscle load, and consequently to determining muscle force applied at the muscle level. The inventors have determined that muscle force can be calculated using non-invasive “external” sensors (i.e., without implantation), and that these sensors can also be integrated into textiles or clothing. Internal muscle load is obtained using two types of data: electromyographic data and inertial kinematic data (e.g., inertial data from an inertial measurement unit).

[0029] Therefore, this disclosure also relates to a textile equipped with electromyographic electrodes and an inertial measurement unit, the electromyographic electrodes being adapted to estimate muscle activation, and the inertial measurement unit being used to acquire joint angles and joint velocities. This disclosure also relates to a method for combining data obtained via the textile with other data to calculate composite data referred to as “muscle stress” (or muscle exertion). Finally, the developed method derives another composite index, called “internal muscle load” or “internal load,” which is the product of “muscle stress” and the duration of exertion during exercise or a set of exercises (muscle exertion × time of exertion).

[0030] like Figure 1 The described and implemented method includes the following steps: - Obtain kinematic data i(t) and electromyographic data emg(t) collected from at least one muscle of A10 during static or dynamic exertion. The data is obtained from sensors (e.g., sensors on clothing worn by the individual). For example, kinematic data i(t) and electromyographic data emg(t) are obtained using an inertial measurement unit (for inertial data) and electromyographic electrodes; - Based on kinematic data (i(t)), using a musculoskeletal model, determine the representative muscle-tendon length l of at least one muscle described in A20. MT Data for (t); -Based on electromyography data emg(t), determine the representative muscle activation level a(t) of at least one muscle described in A30; -Based on muscle-tendon length l MT(t) and muscle activation level a(t), along with two neuromuscular and muscle-tendon models M1 and M2, determine the muscle length l of at least one muscle described in A40. M (t) and representing the contraction velocity v M Data for (t); -Based on muscle length l M (t), determine the muscle deformation d of at least one muscle described in A50. M The estimated value of (t); -Based on muscle activation data a(t) and muscle deformation data d M (t) and depends on the contraction rate v M (t) muscle strength index (f FV (t) Calculate the A60 muscle stress index (s) M (t)); - By using muscle stress indexes M (t) Multiply by the duration t of the force exerted to calculate the internal muscle load C of A70. M Based on the electromyography (EMG) data emg(t) and the number of muscles covered and / or treated by the surface electrodes providing the EMG data emg(t), the internal muscle load of a single muscle M or a group of muscles can be calculated.

[0031] More specifically, in one instance, the method lies in basing the kinematics collected during static or dynamic force application. i ( t ) and electromyography emg ( t Data was used to determine the specific internal muscle load of the muscles involved in the analysis. C M ( t More specifically, the integration of this data captured from the clothing is implemented as follows. For example, kinematic data collected during exertion... i ( t Used in conjunction with a musculoskeletal geometry model (MSQ) to estimate overall skeletal kinematics and obtain the muscle-tendon lengths within the analyzed muscle groups. MT ( t As described, for example, in [Al Borno et al., 2022]. Since the rigid segments to which each inertial measurement unit is connected are known a priori, this kinematics can be estimated. For inertial data, the relative positions of the coordinate axes of the inertial measurement unit can be correlated with the relative positions of the coordinate axes of the musculoskeletal rigid segments. In another implementation, the kinematic data can be replaced by data from resistance goniometers placed within clothing, whose resistance varies with the angles of these goniometers for a given constant input current.

[0032] EMG data emg ( t The data undergoes sequential numerical transformations to estimate the level of neural activation received by each muscle within the analyzed muscle group. e ( t ).

[0033] Neuromuscular and muscle-tendon models are used to estimate activation levels. a ( t Relative muscle length l M ( t Relative muscle contraction speed v M ( t ) and depends only on the pre-obtained data e(t) and l MT ( t The analysis of the contraction velocity of muscles within a muscle group v M ( t Muscle strength index F FV ( t The evolution of time.

[0034] Within the scope of this patent application, a method is proposed that is generalizable and whose predicted quantification is specific to the individual and the muscle being analyzed.

[0035] Finally, the estimates and specific variables for each muscle group within the analyzed muscle group will be presented. a ( t ) 、 l M ( t )and F FV ( t Combined, to obtain the muscle stress index experienced by each of these muscles. s M ( t ).

[0036] Figure 2 The different steps of the general method described above are shown.

[0037] Data obtained from IMU and EMG of muscle groups (which may include, for example, two muscles). i ( t )and emg ( t Independent processing (MSQ, Processing) was performed to obtain the neural activation levels of the muscles that make up the muscle group. e1 ( t ), e2 ( t ) and muscle-tendon length l1MT ( t ), l2 MT ( t General or specific neuromuscular and muscle-tendon models (M1, M2) translate these variables into activation levels for each muscle analyzed. a1 ( t ), a2 ( t ), muscle length l1 M ( t ), l2 M ( t and muscle contraction speed v 1 M ( t ), v 2 M ( t The estimated value of ) Figure 1 (The image shows two muscles). Finally, the three estimated variables (activation level, muscle length, and contraction speed) are combined (MCM1, MCM2) to estimate the internal load of each analyzed muscle.

[0038] More specifically, in order to estimate internal muscle load, the variables activation a(t) and muscle deformation d are combined. M (t) and depends on the contraction rate v M ( t Muscle strength index f FV (t) is used to estimate muscle stress parameters. Based on muscle length, the muscle deformation d is determined using the following calculations. M (t): ,in l opt This is the optimal muscle length. The calculation of muscle strength metrics will be described further below.

[0039] Finally, multiply the muscle stress index by the duration of exertion to obtain the internal muscle load, as shown below. .

[0040] The value of s(%) can be determined in several ways. It can be the muscle stress index s(t) over the duration of exercise. It can be a percentage relative to the maximum muscle stress index (e.g., estimated relative to the maximum value of that exercise). Depending on the circumstances, particularly the physical activity or the desired goal, s(%) can be determined in other ways.

[0041] In this example, the technique integrates kinematic data into an MSQ (simulation) model and EMG data into a mathematical model to infer biomechanical stresses on the muscle (i.e., relative length and relative contraction velocity during muscle contraction), combining this with relative muscle activation and time to provide data representative of internal muscle load. Estimating internal muscle load through the implementation of the described method (activation and kinematic features in the numerical model) may help to better understand the root causes of inter-individual variability in the distribution of muscle injury and hypertrophy. Finally, monitoring internal muscle load during each training cycle is an indirect means of estimating changes in muscle strength (e.g., a decrease in internal muscle load under a given external load).

[0042] The following paragraphs describe various example implementations of the data processing operations and calculations performed to determine internal muscle load. It should be understood that those skilled in the art can adjust the calculations to suit specific conditions based on the chosen implementation. In particular, the musculoskeletal model does not necessarily need to be used in conjunction with a simulation platform; it can be directly integrated into a processing software module that performs simplified calculations, for example, according to a function of variable parameters (such as evaluating the lengths of different bones in the user's skeleton). This can be performed by the user themselves, for example, when a software application is installed on a communication device, performing all or part of the processing operations used to determine internal muscle load.

[0043] Processing kinematic data and determining muscle-tendon length l MT This example demonstrates the muscle-tendon length l used to obtain the muscle-tendon length for each muscle analyzed. MT ( t The methodology. In this example, by using kinematic data... i ( t (Inertial data here) is combined with musculoskeletal modeling (MSQ) to obtain an estimated value l. MT ( t ).

[0044] Raw inertial data is obtained using an inertial measurement unit (IMU). i 原始 ( t These inertial measurement units (IMUs) contain, in a non-restrictive manner, accelerometers, gyroscopes, and magnetometers, all of which are three-dimensional. (Raw data) i 原始 ( t Sensor fusion is used to combine, process, and transform data to obtain quaternions for each IMU. i ( t The time evolution of a rotation matrix in the form of a ) is shown. For example, sensor fusion algorithms are provided by certain IMU vendors (XSens™ or APDM™).

[0045] The desired outcome is one inertial measurement unit for each segment analyzed. For example, if using... Figure 1 To quantify the internal load of the right thigh muscles (including muscles attached to the pelvis and tibia), for example, three IMUs are placed within the clothing worn by the user, positioned on the pelvis, right thigh, and right calf, to obtain three sets of kinematic variable data. i 骨盆 ( t ), i 大腿 ( t )and i 胫骨 ( t These three variables (or groups of variables) are used to reproduce the user's skeletal posture during exercise.

[0046] Therefore, by using kinematic data i ( t Inverse kinematic analysis can virtually recreate the actual skeletal kinematics of an individual during the applied force. This is to allow for the reconstructing of the IMU's orientation history. i ( t ) and angular kinematics of the skeletal system a ( t This is done by linking the user's skeleton to a computer-generated skeletal model (either specifically or generally) that represents the user's bones. This skeletal model models the joint dynamics during the application of forces. This modeling can be performed using platforms such as Opensim (a platform for biomechanical modeling, simulation, and analysis). Under operating conditions, calculations performed via such platforms can be directly integrated into the computing device.

[0047] First, the calibration step adjusts the orientation of the IMU and virtually represents it in an orthogonal coordinate system of the relevant virtual bone segment by eliminating any potential constant deviation between the real IMU and the virtual segment. This step is achieved through brief static acquisitions.

[0048] Once the measurement unit has been calibrated, it can be based on inertial data. i ( t Inverse kinematics simulations were performed. This was achieved by minimizing the MSQ model bone segment and experimental quaternion within each time step. i ( t The differences between orientations are virtually reproduced using the MSQ model to recreate the subject's actual angular kinematics. For example, if using... Figure 1 By studying the internal load of the right thigh muscles (including those attached to the pelvis and tibia), data from three IMUs would allow for accurate and drift-free determination of the angular kinematics of the hip and knee joints. a ( tThese calculations can be performed using software solutions. [Al Borno et al., 2022] describes an example of such a software solution (OpenSense) that allows for the acquisition of angular kinematics of the lower limbs. a ( t ).

[0049] To obtain the muscle-tendon length l of the muscles analyzed during exercise MT ( t The time evolution of ) using angular skeleton kinematics a ( t The relationship between muscle-tendon length and muscle-tendon length, i.e., using (establishing) type l MT ( a The relationship between muscles and bone structures can be described (parameterized) within a (generic or user-specific) skeletal model. To do this, muscles can be defined as unidirectional actuators, attached to the bone structure using two insertion points and an origin, and to supplementary channel points as needed.

[0050] The length of the bone segments and their respective orientations provide angular kinematics a ( t The placement of the muscle-tendon system's defining point on the bone segment three-dimensionally defines the muscle-tendon length. MT ( t ).For example, Figure 3 This displays the quantitative values ​​of the semimembranosus (light gray, dashed line), semitendinosus (black, dashed line), and biceps femoris (dark gray, solid line) muscles during a seven-step running cycle. MT Examples of time evolution (unit: meters).

[0051] Regardless of the actual calculation method used, the muscle-tendon length l, which belongs to the study of muscle groups, is... MT ( t The temporal evolution is based on kinematic data recorded by the IMU during static or dynamic exertion. i ( t Estimated. Quantitative l is estimated using geometric methods (e.g., via the MSQ model). MT Joint kinematics is derived from experimental kinematic data. i ( t It is limited to inverse kinematics and uses the unidirectional geometry of the muscle under consideration.

[0052] Processing electromyography data and determining neural activation levels e ( t ) As explained above, electromyography (EMG) data emg ( t This provides information for studying the neural activation levels of muscles.e ( t ).

[0053] The number of EMG electrodes placed on the muscle group under study varies and is often integrated into the clothing worn by the user. The surface area and shape of the clothing (e.g., bras, cycling shorts, leggings, T-shirts) determine the number of electrodes used. Depending on the operating conditions (e.g., depending on the clothing or the area being studied), between 2 and 256 electrodes can be used, spaced a few millimeters or centimeters apart. Depending on the specific implementation, these electrodes may be mounted in a bipolar configuration. Such mounting provides a representative measurement of muscle activation in a muscle or muscle group. These electrodes collect the surface electromyographic activity of the muscles covered by the electrodes during the application of force. emg ( t ).

[0054] To eliminate ineffective generation components from the collected EMG signal, the original EMG signal was analyzed in the frequency domain. emg ( t The signal is then filtered. The obtained envelope is rectified. Finally, the resulting signal is normalized using the maximum amplitude of the electromyographic signal recorded during maximal voluntary contraction. The processed signal... emg ( t This will provide the level of neural activation in the muscles covered by these electrodes. e ( t ).

[0055] Determine activation level a ( t Relative muscle length l M ( t and relative muscle contraction speed v M ( t ) Once quantitative neural activation has been obtained based on EMG and IMU data... e ( t ) and muscle-tendon length l MT (t) The time evolution of these variables is then integrated into mathematical neuromuscular and muscle-tendon models to identify the three variables necessary for estimating internal muscle load: a ( t ) 、 l M ( t )and v M ( t ).

[0056] The inventors have determined that these variables can be estimated using Hill's muscle model. In this case, a nonlinear passive elastic element representing tendon mechanical properties and a nonlinear active element representing muscle contraction mechanics are modeled in mathematical equilibrium.

[0057] Tendons are composed of their length l T Heli F T Limited, while the active muscle element is determined by its length l M contraction speed v M (Both are to be determined in this paper) and force F M Limitations. Use the following two relations to identify the values ​​acquired by these three variables:

[0058] Force F passively generated by tendon T Following the exponential force-deformation law, its stiffness coefficient and non-force transmission length It is specific to the muscle being studied, and can be measured in tables or experimentally.

[0059] Force F generated by active components (muscles) M Calculated as the product of three factors: muscle activation a ( t Force-length factor f FL ( t ) and force-velocity factor f FV ( t (or muscle strength indicators):

[0060] Muscle activation a ( t The value represents the muscle activation level between 0 and 1. It takes into account the ratio of involved motor units, the firing frequency of these motor units, the type of myofibrils involved, and sarcomere dynamics within the muscle. Using a nonlinear time equation and typical force-EMG nonlinear static relationships obtained from existing literature, the activation level is... a ( t Estimated as muscle nerve activation level e ( t The function of ). Other physiological models using the mass principle can also be conceived as functions under specific conditions.

[0061] Force-Length (FL) Relationship f FL ( tThe FL relationship provides a function of the maximum force a muscle can generate under isometric conditions and its length. While it originates at the molecular level, describing the average overlap of sarcomeres within the muscle, this FL relationship has been measured at both the molecular and muscular levels in humans and exhibits a Gaussian distribution. The maximum force is obtained at a specific muscle length, termed the optimal isometric muscle length. Furthermore, it is specific to the muscles being studied. Optimal isometric muscle length values ​​are available. The table (see, for example, [Rajagopal et al., 2016]). During the submaximal contraction, this FL relationship changes with the activation threshold. a ( t And change, such as Figure 4b The figure shows the dashed lines. In this graph, the solid curves represent the maximum activation level, while the dashed lines represent lower activation levels.

[0062] Force-velocity (FV) relationship f FV ( t ) provides the optimal length The maximum instantaneous force a muscle can generate at its maximum activation level is a function of its contraction velocity. It is generally accepted that during concentric and eccentric exertion, f FV The contraction rate will decrease and increase hyperbolically, respectively. Therefore, this relationship is defined by two hyperbolas that intersect at the point where the contraction rate is zero. C 2 .like Figure 4a As shown, in this FV relationship, the maximum shortening speed v 最大 The hyperbolic increase rate and the hyperbolic increase rate increased and decreased with muscle activation level a, respectively. Figure 4a The force-velocity relationship is shown at different activation levels a and length l. With relative velocity The function. Figure 4a This includes curves for slow-acting fibers and curves for fast-acting fibers. Solid lines represent the maximum activation level, while dashed / dotted lines convey the effect of the activation level. Figure 4b The force-length relationship is shown at different activation levels 'a'. f FL ( l,a ) is a function of length l.

[0063] Using the aforementioned defined neurophysiological model, neural activation can be observed. e ( t This directly estimates the temporal evolution of muscle activation. In this case, it yields... a ( t,e ( t )).

[0064] To estimate muscle length l M ( t ) and its contraction speed v M ( t The temporal evolution of muscles and tendons must address the balance between them at every moment.

[0065] In any case, l T ( t ) = l MT ( t ) - l M ( t ), where l MT ( t As instructed above, l M ( t () is an unknown variable. From muscle-tendon balance and muscle-based active properties, we know that:

[0066] By rearranging the expression, the FV property was reversed, and muscle contraction speed was observed. v M ( t ) is the muscle length l M ( t From the derivative of ), we know that:

[0067] The differential equation is expressed in a known quantitative form. MT ( t )and e ( t (See above) as input, and has its unique variable as (to be determined) quantitative l M ( t At each time step, the equation is solved numerically to obtain the variable l. M ( t Based on this, the muscle contraction velocity can be directly obtained through the time derivative. v M ( t ).

[0068] Furthermore, depending on the implementation method, the tendon length can be set to a fixed constant l. T This method is simplified by assuming that the tendon can transmit muscle force regardless of the contraction condition. In this case, because l M ( t ) = l MT ( t) - l T and You can directly know l M and v M This avoids the aforementioned differential equations (and the computations they generate, thus simplifying the implementation and accelerating computation). This simplification can be made for certain muscles with short or stiff tendons (as is the case for most muscles in the human body), for example, or when computational resources on the measuring device are limited.

[0069] Therefore, using neuromuscular and muscle-tendon models, three variables can be estimated to determine internal muscle load or training load: quantitative neural activation based on EMG and IMU data. e ( t ) and muscle-tendon length l MT ( t The evolution of time a ( t ),l M ( t )and v M ( t ).

[0070] Then, as explained above, these variables are combined to obtain the muscle stress index. s M ( t ) and internal muscle load C M ( t ).

[0071] The Hill neuromuscular-tendon model defined above is universal because the muscle and tendon lengths used above are l. M and l T It is normalized / relative. This is achieved by introducing the aforementioned defined length. and The absolute values ​​of these lengths can scale the neuromuscular-tendon model to a specific muscle. According to sensitivity studies, these values ​​are muscle-specific and are key factors in the accuracy of muscle dynamics predictions ([Scovil & Ronsky, 2006]; [Redl et al., 2007]; [Xiao & Higginson, 2010]; [Carbone et al., 2016]). and Previous estimations were based on cadaver data ([Ward et al., 2009]). Therefore, each muscle has its own specific ratio. The table contains the typical value for an adult human of 170 cm.

[0072] Other parameters in the neuromuscular-tendon model can be adjusted for each muscle, such as the rate of increase in the force-velocity relationship or the slope of the EMG-force relationship. However, these values ​​are not applicable to all muscles in the human body, and vary with muscle length l. M ( t The predictive effect of these values ​​is weak, therefore they are generally applicable to the muscles being studied. These values ​​can be calibrated by minimizing the difference between the experimental and estimated times. Other parameters, such as tendon stiffness, can be estimated experimentally.

[0073] Finally, the maximum isometric force that a muscle can generate varies from muscle to muscle, but a quantitative "absolute force" is not useful in the method developed in this paper.

[0074] In addition, it is used for kinematic data i ( t Obtain muscle-tendon length l MT of Figure 3 The MSQ model used in this study is also general and does not represent the bone geometry of subjects with specific bone segment lengths. Therefore, this general MSQ model is not suitable for describing muscle-tendon length l, which varies non-linearly with bone length. MT The scaling may be limited. It can be scaled by giving the generic MSQ model the same dimensions as the individual being studied. In this case, all skeletal segment lengths are linearly adjusted and multiplied by a factor k, which is the ratio of the individual's height to the generic model height (individual height 170cm). The muscle-tendon length l can then be recalculated using the scaled MSQ model. MT Nonlinear scaling is used. This scaling can be automated, for example, using the simulation platform OpenSim. However, this scaling method is limited because the bone length ratio remains constant before and after scaling and is specific to a general MSQ model (rather than an individual). Higher specificity can be achieved by measuring individual segments, using nonlinear bone length relationships from existing literature, or automatically creating subject-specific MSQ models based on bone and muscle volumes obtained after segmenting medical images of the subject. Tools exist for creating such models (see, for example, [Modenese & Renault, 2021]). However, this approach requires medical images and a considerable amount of time-segmented medical images.

[0075] In any case, the muscle-tendon length obtained experimentally is used for general human use. and ([Ward et al., 2009]) will be determined by the length ratio before and after bone scaling l MT Alternatively, more advanced calibration methods can be used for linear adjustment ([Modenese et al., 2016]).

[0076] Other features and advantages In one example implementation, the method can be implemented within the following overall framework.

[0077] Users are equipped with smart clothing, such as cycling shorts or leggings, incorporating the aforementioned sensors. Following guidance from a third party (e.g., a sports trainer, coach, etc.) or according to a user-defined plan, the user performs one or more exercises to target all or part of the muscle and tendon groups covered by the smart clothing. During the exercise, the smart clothing captures data from one or more of the targeted muscle groups. This data is recorded in a suitable recording device. According to an example implementation, this recording device can be located within the smart clothing itself or in other devices connected to the smart clothing, such as, for example, a tablet or smartphone connected to the smart clothing via a wired or wireless connection. A wired connection can consist of a USB cable connecting a smartphone or tablet to the smart clothing. This option is attractive because the smart clothing therefore does not need its own power source, which is provided via the smartphone or tablet. This makes the clothing easier to manufacture. Wireless connections can take the form of BLE™ (Bluetooth Low Energy) or Zigbee™ connections. While this solution makes the production of smart clothing more complex, it offers greater flexibility in its use.

[0078] In any case, data from a user's physical exercise is stored as it is performed. Once stored, this data can be processed according to the methods described above for determining internal muscle load in order to determine the internal muscle load of the user's exercise.

[0079] refer to Figure 6 This describes an instance system of implementations of all or part of the methods described in this disclosure.

[0080] As explained above, in this example, the user is equipped with a smart garment (VeTI) featuring BLE-type wireless connectivity. The smart garment (VeTI) is paired with a tablet-type communication device (DisPC) via this wireless connection. The communication device (DisPC) is then connected to a communication network, such as via WiFi or a 4G / 5G wireless connection. The communication device (DisPC) also includes an application, which is installed before using the smart garment (VeTI), and is capable of: storing data captured from the smart garment (VeTI) during exercise; transmitting all or part of this data to a processing server (SrVT); and receiving data representing internal muscle load from the server (SrvT). Therefore, in this example, a remote server is responsible for calculating the internal muscle load on behalf of the communication device (DisPC). According to the example implementation, the communication device (DisPC) can transmit all data for a given exercise recording. In another implementation, the communication device (DisPC) can also transmit pre-processed data: once exercise data has been recorded, the DisPC can clean and normalize the data to facilitate processing by the processing server, and limit the amount of data to be transmitted to the server to reduce the bandwidth required for data transmission and save resources overall. The transmitted data can also be compressed to further reduce the amount of data to be transmitted. The processing server (SrVT) then receives the data from the DisPC and decompresses it if necessary. The processing server then identifies the physiological and morphological data of the user associated with these exercise records, as well as the muscle groups involved during exercise, according to the request content passed to it by the DisPC. The request transmitted by the DisPC may include these information elements. The request transmitted by the DisPC may additionally or alternatively include a user identifier. The processing server then extracts the user identifier to retrieve the user's personalized characteristics (e.g., physiological and morphological data) from a database accessible to the server.

[0081] Based on this data, the server implements the method described above to calculate the internal muscle load of one or more muscles or muscle groups as a function of the exercise performed (the transmitted records). Once the internal muscle load is calculated, the server returns it to the communication device (DisPC), allowing it to be processed according to program instructions from an application installed on the DisPC. This application can process the internal muscle load calculated by the server, for example, by incorporating it into an internal muscle load history or cumulative index. This data can then be integrated into a training plan application module, enabling the module to plan one or more additional training sessions based on the calculated internal muscle load as a function, aimed at reducing or conversely increasing that internal muscle load.

[0082] refer to Figure 5This paper describes a simplified architecture of a device for determining internal muscle load, capable of implementing the methods for determining internal muscle load shown above. Such a device for determining internal muscle load includes a memory 51 and a processing unit 52, the processing unit being equipped with, for example, a microprocessor and controlled by a computer program 53 to implement the method according to the invention. In at least one embodiment, the invention is implemented as an application program installed on a communication device specifically for determining internal muscle load. Such devices for determining internal muscle load include, for example, all or part of the following means: - A device for acquiring kinematic data i(t) and electromyographic data emg(t) collected during static or dynamic exertion of the at least one muscle; - A determining device for determining the muscle-tendon length l of the at least one muscle based on kinematic data i(t) and electromyographic data emg(t). MT Data representing activation level a(t), and muscle length l. M The data (t) represents the contraction speed v. M The data for (t) and the data that depend only on the contraction rate v M The force index F of (t) FV (t); - A computing device, configured to base on muscle activation data a(t) and muscle length l M (t) and force index F FV (t) Calculate the muscle stress index s M (t), and configured to be based on muscle stress s M (t) and its duration t are used to calculate the internal muscle load C. M The data represents the internal muscle load C. M .

[0083] These devices can take the form of specific software applications or hardware components specifically built to perform these functions, such as secure elements (SEs) or secure execution environments. Secure elements can take the form of SIM cards, USIM cards, UICC cards, or specific secure components.

[0084] Explanation of the relationship between the location of muscle soreness and the location of internal muscle load Figure 7 The relationships between muscle injury distribution and (a) the internal muscle load distribution measured according to the method of the present invention, and (b) the muscle activation distribution, are illustrated. These relationships are illustrated for three muscles: semitendinosus, biceps femoris, and semimembranosus. These figures are generated based on data on muscle soreness distribution described in the inventors’ article [Goreau et al. (2022)].

[0085] This assumes a maximum muscle contraction velocity of 10 ms. - ¹, The force-velocity factor for maximum eccentric force is 1.4 (1.0 = maximum isometric force). The duration of force (d, sec) is calculated by identifying the start and end of contraction based on EMG signals. The duration of each contraction is considered to be the same between muscles. The internal muscle load (a) as described above is calculated using muscle activation measured by surface EMG during maximum eccentric contraction and joint angles measured by a force gauge. M f FV l M Factor a) M f FV l M The product of d and d yields the internal muscle load (in arbitrary units) for each muscle.

[0086] Figure 7 As shown in (a), there is a close relationship between the location of muscle soreness and the location of internal muscle load. The location of muscle soreness was estimated using elastography. Muscle contrast analysis showed that the semitendinosus muscle received the highest internal muscle load and exhibited the highest level of muscle soreness. This implies that the measured internal muscle load does indeed have a structural impact on the integrity of the trained muscles. This clearly demonstrates that the internal muscle load calculated according to the present invention does indeed have physiological values ​​that can be verified by indicators that can be calculated from muscle damage. It should be noted that damage distributions of less than 0% or greater than 100% can be explained by negative changes in the shear modulus of one of the three muscles measured (i.e., no muscle damage).

[0087] Figure 7 As shown in (b), activation alone is insufficient to pinpoint the location of muscle damage. As explained by [Goreau et al. (2022)], there is indeed a relationship between muscle soreness and muscle activation when considering a single muscle as well as comparing multiple individuals. However, to compare different muscles, it is also necessary to consider the biomechanical stresses specific to each muscle and each individual. More specifically, it is necessary to consider other factors and calculate the internal muscle load as described in this invention in order to estimate muscle stress or damage.

[0088] References [Rajagopal et al., 2016]: Rajagopal, Apoorva, et al. “Full-bodymusculoskeletal model for muscle-driven simulation of human gait.” IEEE Transactions on Biomedical Engineering 63.10 (2016): 2068-2079. [Modenese&Renault, 2021]:Modenese, Luca, and Jean-Baptiste Renault. “Automatic generation of personalised skeletal models of the lower limb from three-dimensional bone geometries.” Journal of Biomechanics 116 (2021):110186. [Scovil&Ronsky, 2006]:Scovil, Carol Y., and Janet L. Ronsky. “Sensitivity of a Hill-based muscle model to perturbations in model parameters.” Journal of Biomechanics 39.11 (2006): 2055-2063 [Redl et al., 2007]:Redl, Christian, Margit Gfoehler, and Marcus G. Pandy. “Sensitivity of muscle force estimates to variations in muscle–tendon properties.” Human Movement Science 26.2 (2007): 306-319. [Xiao&Higginson, 2010]:Xiao, Ming, and Jill Higginson. “Sensitivity of estimated muscle force in forward simulation of normal walking.” Journal of applied biomechanics 26.2 (2010): 142-149. [Carbone et al., 2016]:Carbone, Vincenzo, et al. “Sensitivity of subject-specific models to Hill muscle–tendon model parameters in simulations of gait.” Journal of Biomechanics 49.9 (2016): 1953-1960. [Ward et al., 2009]: Ward, Samuel R., et al. “Are current measurements of lower extremity muscle architecture accurate?” Clinical Orthopaedics and Related Research 467.4 (2009): 1074 - 1082. [Modenese et al., 2016]: Modenese, Luca, et al. “Estimation of musculotendon parameters for scaled and subject specific musculoskeletal models using an optimisation technique.” Journal of Biomechanics 49.2 (2016):141 - 148. [Al Borno et al., 2022]: Al Borno, M., O’Day, J., Ibarra, V. et al. OpenSense: An open - source toolbox for inertial - measurement - unit - based measurement of lower extremity kinematics over long durations. J NeuroEngineering Rehabil 19, 22 (2022). https: / / doi.org / 10.1186 / s12984 - 022 - 01001 - x [Goreau et al. (2022)]: Goreau, Valentin, et al. “Hamstring muscle activation strategies during eccentric contractions are related to the distribution of muscle damage.” Scandinavian Journal of Medicine&Science in Sports 32.9 (2022): 1335 - 1345。

Claims

1. A method for determining internal muscle loads representing those generated by at least one muscle belonging to a muscle group. C M The method for processing data, the method being implemented by means of an electronic device including a processor and a memory, wherein, The method includes the following steps: - Obtain (A10) kinematic data collected during the contraction of at least one muscle. i(t)) and electromyography data ( emg(t)) ; -Based on the kinematic data ( i(t)) Using a musculoskeletal model, determine the representative muscle-tendon length of at least one muscle (A20). l MT Data for (t)); -Based on the electromyography data ( emg(t)) Determine data representing the muscle activation level (a(t)) of at least one muscle as described in (A30); -Based on the muscle-tendon length ( l MT (t)) and muscle activation level (a(t)) and two neuromuscular and muscle-tendon models (M1, M2) to determine the muscle length of at least one muscle (A40). l M (t) and representing the contraction rate ( v M Data for (t)); -Based on the muscle length ( l M (t)), determine the muscle deformation (d) of at least one muscle as described in (A50). M The estimated value of (t)); -Based on the muscle activation data (a(t)) and the muscle deformation (d) M (t) and depends on the contraction rate ( v M (t)) muscle strength index (f FV (t) Calculate (A60) muscle stress index ( s M (t)); and - By using the muscle stress index ( s M Multiply the force applied (t) by the duration of the force applied (t) to calculate the internal muscle load (A70). C M ).

2. The method according to claim 1, wherein, Determine (A20) the muscle-tendon length of the at least one muscle. l MT The steps of (t) include sub-steps: - Simulate the joint angle (a(t)) of joint movement during the exertion according to the kinematic data (i(t)) and the function of the musculoskeletal geometry model; and -Based on the joint angle (a(t)) described by the musculoskeletal model and the musculoskeletal geometry, calculate the muscle-tendon length ( l MT The temporal evolution of (t)).

3. The method according to claim 1 or claim 2, wherein, The step of determining (A30) data representing the muscle activation level (a(t)) of said at least one muscle includes the following sub-steps: - In the frequency domain, the electromyography data ( emg(t)) Filtering is performed to provide the signal envelope. emg(t)) ; - Rectify the signal envelope ( emg(t)) ; -Normalize the signal envelope ( emg(t)) This provides a normalized envelope describing the level of neural activation (e(t)) of the muscle. and - Based on the neural activation level (e(t)) of the muscle, calculate data representing muscle activation (a(t)).

4. An electronic device comprising a processor and a memory, arranged to determine an internal muscle load representing that generated by at least one muscle belonging to a muscle group. C M The data, among which, The device includes: - A device for acquiring kinematic data of the at least one muscle during static or dynamic exertion. i (t)) and electromyography data ( emg(t)) ; - Determining device, for determining based on the kinematic data ( i(t)) and electromyography data ( emg(t)) Determine the muscle-tendon length of the at least one muscle. l MT (t)), data representing muscle activation level a(t), and data representing muscle length ( l M (t) data, muscle deformation (d) M The estimated value of (t) and the value representing the contraction rate ( v M The data for (t)); and - A computing device, configured to base on the muscle activation data (a(t)) and the muscle deformation (d) M (t) and depends on the contraction rate ( v M (t)) muscle strength index (f FV (t)), calculate muscle stress index ( s M (t)), and configured to calculate the internal muscle load (t) C M (Data).

5. An electronic system comprising at least one processor and at least one memory, arranged to determine internal muscle load representing that generated by at least one muscle belonging to a muscle group. C M The system includes smart clothing (VeTI), which is paired with a communication device (DisPC) and connected to a processing server (SrVT) via a communication network. The electronic system further includes: - A device for acquiring kinematic data collected during static or dynamic exertion of at least one muscle in contact with the smart garment (VeTI). i(t)) and electromyography data ( emg(t)) ; - Determining device, for determining based on the kinematic data ( i(t)) and electromyography data ( emg(t)) Determine the muscle-tendon length of the at least one muscle. l MT (t)), data representing activation level (a(t)), and data representing muscle length ( l M (t) data, muscle deformation (d) M The estimated value of (t) and the value representing the contraction rate ( v M The data for (t)); and - A computing device, configured to base on the muscle activation data (a(t)) and the muscle deformation (d) M (t) and depends on the contraction rate ( v M (t)) force index (f FV (t)), calculate muscle stress index ( s M (t)), and configured to calculate the internal muscle load (t) C M (data) -The obtaining device, the determining device, and the computing device are distributed among the various components of the system.

6. A computer program product comprising program code instructions for implementing the determination method when the determination method according to any one of claims 1 to 3 is executed by a processor.

Images