How to estimate interest point trajectories

Abstract

This method for estimating the trajectory of a point of interest belonging to an individual's limb over a displacement interval includes estimating (1320) a change in velocity of a measurement point attached to the point of interest during the displacement interval, estimating (1310) a change in orientation of the measurement point during the displacement interval, and estimating the trajectory of the point of interest during the displacement interval from the estimated velocity and orientation of the measurement point.

Description

[Technical Field]

[0001] The present invention relates to the field of estimating the change in linear velocity of a point from measurements made using inertial devices.

[0002] The invention also relates to estimating the trajectory of a point of interest belonging to a limb of an individual from measurements made by an inertial device, and using said estimation to evaluate and analyze the forces exerted by said limb, in particular when said limb is an upper limb. [Background technology]

[0003] For an individual, moving the upper limbs depends on the use and coordination of multiple muscles. Therefore, analysis of an individual's movements can be used to characterize the individual's fitness level, and in particular the muscle strength that the latter (individual) is able to develop. This muscle strength can vary depending on many factors, such as the individual's fitness level, age, physical training, or whether they have undergone treatment that affects muscle efficiency.

[0004] Changes in an individual's movement over time can indicate changes in physical fitness, especially in individuals suffering from neuromuscular or neurodegenerative diseases. These diseases, such as Duchenne muscular dystrophy (DMD), cause an individual's muscle strength to decline as the disease progresses, resulting in a loss of mobility for affected patients.

[0005] WO2019 / 243609 discloses a method for analyzing a walker's stride length from measurements obtained using an inertial device attached to the walker's ankle, with the aim of characterizing changes in the walker's fitness. Such a method is minimally invasive and has very particular application in tracking neuromuscular or neurodegenerative diseases. However, this method is not suitable for tracking such diseases only if the individual retains sufficient motor function in the lower limbs. Another method is needed to be able to track the disease when the individual can no longer walk (typically because the disease is in an advanced phase).

[0006] Thus, a consensus has emerged in the scientific community regarding the need to be able to accurately quantify upper limb motor function in individuals suffering from neuromuscular or neurodegenerative diseases. This consensus has led to several methods for assessing this motor function.

[0007] Most of these assessment methods consist of traditional methods based on functional scales assessed by specialized clinicians in controlled environments. Typically, individuals are referred to a clinic where they are asked to perform specific tasks, and based on the success or failure of these tasks, the clinician assigns a motor score. These methods include, for example: the Brooke scale (Brooke, MH et al. “Clinical investigation in Duchenne dystrophy: 2. Determination of the “power” of therapeutic trials based on the natural history.” Muscle & nerve vol. 6,2 (1983): 91-103. doi:10.10002 / mus.880060204) the revised upper limb module (better known by the acronym RULM) (Mazzone, Elena S et al. “Revised upper limb module for spinal muscular atrophy: Development of a new module.” Muscle & nerve vol. 55,6(2017): 869-874. doi: 10.10002 / mus.25430) the performance of upper limb module (better known by the acronym PULM) (Mayhew, Anna G et al. “Performance of Upper Limb module for Duchenne muscular dystrophy.” Developmental medicine and child neurology vol. 62,5 (2020): 633-639. doi:10.1111 / dmcn.14361).

[0008] However, these methods have several drawbacks: they are restrictive for patients, require them to travel to the hospital, and the results may vary depending on the patient's health condition on the day of the evaluation and the evaluator.

[0009] Recently, other methods based on the use of information technology have emerged. Most involve the analysis of data recorded in a controlled environment, typically in a hospital, using optical capture systems under the supervision of a medical professional, as described, for example, in Han, JJ et al. "Reachable workspace and performance of upper limb (PUL) in Duchenne muscular dystrophy." Muscle & nerve vol. 53 (2016): 545-554. doi: 10.10002 / mus.24894. However, these methods suffer from many of the drawbacks already identified with previous methods, particularly limitations for patients and variable results depending on the patient's fitness level on the day of assessment.

[0010] To overcome these problems, solutions have been sought that allow the assessment of upper limb motor function over long acquisition times in the patient's daily life. In most cases, assessments are performed by measurements performed by inertial devices attached to the upper limb. These inertial devices must meet particularly strict constraints regarding weight, size, and autonomy to allow portability, and are composed of MEMS-type sensors. However, such sensors have a large time drift, which makes tracking movements using the measurements returned by these sensors very complicated.

[0011] To circumvent this difficulty, WO2017 / 129890 proposes assessing the motor function of the upper limbs only for a very specific movement in which the elbow is placed motionless on a surface and the movement of the wrist, to which an inertial device is attached, traces an arc centered on the elbow. However, this solution is not entirely satisfactory because the assessment of motor function does not take into account many movements of the upper limbs.

[0012] El-Gohary, Mahmoud, and James McNames. “Shoulder and elbow joint angle tracking with inertial sensors.” IEEE transactions on bio-medical engineering vol. 59,9 (2012): 2635-41. doi: 10.11009 / TBME.2012.2208750 also proposes correcting sensor drift by combining a kinematic model of the upper limb with measurements acquired by two inertial sensors, one positioned on the arm and one on the forearm. However, this solution using multiple inertial sensors is too invasive for routine use.

[0013] Finally, Villeneuve, Emma et al. "Reconstruction of Angular Kinematics From Wrist-Worn Inertial Sensor Data for Smart Home Healthcare." IEEE Access, 5 (2017): 2351-2363. doi: 10.11009 / access.2016.2640559 proposed a method to estimate an individual's wrist kinematics from measurements provided by a simple wrist-worn accelerometer using a particle filter and imposing constraints related to respecting a biomechanical model of the individual's upper limbs. However, even though this method can estimate wrist kinematics without using a gyrometer, it does not solve the problem of sensor drift over time.

[0014] Another difficulty encountered when using inertial devices to assess upper limb motor function is that upper limb movements are much more diverse and difficult to recognize than lower limb movements, and that, unlike the latter, upper limb movements are often related not only to their own muscle activity but also to muscle activity in other parts of the body, particularly the trunk or lower limbs. Summary of the Invention

[0015] One of the goals of the present invention is to enable tracking of an individual's physical fitness over time. Another goal is to enable proper estimation of the physical forces exerted by an individual's upper limbs for various movements over an extended period of time in an uncontrolled environment. Other goals are to minimize the invasiveness of the device used to perform this estimation and to improve the quality of the estimation derived from inertial device measurements.

[0016] Therefore, the present invention relates according to a first aspect to a method for estimating the change in linear speed of a measurement point during a displacement interval, said method being implemented by a data processing unit, said method comprising: determining initial and final velocities of the measurement point at the beginning and end of a displacement interval, respectively; and estimating the evolution of the velocity of the measurement point during the displacement interval, the estimation comprising: performing a forward integration of displacement parameters of the measurement points acquired by the inertial device during a displacement interval, taking an initial velocity value as an initial velocity value; performing a backward integration of the displacement parameters, taking the value of the final velocity as the initial value of the velocity; and calculating an estimated velocity by merging the backward integral and the forward integral.

[0017] According to a second aspect, the present invention also relates to a method for estimating a trajectory over a displacement interval of a point of interest belonging to a limb of an individual, the method being implemented by a data processing unit, the method comprising the steps of: estimating the change in velocity of a measurement point attached to the point of interest during a displacement interval; estimating a change in orientation of the measurement point during a displacement interval; and c) estimating the trajectory of the point of interest during the displacement interval from the estimated velocity and orientation of the measurement points.

[0018] According to particular embodiments of the invention, the orbit estimation method also has one or more of the following characteristics, taken alone or in any technically possible combination: An individual is a human being; The limb is the individual's upper limb; The points of interest are points on the forearm of the upper limb; Limbs are the lower limbs of an individual; The points of interest are points on the legs of the lower limbs; The change in velocity of the measurement point during the displacement interval is estimated by the method according to the first aspect; The orbit estimation is calculating a number of candidate trajectories by applying the estimated velocity and orientation to different pairs of starting and ending configurations of a biomechanical model of the limb; assessing the compatibility of each candidate trajectory with a biomechanical model; and selecting the most suitable candidate trajectory.

[0019] The points of interest are points on the forearm belonging to the upper limb of an individual, and each starting configuration of the biomechanical model of the limb is defined by a pair of starting angles formed by a first starting angle between the forearm and a vertical line and a second starting angle between the arm and the vertical line, and each ending configuration of the biomechanical model of the limb is defined by a pair of ending angles formed by a first ending angle between the forearm and the vertical line and a second ending angle between the arm and the vertical line.

[0020] For each starting or ending configuration of the limb biomechanical model: a first starting angle or finishing angle (respectively) is estimated from the orientation of the measurement point at the beginning or end of the displacement interval; The second starting or ending angle is less than the first starting or ending angle.

[0021] For each value of the second start angle that is less than the first start angle, and for each second end angle value within the predetermined interval, there is a pair of start and end configurations from which candidate trajectories are calculated.

[0022] The predetermined interval is the interval between 0° and the value of the first end angle, or the angle β t and the following relationship holds:

number

number

number

number

number

number

[0023] The calculation of each candidate trajectory is as follows: performing a forward integration of the estimated velocity using the position of the measurement point in the starting configuration as an initial position; a step of performing backward integration of the estimated velocity using the position of the measurement point in the final configuration as an initial position; and calculating a candidate trajectory by merging the backward and forward integrals.

[0024] The present invention also relates according to a third aspect to a method for assessing forces exerted by a limb of an individual, the method being implemented by a data processing unit, the method comprising: - estimating a trajectory of a point of interest belonging to a limb over at least one primary displacement interval by a method according to the second aspect; and estimating from the estimated trajectory and velocity changes the primary force exerted by the limb during the or each primary displacement interval.

[0025] According to a particular embodiment of the invention, the force assessment method also has one or more of the following characteristics, taken alone or in any technically possible combination: the method estimating a rotational component of primary kinetic energy acquired by at least a portion of the limb during the primary displacement interval from the displacement parameters acquired during the primary displacement interval, wherein the primary force is a function of the rotational component of the primary kinetic energy; determining a relationship between the primary force and the rotational component of the primary kinetic energy; estimating a change in orientation of a measurement point attached to the point of interest during at least one secondary displacement interval; estimating a rotational component of secondary kinetic energy acquired by at least a portion of the limb during the secondary displacement interval from displacement parameters acquired by the inertial device during the secondary displacement interval; and estimating from the determined relationship and the rotational component of the secondary kinetic energy the secondary force exerted by the limb during the or each secondary displacement interval.

[0026] This method involves the following for each of the primary and secondary displacement intervals: estimating a change in velocity of a measurement point attached to the point of interest during a primary displacement interval or a secondary displacement interval; calculating a number of candidate trajectories by applying the estimated velocity and orientation to different pairs of starting and ending configurations of a biomechanical model of the limb; evaluating the fit of each candidate trajectory with a biomechanical model; for each primary displacement interval or each secondary displacement interval, verifying the acceptability of the suitability of at least one candidate trajectory; A trajectory of an interest point is estimated only if at least one candidate trajectory has an acceptable fit. The change in velocity of the measurement point during the primary or secondary displacement interval is estimated by the method according to the first aspect.

[0027] The present invention also relates according to a fourth aspect to a method for analysing forces exerted by a limb of an individual, the method being implemented by a data processing unit, the method comprising: a) estimating the force exerted by a limb by a method according to the third aspect for a number of intervals of displacement of a point of interest of the limb occurring during a predetermined period of time; b) determining the value of at least one statistic that is calculated based on the set formed by the estimated forces and that represents the fitness state of the individual.

[0028] According to a particular embodiment of the invention, the force analysis method also has the following features: The method includes repeating steps a) and b) over a number of predetermined time periods and observing changes in the or each statistic over different predetermined time periods.

[0029] The present invention also relates, according to a fifth aspect, to a method of tracking the fitness level of an individual, the method comprising the steps of analysing forces exerted by the limbs of the individual by an analysis method according to the fourth aspect and inferring the fitness level of the individual from the analysis.

[0030] According to a particular embodiment of the invention, the tracking method also has the following characteristics: The method includes administering a treatment to the individual and assessing the effectiveness of the treatment from the estimated fitness state.

[0031] Finally, the object of the present invention, according to the fifth, sixth and seventh aspects of the present invention, respectively, is as follows: An apparatus comprising: an inertial device for acquiring displacement parameters of a measurement point during at least one displacement interval; and a data processing unit configured to estimate, from the displacement parameters acquired by the inertial device, at least one of a change in velocity of the measurement point during the displacement interval, a trajectory of a point of interest attached to the measurement point during the displacement interval, and a force exerted by a limb of an individual to which the point of interest belongs, by carrying out a method according to any of the first, second and third aspects. A computer program product comprising code instructions for carrying out a method according to any of the first, second, third, fourth and fifth aspects when the program is executed by a computer. A storage medium readable by a computing device, having stored thereon a computer program product comprising code instructions for performing the method according to any one of the first, second and third aspects when the program is executed by a computer. [Brief explanation of the drawings]

[0032] Other characteristics and advantages of the invention will become apparent on reading the following description, given by way of example only and with reference to the accompanying drawings, in which: [Figure 1] 1 is a diagram of an apparatus in accordance with an exemplary embodiment of the present invention; [Figure 2] FIG. 2 is a diagram of the measuring device of the device of FIG. 1. [Figure 3] FIG. 2 is a diagram of a motion sensor of the device of FIG. 1. [Figure 4] FIG. 2 illustrates a force analysis method implemented by the apparatus of FIG. 1. [Figure 5] 5 shows in detail the steps of the method of FIG. 4. [Figure 6] FIG. 6 shows in detail the first sub-step of the step of FIG. 5. [Figure 7] FIG. 6 shows in detail the second substep of the step of FIG. 5. [Figure 8] FIG. 6 shows in detail the third substep of the step of FIG. 5; [Figure 9]FIG. 5 is a diagram of a biomechanical model of the upper limb used to implement the method of FIG. 4. [Figure 10] 10A and 10B show pairs of starting and ending configurations of the biomechanical model of FIG. 9. DETAILED DESCRIPTION OF THE INVENTION

[0033] 1 is for assessing and analyzing forces exerted by muscles of an upper limb 12 of an individual 14. Here and hereinafter, "individual" means a human being. To this end, the device 10 includes a measurement device 20 attached to a forearm 22 of the upper limb 12. In the illustrated example, the device 10 also includes a movement sensor 24 intended to capture the displacement of at least one lower limb of the individual 14.

[0034] It should be recalled here that, anatomically, an upper limb, such as upper limb 12, is composed of three parts: an arm 25 articulated to a torso 26 by a shoulder 27, a forearm 22 articulated to the arm 25 by an elbow 28, and a hand 29 articulated to the forearm 22 by a wrist 30. Here and hereinafter, these terms will be used in this sense, and the term "arm" will be used specifically to refer only to the portion of the upper limb between the elbow and the shoulder, rather than the entire upper limb as in common parlance.

[0035] The measuring device 20 is fixed to the forearm 22 of the upper limb 12, i.e., in the ground reference frame, it moves substantially the same as the forearm 12. The measuring device 20 is particularly located between the elbow 28 (exclusive) and the wrist 30 (inclusive), for example, as shown, substantially at the wrist 30.

[0036] 2 , the measurement device 20 includes a support 31 and an attachment element 32 for attaching the support 31 to the forearm 22. The measurement device 20 also includes an inertial device 33 and a data processing unit 34 attached to the support 31. In the illustrated example, the measurement device 20 also includes a communication system 36 (typically a wireless communication system) attached to the support 31 for communication between the measurement device 20 and a second measurement device 24 and / or an external device, for example a mobile terminal such as a smart phone (not shown) or a remote server (not shown). Optionally, the measurement device 20 further includes a storage module 38 attached to the support 31. The support 31 is typically constituted by a case.

[0037] The attachment element 32 here consists, for example, of a bracelet with a hook-and-loop strip (which encircles the forearm 22 and allows for an integral connection). Alternatively (not shown), the attachment element 32 consists of any element which allows for an integral connection with the forearm 22.

[0038] The inertial device 33 is a moving reference frame (R m ), i.e., a moving reference frame (R m ) and measures three components of an angular velocity vector within the angular velocity vector. Thus, it can be seen that gyrometer 40 typically comprises a set of three gyrometers associated with one of three axes, and in particular three axes (i.e., capable of measuring one of three components of an angular velocity vector).

[0039] Preferably, a moving reference frame (R m ) is chosen so that when the measurement device 20 is worn, one of its axes is colinear with the forearm 22. Here and below, the moving reference frame (R m ) is defined as the direct orthogonal reference frame formed by three axes, including: an axis U collinear with the forearm 22 and pointing towards the hand 29; Axis V perpendicular to axis U, and Axis W is perpendicular to axes U and V.

[0040] The inertial device 33 also includes an accelerometer 42 for measuring the acceleration of the measurement device 20 according to a system of three orthogonal axes defining a moving reference frame integral with the measurement device 20, which is connected to the moving reference frame (R m ) In other words, the accelerometers are capable of measuring three components of an acceleration vector in a moving reference frame. It will thus be understood that the accelerometer 42 typically comprises three axes, in particular a set of three accelerometers each associated with one of the three axes (i.e., each capable of measuring one of the three components of the acceleration vector). These accelerometers are sensitive to external forces, including gravity, applied to the inertial unit 33 and are capable of measuring a specific acceleration.

[0041] The data processing unit 34 is configured to estimate from the measurements of the inertial device 33 the orientation, linear velocity and position of a measurement point integral with the measuring device 20, e.g., relative to the moving reference frame (R m ) The data processing unit 34 is also configured to estimate the linear velocity and the position of the center of gravity of the forearm 22. The data processing unit 34 is further configured to evaluate and analyze the forces of the muscles of the upper limb 12.

[0042] For this purpose, the data processing unit 34 is constituted in the illustrated example by a programmable machine such as a DSP ("digital signal processor") or a microcontroller. The data processing unit 34 comprises a processor or CPU ("central processing unit") 44 and a memory 46 of the RAM ("random access memory") and / or ROM ("read only memory") type. The processor 44 is configured to execute instructions loaded into the memory 46. When the measuring device 20 is powered on, the processor 44 is able to read and execute instructions from the memory 46. These instructions form a computer program that causes the processor 44 to carry out certain steps of a method 1000 (FIG. 4), which will be described in more detail below.

[0043] Alternatively (not shown), the data processing unit 34 is constituted by a machine or dedicated component such as an FPGA ("Field Programmable Gate Array") or an ASIC ("Application Specific Integrated Circuit"). Furthermore, the data processing unit 34 also includes a buffer memory 48 for temporarily storing information required for the implementation of the method 1000 .

[0044] The communication system 36 may be configured to implement short-range wireless communication, such as Bluetooth or Wi-Fi (particularly as shown in the example of the motion sensor 24), and / or connect to a mobile network (typically UMTS / LTE / 5G) for long-range communication. Alternatively (not shown), the communication system 36 may be a wired connection (typically USB) for transferring data from the storage module 38 to another storage module (typically a server (not shown)).

[0045] For example, communication system 36 may be configured to receive displacement data of at least one lower limb of individual 14 from motion sensor 24 and / or transmit analyzed data to an external device for presentation to a human operator, typically by displaying it on a screen.

[0046] Optionally, the measurement device 20 also includes an array of magnetometers (not shown) connected to the support 31, i.e., each magnetometer moves substantially in the same way as the support 31 in the ground reference frame and is spatially separated from one another. Each magnetometer is a three-axis magnetometer capable of measuring magnetic fields along three axes. For this purpose, each magnetometer typically consists of three single-axis magnetometers (not shown) oriented along axes that are substantially orthogonal to one another. These axes are preferably the same as the axes of the three orthogonal axes of the gyrometer 40 and / or accelerometer 42 system.

[0047] The magnetometer array, due to its particular geometry, is able to measure, at each measurement time of the magnetometer, the spatial gradient of the measured magnetic field, in particular the spatial gradient of the moving reference frame (R m ) are suitable for determining the coefficients of this gradient along each of the axes U, V, W of the magnetic field. The coefficients of this gradient are determined, for example, by a method using vector measurements of the magnetic field performed by a magnetometer, by a method related to optimization methods of the least squares or median filter type, or by a method related to the intrinsic properties of the magnetic field described by Maxwell's equations. However, any other conventional method suitable for calculating the coefficients of the spatial gradient of a magnetic field is also suitable.

[0048] The data processing unit 34 is then configured to determine the linear velocity of the measurement point over a particular displacement phase, typically by implementing the method described in EP 2 541 199.

[0049] In the above example, the processing unit 34 of the measuring device 20 is mounted on the support 31. Alternatively (not shown), at least part of the data processing unit 34 is located remotely, for example in a mobile terminal (not shown) and / or a remote server (not shown). In other words, at least part of the steps of the method 1000 are performed by the mobile terminal and / or the remote server. The communication system 36 is then configured to transmit data from the inertial device 33, and optionally also data from the magnetometer array of the first measuring device 20, to the mobile terminal and / or the remote server.

[0050] Returning to FIG. 1, here the motion sensor 24 is secured to the lower limb 50 of the individual 14 . Here, anatomically, a lower limb, such as lower leg 50, is composed of three segments: a thigh 51 articulated to torso 26 by hip 52, a leg 53 articulated to thigh 51 by knee 54, and a foot 55 articulated to leg 53 by ankle 56. Here and hereinafter, these terms are used in this sense. In particular, the term "leg" is used to refer only to the segment of the upper limb between the knee and the ankle, rather than the entire lower limb, as is common parlance.

[0051] The motion sensor 24 is particularly fixed to the leg 53 of the lower leg 50, i.e., in the ground reference frame, it moves substantially the same as the movement of the leg 53. The motion sensor 24 is particularly located between the knee 54 (exclusive) and the ankle 56 (inclusive), for example, as shown in the figure, substantially at the ankle 56. Alternatively, the motion sensor 24 is fixed to the wheelchair 58 in which the individual 14 is seated.

[0052] 3 , the motion sensor 24 here includes a support 61 and an attachment element 62 for attaching the support 61 to the leg 53. The motion sensor 24 also includes a gyrometer 63 and a data processing unit 64 attached to the support 61. In the illustrated example, the motion sensor 24 further includes a communication system 66 (typically a wireless communication system) attached to the support 61 for communication between the motion sensor 24 and the measuring device 20 and / or an external device, for example a mobile terminal such as a multifunctional mobile phone (not shown) or a remote server (not shown). Optionally, the motion sensor 24 further includes a storage module 68 attached to the support 61. The support 61 is typically constituted by a case.

[0053] The attachment element 62 here consists, for example, of a bracelet with hook-and-loop strips, configured to surround the legs 53 and allow an integral connection. Alternatively (not shown), the attachment element 62 consists of any element that can be integrally connected to the legs 53.

[0054] The gyrometer 63 is capable of measuring the angular velocity of the motion sensor 24 according to a system of three orthogonal axes that define a moving reference frame integral with the motion sensor 24, i.e., is capable of measuring the three components of an angular velocity vector in the moving reference frame. It will thus be understood that the gyrometer 63 is typically constituted by a set of three gyrometers, in particular triaxial, each associated with one of the three axes (i.e., each capable of measuring one of the three components of an angular velocity vector).

[0055] The data processing unit 64 is configured to deduce moving and immobile phases of the motion sensor 24 from the measurements of the gyrometer 63. For this purpose, the data processing unit 64 is typically configured to compare the norm of the angular velocity measured by the gyrometer 63 with a threshold and to define phases for which this norm is smaller than the threshold as immobile phases, and phases for which this norm is smaller (larger) than the threshold as moving phases.

[0056] In the example shown, the data processing unit 64 is constituted by a programmable machine such as a DSP (digital signal processor) or a microcontroller. The data processing unit 64 includes a processor or CPU (central processing unit) 74 and a memory 76 of the RAM (random access memory) and / or ROM (read only memory) type. The processor 74 is configured to execute instructions loaded into the memory 76. When the motion sensor 24 is powered on, the processor 74 can read and execute instructions from the memory 76. These instructions form a computer program that causes the processor 74 to implement a method for detecting displacement of the motion sensor 24.

[0057] Alternatively (not shown), the data processing unit 64 is constituted by a machine or dedicated component such as an FPGA ("Field Programmable Gate Array") or an ASIC ("Application Specific Integrated Circuit"). The data processing unit 64 further includes a buffer memory 78 for temporarily storing information required to detect the displacement of the motion sensor 24 .

[0058] The communication system 66 is configured to perform short-range wireless communication, such as Bluetooth or Wi-Fi (particularly the example shown in measuring device 20), and / or to connect to a mobile network (typically UMTS / LTE / 5G) for long-range communication. Alternatively (not shown), the communication system 66 is a wired connection (typically USB), for example for transferring data from the storage module 68 to another storage module, typically a server (not shown).

[0059] For example, the communication system 66 is configured to transmit the start and end times of the immobility phase to the measurement device 20 and / or an external device, typically a mobile phone or a server.

[0060] A method 1000 performed by the device 10 will now be described with reference to FIGS. 4, the method 1000 begins with step 1002 of acquiring displacement parameters of a measurement point. During this step, the inertial device 33 acquires displacement parameters at a number of sampling times t k At each point, the acceleration at the measurement point

number

number

number

number

[0061] Step 1002 follows power-up of the measurement device 20 and typically continues until the measurement device 20 is shut down.

[0062] The method 1000 continues after step 1002, or more precisely after a sufficient number of sampling times t k Acceleration over

number

number

[0063] For example, step 1004 involves the processing unit 34 detecting an immobility phase of the measuring device 20 and subsequently measuring the acceleration during the immobility phase. The detection of the immobility phase is typically performed by measuring the rotational speed.

number

number

[0064] If this guess shows that axis U points in the direction of hand 29, then the moving reference frame (R m ) is maintained. Conversely, if axis U points towards elbow 28, then the moving reference frame (R m ) is reoriented (typically rotated 180° about axis V or axis W) so that axis U points toward hand 29.

[0065] In parallel with steps 1002 and 1004, method 1000 also includes a step 1006 of detecting immobility phases of the lower limbs of individual 14, followed by a step 1008 of transmitting the start and end times of these immobility phases to measurement device 20.

[0066] In step 1006, typically, the processing unit 64 of the motion sensor 24 compares the rotational speed criterion measured by the gyrometer 63 to a threshold and determines periods when the criterion is less than the threshold as immobility phases.

[0067] Next, in step 1008, the motion sensor 24 transmits the thus detected start and end times of the immobility phase to the measurement device 20 via the communication systems 66, 36.

[0068] Steps 1004 and 1008 are followed by step 1100 of selecting a period of interest for the acquired parameters. During this step 1100, the processing unit 34 of the measurement device 20 identifies the periods between the start and end times of the immobility phase of the lower limbs as forming the period of interest for the acquired parameters and selects these periods.

number

number

[0069] Next, for each period of interest, method 1000 includes a step 1200 of determining the linear velocity of the measurement point at various times in the period of interest. During this step 1200, processing unit 34 determines the linear velocity of the measurement point during a particular phase.

[0070] These phases are, for example, stationary phases of the measuring device 20 that the processing unit 34 detects and determines that the linear velocity of the measuring point during these phases is zero. These stationary phases are typically determined by the rotational speed norm

number

number

[0071] Preferably, these fields also include immobility phases of elbow 28, and processing unit 34 detects an immobility phase when the movement of the measurement point is a pure rotation with a radius equal to the distance between the measurement point and elbow 28. The linear velocity of the measurement point can then be easily determined by methods known to those skilled in the art.

[0072] Advantageously, these phases also include phases in which the linear velocity of the measurement point is determined by data other than that provided by the inertial device 33. These other (other) data consist, for example, of magnetic field gradient measurement data provided by a magnetometer array. The phases in which the linear velocity of the measurement point is determined include steady-state phases of the surrounding magnetic field, and the processing unit 34 implements the method described in WO2019 / 016474 in order to determine the linear velocity of the measurement point during these phases.

[0073] The phases in which the linear velocities of the measurement points are determined are spaced apart by time intervals in which the linear velocities of the measurement points are not known a priori. For each of these intervals, the method 1000 includes a step 1300 of estimating the forces exerted by the muscles of the upper limb during this interval. This step 1300 is performed by the processing unit 34.

[0074] 5, step 1300 includes a first sub-step 1310 of estimating the change in orientation of the measurement point during the interval. Here and below, the "orientation of the measurement point" refers to the change in orientation of the measurement point relative to the ground inertial reference frame (R) whose axes are aligned with the vertical. i ) in a moving reference frame (R m ) and hence the angular velocity

number

[0075] The orientation of a measurement point can be expressed as a rotation matrix (denoted by R), an orientation quaternion (denoted by Q), or Euler angles (roll φ, pitch θ, yaw ψ). These three notations are equivalent and are therefore used interchangeably in this document.

[0076] In particular, the inertial reference frame (R i ) to the moving reference frame (R m ) from Euler angles is known to be expressed as follows:

number

[0077] Referring to FIG. 6, substep 1310 begins with determining the initial time t i The method includes a first sub-step 1312 of determining an initial orientation of the measurement point in

[0078] In this substep 1312, the processing unit 34 first determines the starting orientation of the measurement point at a starting time t0, at which the forearm 22, and therefore the measurement device 20, are considered immobile. This determination is typically performed from acceleration measurements,

number

number

number

number

number

number

number

[0079] Note that this equation is not completely deterministic and there are degrees of freedom remaining in some components of the rotation matrix R(t0). This is because the yaw angle ψ(t0) at the start time t0 is not very important for the rest of the calculations, and therefore the axis X can be set arbitrarily. For example, the axis X is chosen so that the yaw angle ψ(t0) at the start time t0 is zero. That is, the rotation matrix at the start time t0 is chosen so that it solves the following additional equation:

number

number

number

number

number

number

[0080] The start time t0 is the initial time t i In that case, the initial orientation is equal to the starting orientation. Alternatively, the starting time t0 is equal to the initial time t i and the initial orientation is then estimated from the starting orientation by forward integration of the measurements provided by gyrometer 40. By "forward integration" here and below it is meant that the integration is performed in positive time and the initial value is chosen in the past.

[0081] Sub-step 1312 is followed by sub-step 1313 of performing a forward integration of the measured orientation based on the initial orientation.

[0082] During this substep 1313, the processing unit 34 k The rotation speed measurement values ​​up to time t are integrated. i The initial orientation at each sampling time t k A first estimate of the orientation of the measurement point at R f Generate.

[0083] Preferably, this determination involves the implementation of a linear (Luenberger filter, Kalman filter, etc.) or non-linear (Extended Kalman filter, Invariant Observer, etc.) state estimation filter. Although an Extended Kalman filter implementation is described herein, one skilled in the art will be able to substitute other filters.

[0084] The implementation of the extended Kalman filter first involves a prediction step during which processing unit 34 uses the following equation:

number

number

number

[0085] Matrix estimate R after n sample steps f of

number

number

number

number

[0086] Although this equation is presented in terms of the rotation matrix, an equivalent differential equation exists between the attitude quaternion and the angular velocity. In parallel, the processing unit 34 modifies the covariance matrix, which estimates the covariance between each state of the filter, by linearization of the differential equation.

[0087] The prediction step is followed by an update step. In this update step, the acceleration measured by the accelerometer 42

number

number

number

number

[0088] Furthermore, acceleration

number

number

number

[0089] Rotation matrix

number

number

number

[0090] The difference between the estimated gravity field and the actual gravity field of the Earth is expressed as:

number

[0091] Furthermore, by linearizing this equation, the Kalman gain can be calculated, resulting in updates to the state and covariance matrices. The covariance matrix update is based on the assumption that the sensor and approximation errors are modeled as Gaussian distributed noise. The variance is estimated from measuring the sensor noise at rest and from assumptions about the movement of the forearm 22. Thus, at each sampling time t k A first estimate of the orientation of the measurement point at is obtained.

[0092] After substep 1313, the final time t f This is followed by sub-step 1314 of determining the final orientation of the measurement point in .

[0093] In this sub-step 1314, the processing unit 34 first determines a reference time t during which the forearm 22, and therefore the measurement device 20, is considered to be immobile. r This determination is typically performed from acceleration measurements, where the measured acceleration γ(t r ) is equal to the opposite of the gravitational field:

number

number

number

number

number

number

[0094] This formula allows us to calculate the reference time t r The roll and pitch at the point can be obtained by the following formula:

number

number

number

number

[0095] Yaw angle ψ(t r ) is the time from the start time t0 to the reference time t r The yaw angle is set equal to the yaw angle obtained by forward integration of the rotation rate measurement up to

[0096] Preferably, the reference time t r is the final time t f and the final orientation is equal to the reference orientation. Alternatively, at the reference time t r is the final time t fand the final orientation is then estimated from the reference orientation by backward integration of the measurements provided by gyrometer 40. By "backward integration" we mean here and below that the integration is performed in negative time and the initial value is chosen in the future.

[0097] Sub-step 1314 is followed by sub-step 1315 of backward integrating the measured orientation based on the final orientation.

[0098] During this substep 1315, processing unit 34 k The subsequent rotation speed measurements are integrated and the time t f The final orientation in is used as the initial value, and at each sampling time t k A second estimate of the orientation of the measurement point at R b Generate.

[0099] Preferably, this determination involves implementing a linear state estimation filter (e.g., Luenberger filter, Kalman filter, etc.) or a non-linear (e.g., extended Kalman filter, invariant observer, etc.) state estimation filter. Although an extended Kalman filter implementation is described herein, one skilled in the art could substitute other filters.

[0100] This implementation is substantially the same as the implementation of the forward integration step 1313 described above, with the following differences. The formula used in the prediction step is:

number

number

number

[0101] Essentially, the differences can be summarized as follows: forward integration uses a priori estimates for the prediction step of the extended Kalman filter, whereas backward integration uses a posteriori estimates. Thus, at each sampling time t k A second estimate of the orientation of the measurement point at is obtained.

[0102] Sub-step 1310 ends with sub-step 1316 of merging the backward and forward integrals, during which each sampling time t k The final estimate of the orientation of the measurement point at is calculated by linear spherical interpolation of the first and second estimates:

number

number

[0103] Returning to FIG. 5, step 1300 includes, in parallel with step 1310, step 1320 of estimating the linear velocity of the measurement point during the interval.

[0104] Referring to FIG. 7, this sub-step 1320 includes a first sub-step 1322 of forward integrating a displacement parameter based on the linear velocity at the start of the interval.

[0105] During this substep 1322, processing unit 34 k The acceleration and rotational speed measurements from time t are integrated. i The initial velocity is set as the initial value at each sampling time t k A first estimate of the linear velocity V of the measurement point at f This initial velocity is determined in step 1200 and is equal to the linear velocity of the measurement point at the end of the immediately preceding phase of the interval.

[0106] A first estimate of the linear velocity of the measurement point is calculated for each sampling time t, starting with the oldest, using the following formula: k is iteratively estimated by:

number

number

number

number

number

number

number

number

number

[0107] Sub-step 1322 is followed by sub-step 1324 of back-integrating the displacement parameters based on the linear velocity at the end of the interval.

[0108] During this substep 1324, processing unit 34 k The subsequent acceleration and rotational speed measurements are integrated and calculated over time t f The final velocity at each sampling time t k A second estimate of the linear velocity V at the measurement point b This final velocity is equal to the linear velocity at the measurement point at the start of the phase immediately following the interval whose linear velocity was determined in step 1200.

[0109] A second estimate of the linear velocity of the measurement point is calculated for each sampling time t, starting with the most recent one. kis iteratively estimated using the following formula:

number

number

number

number

number

number

number

number

number

[0110] Sub-step 1320 ends with sub-step 1326 of merging the backward and forward integrals, during which each sampling time t k The final estimate of the linear velocity of the measurement point at is calculated by linear interpolation of the first and second estimates:

number

number

number

number

[0111] Returning to FIG. 5, sub-steps 1310 and 1320 are followed by sub-step 1330 of defining a computational inertial reference frame.

[0112] This computational inertial reference frame is the inertial reference frame (R) used to estimate, for example, orientation and linear velocity. i ) Alternatively, the computational inertial reference frame is another inertial reference frame, preferably having the following reference frame: The first axis Z coincides with the vertical line, A second horizontal axis X, which is collinear with the forearm 22 when the axis is in the most horizontal orientation of the interval, is located in the moving reference frame (R m a second horizontal axis X, oriented in the direction of the axis of A third horizontal axis Y which together with the first and second axes forms a directly orthogonal reference frame X, Y, Z.

[0113] Next, the direction R(t k ) and previously estimated speeds

number

[0114] Sub-step 1330 is followed by sub-step 1340 of calculating candidate trajectories for the center of mass of the forearm 22. The calculations are based on a biomechanical model 100 of the upper limb 12 shown in FIG.

[0115] As can be seen in this figure, the biomechanical model 100 includes a first segment 102 having a free end 104. The biomechanical model 100 also includes a second segment 106 articulated by a first one of its ends 108, 110 to a proximal end 114 opposite the free end 104 of the first segment 102 via a first ball joint 112, and by its second end 110 to a fixed point via a second ball joint 1116.

[0116] The first segment 102 models the forearm 22 of the upper limb 12. It is constituted by a cylinder with radius r and length l equal to the distance between the measuring device 20 and the elbow 28. It forms with a vertical line Z a first primary angle α that models the angle of the forearm 22 with the vertical, and with a plane (X, Z) a first secondary angle ζ (not shown) that models the angle of the forearm 22 with said plane.

[0117] The second segment 106 models the arm 25 of the upper limb 12. It is constituted by a cylinder with a length 1 equal to the length of the arm 25. It forms a vertical line and a second primary angle β that models the angle of the arm 25 with the vertical line, and a plane (X, Z) and a second secondary angle η (not shown) that models the angle of the arm 25 with said plane.

[0118] Each segment 102, 106 is rigid, i.e., it cannot bend. Preferably, at least one of the segments 102, 106 is extensible. If both segments 102, 106 are extensible, their extensibility is advantageously constrained to be equal to one another.

[0119] Center of gravity C of forearm 22 G is positioned on the first segment 102 at a distance of approximately half the length of the forearm 22 from the joint 112.

[0120] Thus, biomechanical model 100 has a number of possible configurations defined by the values ​​of the first primary angle α and secondary angle ζ, and the values ​​of the second primary angle β and secondary angle η, respectively.

[0121] In sub-step 1340, a candidate trajectory is calculated for each of a number of pairs of starting and ending configurations 130, 132 of the biomechanical model 100 shown in Figure 10, modeling the configuration of the upper limb 12 at the beginning and end of the interval, respectively. By hypothesis, in each of these configurations, the first secondary angle ζ is defined to be equal to the orientation ψ of the measurement point, and the second secondary angle η is set equal to the first secondary angle ζ. Thus, each of the starting and ending configurations of the biomechanical model 100 is directly defined by the first primary angle α and the second primary angle β.

[0122] Referring to FIG. 8, sub-step 1340 thus determines the starting value α of the first primary angle α for the purpose of defining a starting configuration and an ending configuration. i (FIG. 10) and the end value α of the first primary angle α. f (FIG. 10) starts with sub-step 1344 of estimating.

[0123] In sub-step 1342, a first primary starting angle α i is calculated at the initial time t using the following formula: i is estimated from the orientation of the measurement points at:

number

number

number

number

number

number

[0124] In sub-step 1344, a first primary end angle α if is calculated at the final time t using the following formula: f is estimated from the orientation of the measurement points at:

number

number

number

number

number

number

[0125] These sub-steps 1342, 1344 are shown as being performed sequentially. Alternatively, the sub-steps may be performed in parallel with one another.

[0126] Sub-steps 1342, 1344 are followed by sub-step 1346, which calculates the displacement of elbow 28. Alternatively, sub-steps 1342, 1344 are performed in parallel with sub-step 1346.

[0127] In sub-step 1346, processing unit 34 calculates an initial time t i and the final time t f Displacement of elbow 28 between

number

number

number

number

number

number

[0128] Substep 1340, after substep 1342, determines the starting value α of the first primary angle α. i The method also includes a sub-step 1348 of determining a range of starting values ​​for the second primary angle β that are compatible with the above.

[0129] This sub-step 1348 preferably includes: Displacement norm

number

number

number

[0130] In that case, there are various scenarios. Displacement norm

number

number

number

number

[0131] Alternatively, sub-step 1348 may include only one of these comparisons, with the range [0,α i ] or [0,α i +δ] depends on the result of this single comparison. Alternatively, sub-step 1348 does not involve a comparison, and the range of suitable starting values ​​is the range [0,α i ] is directly determined to be equal to

[0132] Sub-step 1348 is followed by sub-step 1350 of building a starting configuration of the biomechanical model 100 .

[0133] This substep 1350 selects a starting value β of the second primary angle β within a range of compatible starting values. i The step 1352 includes selecting

[0134] To build a starting configuration of the biomechanical model 100, a first primary angle α is assigned a starting value α i is assigned, and the second primary angle β is assigned a starting value β i is assigned.

[0135] Sub-step 1350 is followed by sub-step 1354 of forward integrating the estimated linear velocity V based on the starting configuration.

[0136] During this substep 1354, processing unit 34 k The linear velocity V is estimated by integrating the linear velocity V estimates from the previous measurement points up to time t. i The initial position P of the measurement point i is the initial value, and at each sampling time t k A first estimate P of the position of the measurement point at f (t k ) whose initial position is given by:

number

number

number

number

[0137] A first estimate of the location of the measurement point is calculated at each sampling time t using the following formula: k is estimated at:

number

[0138] In parallel with this sub-step 1354, sub-step 1340 determines the starting value β of the angle β. i and / or the end value α of the first primary angle α f The method includes a sub-step 1355 of determining a range of end values ​​of the second primary angle β that are compatible with the above.

[0139] Preferably, this sub-step 1355 includes: Displacement norm

number

number

number

[0140] In that case, there are various scenarios: Displacement norm

number

number

number

number

number

[0141] Alternatively, sub-step 1355 may include only one of these comparisons, with angle βt Angle interval or range [0,α i +ε] depends on the result of this single comparison. Alternatively, sub-step 1355 does not involve a comparison, and the range of suitable starting values ​​is determined by the angle β t is directly determined to be equal to the angular interval centered on

[0142] Sub-step 1355 is followed by sub-step 1356 of constructing a final configuration of the biomechanical model 100 .

[0143] This substep 1356 determines the end value β of the second primary angle β within the range of compatible end values. f The step 1357 includes selecting

[0144] To construct the final configuration of the biomechanical model 100, the final value α f is assigned to the first primary angle α and the end value β f is assigned to the second primary angle β.

[0145] Sub-step 1356 is followed by sub-step 1358 of back-integrating the estimated linear velocity V based on the ending configuration.

[0146] During this substep 1358, processing unit 34 determines the final time t f The final position P of the measurement point at f is the initial value, and at time t k At each sampling time t in the interval, the linear velocity V is estimated by integrating the linear velocity V estimates for subsequent measurement points. k A second estimate P of the position of the measurement point at b (t k ) whose initial position is given by:

number

number

number

number

[0147] A second estimate of the location of the measurement point is calculated at each sampling time t using the following equation: k is estimated at:

number

[0148] Sub-step 1340 ends with sub-step 1359 of merging the backward and forward integrals, during which each sampling time t k The final estimate of the position of the measurement point at is calculated by linear interpolation of the first and second estimates:

number

number

[0149] From this final estimate of the trajectory of the measurement points, the processing unit 34 calculates the center of gravity C G This trajectory is calculated from the final estimate of the measurement point position by the following formula: k It consists of a set of candidate locations in:

number

number

number

number

number

[0150] Substeps 1356, 1358, and 1359 determine the end value β that falls within the range of fitness values ​​determined in substep 1355. f and sub-steps 1350, 1354, 1355, 1356, 1358, and 1359 are repeated for each starting value β that falls within the range of fit values ​​determined in sub-step 1348. iIt should be noted that "each value in the range" refers to a set of discrete values ​​spaced apart from one another by a predetermined step (e.g., substantially equal to 0.1°).

[0151] Thus, each starting value β within the range of fit values ​​determined in substep 1348 i and each end value β included in the range of fitness values ​​determined in substep 1355 f For the center of gravity C G There is a pair of starting and ending configurations for which candidate trajectories are calculated.

[0152] Returning to FIG. 5, sub-step 1340 is followed by sub-step 1360 of evaluating the compatibility of each candidate trajectory with the biomechanical model 100 .

[0153] During this sub-step 1360, processing unit 34 evaluates the compatibility of each candidate trajectory with biomechanical model 100. This compatibility is typically evaluated by a cost function that depends on the deformation imposed on biomechanical model 100 by tracking the candidate trajectory. Typically, the compatibility is inversely proportional to the result of the cost function. For example, if segment 106 (corresponding to arm 25) is extensible, the cost function may be the difference between the maximum extension of segment 106 and length L, and the compatibility is inversely proportional to said difference. Alternatively, the cost function is configured such that the compatibility increases depending on the result of the cost function.

[0154] Sub-step 1360 is followed by sub-step 1365 of selecting the best-fit candidate trajectory.

[0155] During this sub-step 1365, processing unit 34 selects the candidate trajectory whose cost function result reflects the best fitness, i.e., whose result is either lowest (if fitness is inversely proportional to the cost function result) or highest (if fitness increases with the cost function result).

[0156] Sub-step 1365 is followed by sub-step 1370 of verifying the acceptability of the suitability of the selected trajectory.

[0157] In this sub-step 1370, processing unit 34 compares the result of the cost function of the selected trajectory with a predetermined threshold. The threshold is selected such that a comparison of the result of the cost function of the selected trajectory with the threshold reflects the acceptable or unacceptable nature of the fitness of the selected trajectory. Typically, the threshold is selected such that the fitness of the selected trajectory is considered acceptable only if the result of the cost function is below the threshold (if fitness is inversely proportional to the cost function result) or above the threshold (if fitness increases with the cost function result).

[0158] For each interval (hereinafter referred to as a "primary interval") in which the suitability of the selected trajectory is acceptable, substep 1370 is followed by substep 1372 of validating the selected trajectory. Thus, the selected trajectory is determined by the center of gravity C G construct a validated estimate of the trajectory of

[0159] Sub-step 1372 is followed by sub-step 1374 of estimating the rotational component of the kinetic energy of the forearm 22 during the primary interval.

[0160] During this substep 1374, processing unit 34 derives from the rotational speed measurements during the primary interval the rotational component E of the mass kinetic energy acquired by forearm 22 during the primary interval. k,r This rotation component is typically given by:

number

number

number

number

[0161] After substep 1374, the primary force W exerted by upper limb 12 during the primary interval is p This is followed by substep 1376 of evaluating

[0162] During this step, processing unit 34 calculates the primary force W from the estimated trajectory and change in linear velocity, typically using the following equation: p Rate:

number

number

number

number

number

number

number

[0163] After substep 1376, the primary force W of kinetic energy acquired by forearm 22 during each primary interval is p and the rotation component E k,r This is followed by substep 1378 of determining the relationship f between the linear force W and the linear force W. This relationship f is typically determined by multiplying all pairs (linear forces W) taken at various linear intervals. p , the rotational component of kinetic energy E k,r ) is determined by regression, preferably linear regression.

[0164] For each interval (hereinafter referred to as a "secondary interval") in which the suitability of the selected trajectory is not acceptable, sub-step 1370 is followed by sub-step 1382 of rejecting the candidate trajectory. None of the candidate trajectories has a centroid C in the secondary interval. G are not considered valid estimates of the orbit of the

[0165] Sub-step 1382 is followed by sub-step 1384 of estimating the rotational component of the kinetic energy of the forearm 22 during the secondary interval.

[0166] During this substep 1384, processing unit 34 derives from the measurements of rotational speed during the secondary interval the rotational component E of the kinetic energy acquired by forearm 22 during the secondary interval. k,r This rotation component is typically obtained by the formula already explained in step 1374.

[0167] After substep 1384, the secondary force W exerted by upper limb 12 during the secondary interval is s This is followed by substep 1386 of evaluating

[0168] During this step, the processing unit 34 calculates the rotational component E of the kinetic energy acquired by the forearm 22 during the secondary interval. k,r and the primary force W of kinetic energy acquired by the forearm 22 determined in sub-step 1378 p and the rotation component E k,r From the relationship between f and the primary force W s Typically, the processing unit 34 evaluates the primary force W using the following equation: s Rate:

number

[0169] Step 1300 is repeated for each time interval of the period of interest where the linear velocity of the measurement point is not known a priori and falls between the two phases for which the linear velocity of the measurement point was determined in step 1200 .

[0170] Steps 1200 and 1300 are repeated for each period of interest included in the predetermined period, which is typically greater than one week and preferably less than three months in duration.

[0171] After this repetition of steps 1200, 1300, the method 1000 calculates the primary force W evaluated for the various time intervals included in the predetermined time period. p and / or secondary force W s The method includes a step 1400 of determining the value of a statistic representative of the morphological state of the individual 14 over the predetermined period, calculated based on the set formed by

[0172] During this step 1400, the processing unit 34 calculates the primary force W evaluated for the various time intervals included in the predetermined period. p and / or secondary force W s Statistical calculations are performed on the set formed by (a) and (b). Typically, processing unit 34 calculates the mean and / or percentiles of these forces for all forces assessed over a given time period. For example, processing unit 34 calculates the 50th percentile, which corresponds to the median, 80th, 95th, 99th, or other percentile value of the forces assessed over a given time period.

[0173] Preferably, processing unit 34 selects statistics from a predetermined percentile range, particularly the high percentile range, i.e., percentiles greater than the 70th percentile. Indeed, these high percentiles reflect the individual's fastest and / or longest movements, and therefore represent the maximum force and maximum muscle strength that individual 14 can exert, with the speed and duration of upper limb movements being limited by muscle strength status. Thus, these high percentiles are particularly sensitive to the fitness state of individual 14.

[0174] Advantageously, the statistic comprises the 99th percentile of the forces assessed over a given period of time. Optionally, step 1400 includes selecting at least one other statistic representative of the fitness state of the individual 14 over a predetermined period of time.

[0175] Steps 1002-1400 are repeated over a predetermined period of time, so that processing unit 34 generates a set of statistical values ​​representative of the fitness state of individual 14 over the predetermined period of time.

[0176] Following this repetition of steps 1002-1400, the method 1000 includes a step 1500 of observing the change in the statistics over different predetermined time periods. In this way, the fitness state of the individual 14 can be tracked.

[0177] Advantageously, this tracking of the fitness level of the individual 14 is used to determine the effectiveness of treatment administered to the individual 14 .

[0178] By virtue of the above exemplary embodiments, the physical fitness of individuals who lack the use of their lower limbs can be tracked over time with little or no restriction on the individual, particularly over long periods in uncontrolled environments, and can closely track the individual's actual fitness level.

[0179] The exemplary embodiment also allows for good estimation of the physical forces exerted by an individual's upper limbs for various movements over long periods of time in an uncontrolled environment. The exemplary embodiment can further improve the quality of linear velocity and position estimates inferred from inertial device measurements, especially when the inertial device is attached to the individual's limbs.

[0180] While the above description focuses on an exemplary embodiment in which the measurement device 20 is attached to the upper limb 12 of an individual 14 and the force being assessed is the force of this upper limb 12, the present invention also applies to methods for assessing the force of a lower limb of an individual, in which case the measurement device 20 is attached to the leg of this lower limb. The present invention also applies to methods for assessing the force of an animal's forelimbs or hind limbs. Those skilled in the art will be able to readily make any necessary modifications.

Claims

1. 1. A method for estimating a trajectory over a displacement interval of a point of interest belonging to a limb (12) of an individual (14), the method being executed by a data processing unit (34), the method comprising the steps of: estimating (1320) the change in velocity of a measurement point attached to said point of interest during said displacement interval; estimating (1310) a change in orientation of the measurement point during the displacement interval; and estimating the trajectory of the point of interest during the displacement interval from the estimated velocity and orientation of the measurement points. method.

2. 2. The method of claim 1, wherein the individual (14) is a human, the limb (12) is an upper limb of the individual (14), and the point of interest is preferably a point on a forearm (22) of the upper limb.

3. The step of estimating the trajectory includes: calculating (1340) several candidate trajectories by applying the estimated velocity and orientation to different pairs of starting and ending configurations (130, 132) of a biomechanical model (100) of the limb (12); assessing 1360 the compatibility of each candidate trajectory with the biomechanical model; and selecting (1365) the candidate trajectory that is most suitable.

4. The points of interest are points on the forearm (22) belonging to the upper limb of the individual (14), and each starting configuration (130) of the biomechanical model (100) of the limb (12) has a first starting angle (α ｉ ) and a second starting angle (β ｉ ), and each end configuration (132) of the biomechanical model (100) of the limb (12) is defined by a pair of start angles formed by a first end angle (α) between the forearm (22) and the vertical line. ｆ ) and a second end angle (β ｆ 4. The method of claim 3, wherein the angle is defined by a pair of end angles formed by

5. For each starting configuration (130) or each ending configuration (132) of the biomechanical model (100) of the limb (12), The first starting angle (α ｉ ), or the end angle (α ｆ ) is estimated from the orientation of the measurement point at the start or end of the displacement interval; The second starting angle (β ｉ ), or the end angle (β ｆ ) is the first starting angle (α ｉ ), or the end angle (α ｆ 5. The method of claim 4, wherein the .lambda.

6. The first starting angle (α ｉ ) is smaller than the second starting angle (β ｉ ) and each value of the second end angle (β ｆ 6. The method of claim 4, wherein for each of the first and second configurations, there exists a pair of starting and ending configurations (130, 132) for which candidate trajectories are calculated.

7. The predetermined interval is between 0° and the first end angle (α ｆ ) the interval, or angle β ｔ and the following relationship holds: [Equation 1] where: L is the length of the arm (25) of the upper limb, β ｉ is the second starting angle (β) in the starting configuration (130) that belongs to the same pair as the ending configuration (132). ｉ ) and t ｉ is the start time of the displacement interval, t ｆ is the end time of the displacement interval, [Equation 2] is the velocity vector of the measurement point at each time t in the displacement interval, l is the distance between the elbow (28) of the upper limb and the measurement point; [Equation 3] is a unit vector indicating the orientation of the major axis of the forearm (22) at the end of the displacement interval; [Equation 4] is a unit vector indicating the orientation of the major axis of the forearm (22) at the start of the displacement interval; [Equation 5] is a unit vector indicating the vertical direction, Vector [Equation 6] The method of claim 6 , wherein:

8. The step of calculating each candidate trajectory (1340) includes: a step (1354) of performing a forward integration of the estimated velocity using the position of the measurement point in the starting configuration (130) as an initial position; a step (1358) of performing backward integration of the estimated velocity using the position of the measurement point in the final configuration (132) as an initial position; and calculating (1359) the candidate trajectory by merging the backward integral with the forward integral.

9. The step of estimating the change in velocity of the measurement point during the displacement interval comprises: determining (1200) an initial velocity and a final velocity of the measurement point at the beginning and end of the displacement interval, respectively; and estimating (1320) the change in velocity of the measurement point during the displacement interval, the estimation comprising: performing a forward integration (1322) of displacement parameters of the measurement points acquired by the inertial device (32) during the displacement interval, taking the value of the initial velocity as the initial value of the velocity; performing a backward integration (1324) of the displacement parameter, taking the value of the final velocity as the initial value of the velocity; and calculating (1326) the estimated velocity by merging the backward integral and the forward integral.

10. 1. A method (1300) for assessing a force exerted by a limb (12) of an individual (14), the method being performed by a data processing unit (34), the method comprising: - estimating a trajectory of a point of interest belonging to said limb (12) over at least one primary displacement interval by a method according to any one of claims 1 to 9; and estimating (1376) a primary force exerted by the limb (12) during the or each primary displacement interval from the estimated trajectory and change in velocity. method.

11. estimating (1374) from the displacement parameters acquired during the primary displacement interval a rotational component of primary kinetic energy acquired by at least a portion of the limb (12) during the primary displacement interval, wherein a primary force is a function of the rotational component of the primary kinetic energy; determining (1378) a relationship between the primary force and the rotational component of the primary kinetic energy; - estimating (1310) a change in orientation of a measurement point attached to said point of interest during at least one secondary displacement interval; estimating a rotational component of secondary kinetic energy acquired by the at least part of the limb (12) during a secondary displacement interval from displacement parameters acquired by an inertial device (32) during the secondary displacement interval; 11. The method (1300) of claim 10, further comprising the step (12) of evaluating a secondary force exerted by the limb during the or each secondary displacement interval from the determined relationship and the rotational component of the secondary kinetic energy.

12. For each of the primary displacement interval and the secondary displacement interval, estimating (1320) a change in velocity of a measurement point attached to said point of interest during said primary displacement interval or said secondary displacement interval; calculating (1340) a plurality of candidate trajectories by applying the estimated velocity and orientation to different pairs of starting and ending configurations (130, 132) of the biomechanical model (100) of the limb (12); evaluating (1360) the compatibility of each candidate trajectory with the biomechanical model (100); and for each primary displacement interval or each secondary displacement interval, verifying (1370) the acceptability of said suitability of at least one candidate trajectory; 12. The method (1300) of claim 11, wherein the trajectory of the point of interest is estimated only if at least one candidate trajectory has an acceptable fit.

13. 1. A method (1000) for analyzing forces exerted by a limb (12) of an individual (14), the method being performed by a data processing unit (34), the method comprising: a) estimating the force exerted by the limb (12) for a number of intervals of displacement of a point of interest of the limb (12) occurring during a predetermined period of time by a method (1300) according to any one of claims 10 to 12; b) determining (1400) the value of at least one statistic representative of the fitness state of the individual (14), the statistic being calculated based on a set formed by the estimated forces; method.

14. 14. The method (1000) of claim 13, comprising the step (1500) of repeating steps a) and b) over a number of predetermined time periods and observing changes in the or each statistic over different predetermined time periods.

15. 15. A method for tracking the fitness status of an individual, comprising the steps of analyzing forces exerted by limbs of the individual by an analysis method according to claim 13 or 14, and inferring the fitness status of the individual from said analysis.

16. 16. The method of claim 15, comprising administering a treatment to the individual and assessing the effectiveness of the treatment from the inferred fitness state.

17. 13. An apparatus (10) comprising: an inertial device (32) for acquiring displacement parameters of a measurement point during at least one displacement interval; and a data processing unit (34) configured to estimate, from the displacement parameters acquired by the inertial device (32), at least one of a trajectory of a point of interest attached to the measurement point over the displacement interval and a force exerted by a limb (12) of an individual (14) to which the point of interest belongs, by implementing the method of any one of claims 1 to 12.

18. A computer program product comprising code instructions for carrying out the method of any one of claims 1 to 16 when executed by a computer.

Images