FR3058538B1 - MULTIPARAMETRIC METHOD FOR QUANTIFYING BALANCE - Google Patents

Abstract

The invention relates to a method for quantifying the equilibrium of an individual to obtain a value representative of the equilibrium of said individual, said method being implemented by a device comprising at least one data processing module connected to a device. storage means and said method comprising: a) a step of recording, on the storage means, at least one statokinesigram of the individual obtained from a platform comprising pressure and / or force sensors, b) a step of extraction, by the data processing module and from the statokinesigramme (s) of the individual recorded on the storage means, values of at least one relative trajectory parameter the position of the center of pressure and the values of at least one path parameter relating to the stability of the center of pressure, c) a step of determination, by the data processing module, of the value of more their quantifiers, from the values of the trajectory parameters extracted in step b), d) a step of comparing, by the data processing module, said values of several quantizers with the values of the same quantizers obtained from statokinesigrams of reference, and e) a step of determining, by the data processing module, said value representative of the equilibrium of the individual at the end of the comparison.

Description

MULTIPARAMETRIC METHOD FOR QUANTIFYING BALANCE

The present invention relates to the field of quantifying the equilibrium of an individual. The present invention more particularly relates to a multiparametric method for the quantification of the balance of a person, a device adapted to implement this method and a system incorporating said device. The present invention makes it possible in particular to monitor the evolution of this equilibrium, for example in the context of a re-education process, in a self-quantification approach or in the elderly, in particular so as to warn about risks of fall, in a process of home support. Previous Art [002] Self-measurement ("quantified self") is an increasingly common practice and includes more and more parameters. Similarly, devices on the market for quantifying different functions relating to self-measurement such as sleep or walking are increasing.

[003] Static and dynamic balances are essential components of our everyday movements and a lack of balance is a major cause of falling. In addition, the balance can be modified (i.e. reduced) as a result of life-related accidents such as lower extremity accidents but also related to the occurrence of mental disorders or strokes such as stroke. This postural instability linked to a disruption of the motor, sensory and / or cognitive functions can favor a fall which can be fatal or lead to serious decompensations.

[004] According to estimates by the World Health Organization in 2012, nearly 424,000 people worldwide die every year as a result of falls, placing falls in second place among the causes of accidental deaths in the world. the world. According to the Institut de Veille Sanitaire, there are 450,000 falls each year in France in people over 65 years of age, and this is the most common cause of death in the elderly with 4000 to 4500 cases per year in France. France.

[005] The fall of people is therefore a major public health problem because of its frequency and its medical and social consequences; especially and especially in the elderly. For example, the post-fall syndrome translates into a phobia of the fall with a loss of self-confidence to perform acts of daily life and leads to gratification. Thus, many cares aim at restoring the balance of the individuals and exercises aim to maintain it.

[006] Despite these stakes, there is currently no intuitive, reliable and inexpensive method or device for quantifying the equilibrium of an individual.

[007] Doctors today appreciate the balance via visual monitoring methods of the patient most often via standardized tests such as the Romberg test. The latter can help the doctor to diagnose and identify the possible causes of static ataxia. Nevertheless, such monitoring makes it possible to qualify and not to objectively quantify the balance of an individual. However, quantitative methods could make it possible to reinforce the objectivity, the homogeneity of the interpretations, to give the possibility to make comparisons of such tests (eg followed in the time or within a group of individuals) and to identify imperceptible behavior via visual tracking methods of the patient.

[008] Today, there are methods for evaluating the equilibrium generally based on the study of the displacement of the pressure center measured from a force platform. In particular, a standardization effort was carried out in the 80s and 90s with the establishment of "norms 85", including the definition of a protocol including the placement of individuals in a cabin with standardized dimensions, imposing a position of the feet so to ensure reproducibility of the parameters with an acquisition duration of 51.2 seconds at a sampling frequency of 5 Hz (Gagey et al 1988 - Statistical studies of the measurements made on the normal man by means of the plate standardized clinical stabilometry platform).

[009] In this context, the parameters usually determined from the pressure center displacement data, generally at a frequency of at least 10 Hz, make it possible to account for the subject's ability to maintain his orthostatic equilibrium and can be selected from among : - the length of the displacement, - the surface of the plot of the statokinesigram corresponding to the surface on which the center of pressure moves and is measured in mm2 by calculating the area of the ellipse of confidence (based on a curve of Gauss) which contains 90% of the pressure center support points, - the length as a function of the area (LFS) which corresponds to the ratio between the distance traveled by the center of pressure and the surface of the ellipse of confidence, - the variance of the velocity along the Y axis (VFY), or - the oscillation frequencies.

If posturography is the most suitable means in the evaluation of standing, its place in the assessment of balance disorders remains controversial. Several reasons make it difficult to use: cost of acquisition, achievement, reproducibility of exams, sensitivity, specificity, difficulties in interpreting results (large number of measurements to be compared between two states). Indeed, these methods are interested in the parameters of the displacement of the center of pressure independently and do not combine them. They do not allow to obtain, in the absence of a professional and by a simple and fast method, a value representative of the balance of the individual.

The force platform designated Wii made balance board (registered trademark) can track the movement of the center of pressure and also offers a measurement function of the "Wii fit age" including based on a measurement of the displacement of the center pressure. Numerous articles have focused on equilibrium assessment via this force platform, and although it provides similar pressure center displacement data to those obtained with force-based force platforms. Wheatone bridges (Clark et al, 2010 - Validity and Reliability of the Balance Scale), the measurement of balance by this type of device is not conclusive (Gobie et al 2014 - Using the Wii Fit as a tool for balance assessment and neurorehabilitation: the first hait decade of "Wii-search").

This lack of performance or at best this reduced performance is probably related to the mono factorial appreciation of the displacement of the pressure center and to random changes in the acquisition frequency over time that may lead to over-acquisition or under-acquisition. The few times that multiparametric methods are used, the latter are not able to produce satisfactory results because the data to be studied are very complex and difficult to model. Thus, the analytical methods previously used to evaluate the balance do not provide a unique and reliable value, easily measurable using simple and inexpensive device.

There is therefore a need for a device for quantifying the balance that can be performed routinely by anyone concerned to follow his balance and that low cost, but also by public authorities or health personnel.

Technical problem! The invention aims to overcome the disadvantages of the prior art. In particular, the object of the invention is to propose a method for quantifying the reliable equilibrium, that is to say, making it possible to establish a value representative of the state of the equilibrium of the individual, which is fast, simple, and does not necessarily require the intervention of a specialist in the field of posturology. Note that this procedure is not intended to replace the general practitioner or specialist and does not pose a diagnosis.

The invention also aims to propose a quantization device of the equilibrium that can be integrated into a complete system for quantifying the equilibrium.

To this end, the invention relates to a method of quantifying the equilibrium of an individual to obtain a value representative of the equilibrium of said individual, said method being implemented by a device comprising at least one data processing module connected to a storage means and said method comprising: a) a step of recording, on the storage means, at least one statokinesigram of the individual obtained from a platform comprising pressure and / or force sensors, b) an extraction step, by the data processing module and from the statokinesigramme (s) of the individual recorded on the means for memorizing values of at least one trajectory parameter relative to the position of the center of pressure and the values of at least one trajectory parameter relating to the stability of the center of pressure; e of determining, by the data processing module, the value of several quantizers, from the values of the trajectory parameters extracted in step b), d) a comparison step, by the data processing module, said values of several quantizers to the values of the same quantizers obtained from reference statokinesigrams, and e) a step of determination by the data processing module of said value representative of the equilibrium of the individual at the end of the comparison.

The implementation of this method includes the determination of several quantizers being then jointly processed in a step of comparing and determining a value representative of the equilibrium. Unlike the prior art, the method according to the invention comprises taking into account and combining trajectory parameters related to the position and the stability of the center of pressure so as to obtain a representative value of the equilibrium.

According to other optional features of the method: step b) further comprises extracting the values of at least one trajectory parameter relating to the dynamics of the center of pressure. Taking into account the values of a trajectory parameter related to the dynamics of the center of pressure makes it possible to improve the relevance of the representative value obtained as can be shown by the curves of the ROC curves (Receiver Operating Characteristic) and more particularly the area under the curve (AUC) data. at least one of the quantizers determined in step c) is selected from: the average value, the median value, the variance value, and an extreme value of said trajectory parameter. Alternatively, one of the quantizers determined in step c) is selected from: the average value, the median value, the variance value, the average value of the squares and an extreme value of the said trajectory parameter. step b) is carried out from two statokinesigrams generated during a Romberg test. The Romberg test is a test used by many posturology professionals for decades and it makes it possible to obtain a value representative of equilibrium in the context of the invention, in particular by comparing the values of the statokinesigrams obtained. eyes closed or eyes open. at least one of the quantizers determined in step c) is an extreme value of said trajectory parameter, said extreme value corresponding to a percentile between a percentile greater than or equal to 5 and a percentile less than or equal to 15, or a percentile between a percentile greater than or equal to 85 and between a percentile less than or equal to 95, values of the trajectory parameter. The analysis of extreme values, which is not generally found in current methods of analyzing the equilibrium, makes it possible to generate results that are particularly relevant in the context of the invention. steps d) and e) are performed by implementing the values of the quantizers determined in step c) in a previously calibrated notation algorithm on the basis of the values of the same quantizers obtained from the reference statokinesigrams. The use of a scoring algorithm incorporating steps d) and e) in particular makes it possible to shorten the duration of implementation of the method. - The previously calibrated scoring algorithm was obtained by implementing a statistical method of supervised learning, preferably a method of RANKING FOREST or RANDOM FOREST. The inventors have discovered that in the context of quantification of the balanced, the methods of RANKING FOREST or RANDOM FOREST are very effective. Preferably, the supervised learning statistical method is used in combination with a Bagging step. Thus, the previously calibrated scoring algorithm can be obtained by implementing a statistical supervised learning method comprising the implementation of a Bagging step. The method comprises in step c) the determination of the value of at least five quantifiers. The methods of the prior art are generally based on the one-to-one comparison of values extracted from the statokinesigrams whereas, in the context of the method according to the invention, the combination of at least five quantizers allows better performances.

The invention also relates to a device for quantifying the equilibrium of an individual, said device comprising: a communication module able to receive a statokinesigram of said individual, a storage means capable of recording the statokinesigram, and at least one data processing module able to connect to the storage means, said data processing module being configured to: extract, from a statokinesigram of said individual transmitted by the communication module, values of at least one trajectory parameter relative to the position of the center of pressure and the values of at least one trajectory parameter related to the stability of the center of pressure, o to determine several quantizers, from the values of the extracted trajectory parameters, o Compare the values of said quantifiers with the values of the same quantifiers obtained from referenced statokinesigrams This, and o Determine a representative value of the equilibrium of the individual on the basis of said comparison.

The invention further relates to a system for quantifying the equilibrium of an individual, comprising: a platform, said platform being adapted to receive an individual and comprising pressure and / or force sensors configured to generating raw data, at a first frequency, a function of a pressure exerted by the feet of the individual on the platform, - a raw data processing unit, arranged to obtain at least one statokinesigram of the individual from the raw data generated by the platform, and - a quantization device of the balance according to the invention, able to communicate with the processing unit.

[001] Other advantages and characteristics of the invention will appear on reading the following description given by way of illustrative and nonlimiting example, with reference to the appended figures which represent: • Figure 1, represents a diagram of the process quantization of the equilibrium according to the invention. FIG. 2 represents a statokinesigram according to the invention in which the axis y corresponds to the anteroposterior axis and the axis x to the medio-lateral axis. 3A and 3B represent a statokinesigram according to the invention, FIG. 3A corresponding to the projection on the X axis of all the values of the parameter "position of the center of pressure along the X axis" and FIG. the projection on the Y axis of all the values of the parameter "position of the center of pressure along the Y axis" • Figure 4 represents the set of values as a function of time (X axis) of the parameter "equilibrium temporal "calculated on the basis of the statokinesigram presented in FIG. 2. The dashed line representing a threshold value for the calculation of the ballistic interval parameter according to the invention. • Figure 5 shows the projection on the X axis of all the values of the parameter "position of the center of pressure along the X axis" and four quantifiers calculated from this parameter. FIG. 6 represents an exemplary implementation of the quantization method of the equilibrium according to the invention. The numbers arranged in oval shapes correspond to process steps whereas those arranged in the rectangles correspond to values obtained after the implementation of these steps. • Figure 7, represents a decision tree built from a Ranking Forest algorithm. F1 corresponds to the closed eyes / open eyes ratio for the 90th percentile of the ballistic interval, F2 is the ratio of eyes closed to eyes open for the variance of the position on the Y axis, and F3 is the variance of the temporal equilibrium of the eyes closed. FIG. 8 represents a device for quantifying the equilibrium according to the invention. FIG. 9 represents a system for quantifying the equilibrium according to the invention.

[002] In the remainder of the description, the "balance" in the sense of the invention corresponds to the postural equilibrium related to the stability of the body and more particularly to the stability of the center of gravity of the body. 'an individual. The concept of equilibrium according to the invention is related to the ability of an individual to avoid falling and includes static equilibrium and dynamic equilibrium.

[003] The "quantization of equilibrium" corresponds, in the sense of the invention, to the attribution of a value, for example a score, a classification or a note to a trajectory or a displacement of the pressure center of an individual. This quantization of the equilibrium makes it possible to obtain a value representative of the equilibrium and can be carried out on the basis of many scales of different sizes (e.g., 1.5, 10, 100), linear or otherwise. The representative value of the equilibrium attributed during the quantization of the equilibrium can also make it possible to assign an individual to a group for example via a decision rule. Quantification according to the invention can be achieved in particular by the implementation of a notation algorithm generated from a learning method.

[004] By "model" or "rule" or "algorithm of notation" it is necessary to understand within the meaning of the invention a finite sequence of operations or instructions making it possible to quantify the equilibrium, that is, to classify one or more individuals within previously defined groups Y, or to rank one or more individuals within a classification. The implementation of this finite sequence of operations makes it possible, for example, to assign a label Yo to an observation described by a set of characteristics Xo by, for example, the implementation of a function f that is capable of reproducing Y having observed X.

where e symbolizes noise or measurement error [005] For the purposes of the invention, the term "supervised learning method" means a method making it possible to define a function f from a base of n labeled observations.

[006] For the purposes of the invention, the term "center of pressure", the projection on the horizontal plane passing through the point of contact between the subject and the ground of the center of gravity of the vertical forces exerted on the ground by the body of the subject (Benda, BJet al 1994. Rehabilitation Engineering, IEEE Transactions on 1994, 2, 3-10, Biomechanical Relationship between Center of Gravity and Rehabilitation. These measurements can be made using a platform analyzing the pressure distribution under the sole such as a platform of force or a shoe or deformable soil. Without being an exact projection of the center of gravity, the center of pressure is strongly related to the center of gravity. The displacement of the center of pressure is generally faster and wider than that of the center of gravity in order to keep it in balance. It reflects the efforts made by an individual to control the position of his center of gravity.

[007] For the purposes of the invention, the term "center of gravity", the body center of gravity of an individual. It corresponds to the meaning of the invention at the center of mass of the individual. The center of gravity can not be maintained in perfect stability and by

example when standing upright the center of gravity oscillates back and forth and from left to right.

[008] The term "statokinesigram" or "trajectory of the center of pressure" means the data relating to the trajectory or displacement of the center of pressure. The statokinesigram can also be called a stabilogram and is generally generated via a platform such as a force platform, a "smart" ground equipped with sensors or soles equipped with pressure sensors. It corresponds to the calculated trajectory of the center of pressure in time. The trajectory of the center of pressure is defined by a set of position data in an orthonormal coordinate system x, y as a function of time and over a defined duration.

[009] The "platform" within the meaning of the invention corresponds to a device resting on the ground comprising sensors, for example of the type of force or pressure sensors, producing an electrical, optical or magnetic signal proportional to the force applied. on the said platform by the feet of an individual. The sensors used may, for example, be Wheatstone bridge-mounted strain gauges in order to generate the 3 force and moment components Fx, Fy, Fz, Mx, My and Mz; piezoelectric pressure sensors, piezoresistive pressure sensors or capacitive pressure sensors. In the sense of the invention, the platform is configured to generate "raw data" from said sensors.

By "parameter" and more particularly by "parameter calculated from the path of the center of pressure" is meant in the sense of the invention a transformation of the trajectory of the center of pressure into a set of values.

By "quantizer" and more particularly by "quantizer calculated from transformation of a parameter obtained" is meant in the sense of the invention a single value obtained by selection or transformation of all the values of a parameter.

By "reference quantizer" means a value obtained from a reference statokinesigram from a person whose balance has been previously qualified.

Within the meaning of the invention, the "Receiver Operating Characteristic (ROC) curve" represents the evolution of the sensitivity (rate of true positives) as a function of the specificity (false positive rate) of a model for each threshold value given. It is an increasing curve between the point (0,0) and the point (1,1) and in principle located above the first bisector. Indeed, a random prediction would give a line corresponding to the first bisector. For a ROC curve, the higher the curve is above the first bisector, the better the prediction and the area under the AUC (Area Under the Curve) curve gives an indicator of the quality of the model (1 for a ideal prediction, 0.5 for a random prediction).

In the following description, the same references are used to designate the same elements.

According to a first aspect, the invention relates to a method for quantifying the equilibrium of an individual to obtain a value representative of the balance of said individual. While, until now, the recognized methods are based on appreciations or visual observations and are not completely objective, the method according to the present invention has the advantage of generating a value also designated by a score, and more particularly an encrypted value. In addition, this value can be generated completely automatically, objectively and without the need to provide the process information on the individual in addition to one or more statokinesigramme (s).

This quantization method 1, shown in Figure 1, is implemented by a device at least one data processing module 220 connected to a storage means 280. Advantageously, the device may be the quantization device 2 according to the invention.

More particularly, as shown in FIG. 1, the quantization method 1 according to the invention comprises: a) a recording step 10, on the storage means 280, of at least one statokinesigram of the individual 110 obtained from a platform 310 comprising sensors 312 of pressure and / or force, b) an extraction step 20, by the data processing module 220 and from the statokinesigram of the individual 110 recorded on the storage means 280, values of at least one trajectory parameter relative to the position 121 of the center of pressure and the values of at least one trajectory parameter relative to the stability 122 of the center of pressure, c) a determination step 30, by the data processing module 220, of the value of several quantizers 130, from the values of the trajectory parameters 121, 122 extracted in step b), d) a comparison step 40, p ar the data processing module 220, said values of several quantizers 130 to the values of the same quantizers obtained from reference statokinesigrams 111, and e) a determination step 50, by the data processing module 220, of said value representative 160 of the equilibrium of the individual at the end of the comparison 40.

The quantization method 1 of the equilibrium according to the invention is based on the processing of the trajectory data of the pressure center of an individual as transcribed in a statokinesigram 110.

These trajectory data of the center of pressure of an individual generally correspond to the trajectory data of the center of pressure and are generally acquired when the individual is standing.

Preferably, the acquisition is done standing, for example the feet together or at the hips.

The influence of visual afference on stability is measured by generating two kinetics of displacement or trajectory of the center of pressure, a first open eyes and a second closed eyes. Thus, preferably, the extraction step is carried out from a statokinesigram 110 obtained while the individual with open eyes and a statokinesigram 110 obtained while the individual with the eyes closed.

In particular, the acquisition time of a statokinesigram 110 may be between 5 and 70 seconds, preferably between 20 and 60 seconds and even more preferably between 20 and 40 seconds.

Advantageously, the trajectory data of the center of pressure can be obtained during a Romberg test. Romberg's test is to place the individual standing motionless, arms along the body, looking straight ahead. The displacement of the center of pressure is recorded for 30 seconds. For example, a 30-second recording is done with the eyes open and another 30-second recording is done with the eyes closed.

Alternatively, it is possible to increase the sensitivity of the process by the use of a foam placed on a support and capable of deforming or disturbing the proprioceptive and tactile information. This foam may for example have a thickness of 1 to 10 millimeters and a density of between 100 and 500 kg / m3.

The method according to the invention can be implemented from data relating to the displacement of the pressure center, that is to say on the basis of at least one statokinesigram 110 or on the basis of the data. raw from the sensors and relating to the displacement of the center of pressure.

The quantization method 1 according to the invention comprises a step of recording 10 of at least one statokinesigram 110 of the individual, on a storage means 280. This recording can be performed on all types of memory such as transient or non-transitory memories. This recording is preferably performed on a non-transient memory.

The statokinesigram 110 may be generated well before the implementation of the method according to the invention and at a remote location. Alternatively, the statokinesigram 110 can be generated just before the implementation of the quantization method 1 of the equilibrium according to the invention and by the same system. Thus, the quantization method 1 according to the invention may first include a step of generating the raw data relating to the displacement of the pressure center 11. This step is nonetheless optional and it may be performed upstream of the quantization method 1 according to FIG. invention by known devices and methods. The raw data relating to the displacement of the center of pressure correspond for example to the pressure values measured by each of the sensors present on the platform.

These raw data can be the subject of a transformation step 12 in trajectory data of the center of pressure (i.e. a statokinesigram). This transformation step is also optional because it can be performed upstream of the method according to the invention by known methods. FIG. 2 is a graphical representation of a statokinesigram 110 according to the invention and more particularly of the trace 125 of the statokinesigram 110.

The quantification method 1 according to the invention comprises a step of extracting the values of several trajectory parameters 120 from a statokinesigram 110 of the individual. This extraction step can be performed by a data processing module 220 from the statokinesigram of the individual 110 recorded on a storage means 280.

This extraction step 20 allows the transformation of the trajectory data of the center of pressure (i.e. a statokinesigram), so as to obtain parameters retranscribing the properties of this displacement of the center of pressure. The parameters thus calculated are, for example, for each of the instants t, the speed, the acceleration, the position on an axis of the center of pressure.

This extraction step 20 can be performed from a set of data relating to the displacements of the pressure center such as for example the trajectory data of the pressure center or the raw data of the force platform. Indeed, alternatively, the extraction of the parameters according to the invention can be directly performed from the raw data of the sensors in a single step including the generation of the trajectory data 12 of the pressure center and then the transformation into parameters 120.

In the context of the implementation of the method according to the invention, the inventors classified these transformations carried out during the extraction step 20 into three categories: i) those associated with the position of the pressure center, ii) those associated with the stability of the center of pressure, and iii) those associated with the dynamics of displacement of the center of pressure. The inventors have discovered that in order to obtain a reliable quantification of the equilibrium of an individual, a multiparametric method combining the analysis of several parameters from different families of parameters was essential.

Among a multitude of trajectory parameters, the inventors have selected the most efficient trajectory parameters to quantify the equilibrium. They discovered that the method must integrate the calculation of at least one trajectory parameter linked to the position 121 of the pressure center and at least one trajectory parameter related to the stability 122 of the center of pressure.

Thus, the trajectory parameters calculated in the context of the quantization method 1 according to the invention comprise: at least one parameter relating to the position 121 of the pressure center, and at least one parameter relative to the stability 122. from the center of pressure.

Indeed, the inventors have determined that the use of at least one parameter of these two classes makes it possible to increase the precision of the quantification and thus to obtain significantly higher performances, both with respect to the literature but also with respect to the notation algorithms according to the invention using the descriptors of the literature.

In addition, the inventors have also shown that the addition of a parameter relative to the dynamics 123 of the pressure center, the stability parameters and the position of the pressure center, could make it possible to improve the quantification. Thus, according to the invention, advantageously, the calculated trajectory parameters may also comprise at least one parameter related to the dynamic 123 of the center of pressure.

Transformations associated with the position 121 of the center of pressure,

The position of the center of pressure along the X axis

This position corresponds to the position of the center of pressure relative to the median line of the orthonormal coordinate system in a plane of the X axis. For example, in the case of offset position of the center of pressure to the left at a time t , the position along the X axis has, for this instant t a negative value sign of a hyper left support. This position can for example be measured in millimeters.

FIG. 3A represents a set of values of the parameter "position of the center of pressure along the X axis" 121a obtained from the statokinesigram 110 presented in FIG. 2.

The position of the center of pressure along the Y axis

This position corresponds to the position of the center of pressure relative to the median line of the orthonormal coordinate system in a plane of the Y axis. For example, in the case of position shifted backwards at a time t, the position according to the Y axis present, for this instant t a negative value sign of a hyper posterior support. This position can for example be measured in millimeters.

FIG. 3B represents the set of values of the parameter "position of the center of pressure along the Y axis" 121b obtained from the statokinesigram 110 presented in FIG. 2.

The radius in polar coordinates [0045] This distance corresponds to the distance from the center of pressure relative to the average position of the center of pressure according to the orthonormal plane (0,0). For example, in case of a shift of 4 mm from the center of pressure relative to the average position of the center of pressure along an axis of 60 °, at a time t, the radius in polar coordinates has, for this instant t, a value of 4 millimeters.

Such a transformation makes it possible to quantify the overall distance of the pressure center with respect to a point of origin without being limited to the X and Y coordinates of the center of gravity.

Transformations associated with the stability 122 of the center of pressure [0048] Radial balance [0049] The radial equilibrium corresponds to the maximum distance of the pressure center from its current value at a given instant over a predefined duration of t seconds. .

The duration taken into account for the calculation of the radial balance may be between 0.05 to 10 seconds, preferably from 0.1 to 2 seconds.

The radial equilibrium is calculated for all the sampling points of the displacement of the center of pressure over the duration of the acquisition. It is thus a function of time and can be measured for example in millimeters.

Time equilibrium The temporal equilibrium corresponds to the time required for the center of pressure to move more than r millimeters from its current position at a given instant. The distance r taken into account for calculating the time equilibrium can be between 0.1 to 20 millimeters, preferably from 1 to 10 millimeters.

The time equilibrium is calculated for all the sampling points of the displacement of the center of pressure over the duration of the acquisition. It is thus a function of time and can be measured for example in seconds.

Ballistic Interval [0057] The ballistic interval corresponds to the time interval between two equilibrium positions. A position of equilibrium corresponds to a strong value of temporal equilibrium. Preferably, an equilibrium position corresponds to a moment when the temporal equilibrium is greater than 60%, preferably greater than 80%, of the maximum observed on the statokinesigram.

The ballistic interval is calculated over the duration of the acquisition and the number of values obtained may vary depending on the acquisitions. It is not a function of time and can be measured for example in seconds. FIG. 4 represents, as a function of time, all the values of the parameter "time equilibrium" calculated on the basis of the statokinesigram 110 presented in FIG. 2. The values of the "ballistic interval" parameter 122a are for example expressed in seconds and correspond to the times separating values between P1 and P3, and P4 and P6. For example, this parameter includes the values corresponding to the time differences between: P3 - P2; P3 - P1; P6 - P1; P6 - P2; P6 - P3; P6 - P5; P5 - P1; P4 - P3.

Transformations Associated with the Dynamics 123 of the Pressure Center [0060] Velocity of the Displacement of the Pressure Center [0061] The velocity of the displacement of the center of pressure is calculated for all the sampling points of the displacement of the center of pressure on the duration of the acquisition. It is thus a function of time and can be measured for example in millimeters per second.

Acceleration of the displacement of the center of pressure The acceleration of the displacement of the center of pressure is calculated for all the sampling points of the displacement of the center of pressure over the duration of the acquisition. It is thus a function of time and can be measured for example in millimeters per second squared.

Power The power according to the invention corresponds to the value of the scalar product of the speed and the acceleration. This parameter represents the energy expended by the individual to change the standard of pressure center velocity.

Deflection The deflection according to the invention corresponds to the standard of the vector product of speed and acceleration. It represents the energy expended by the individual to change the direction of the pressure center velocity (i.e its displacement, trajectory or movement).

In the context of the study of a statokinesigram 110, the power and the deviation are two parameters that can advantageously replace the length of the displacement of the center of pressure, the variance of the speed as a function of the Y axis (VFY ) or the LFS (ie ratio of the length on the surface), providing information according to the prior art on the energy consumption spent by the subject to control his posture. In particular, the deviation proposed by the inventors makes it possible to evaluate the energy expenditure of the individual in a new angle and is more relevant in the method according to the invention than the parameters proposed in the prior art (eg VFY, LFS ).

The quantization method 1 according to the invention may also comprise a step of determining a plurality of quantizers 130 from the trajectory parameters 120. This determination step 30 may be performed by a data processing module 220. Once the trajectory parameters 120 generated during the extraction step 20 of the pressure center trajectory data, the work of the prior art has generally focused exclusively on the average and the variance of these parameters during an acquisition period.

In the process developed by the inventors, the latter were interested in many other methods of transforming these parameters to generate, from these sets of values and in combination with other sets of values, a single representative value of the balance of an individual. Thus, new quantizers never used in the past have been generated.

The step 30 for determining a plurality of quantizers 130 from the trajectory parameters 120 therefore consists in transforming, for each trajectory parameter 120, the set of values into a single value that can be used in the context of a comparison for example via a statistical model of comparison. This transformation into a single value (quantizers) can be repeated for several trajectory parameters or for the same trajectory parameter. FIG. 5 represents some of these quantizers as calculated for the "X-axis pressure center position" parameter.

The most informative quantifiers in the context of the quantification method according to the invention are the average value, the median value, the variance, the mean of the squares, or an extreme value, of a trajectory parameter 120. extract. Thus, preferably, the quantization method 1 according to the invention comprises, for at least one parameter, the calculation of the average value 127, the median value 128, the variance, the average of the squares or of minus an extreme value, of said path parameter (121, 122, 123). Even more preferably, the quantization method 1 according to the invention comprises, for at least one parameter, the calculation of at least one extreme value of said parameter.

The extreme values of a parameter are obtained via the determination of a percentile. A percentile, or percentile, can for example be calculated by ordering and then sorting the set of values of a parameter into 100 subsets comprising the same number of values. FIG. 5 shows the percentile O 126 and the percentile 95 129 of the X-axis pressure center position respectively corresponding to the highest value of the X-axis pressure center position within the X axis. % of the lowest values and the lowest value of the center of pressure position along the X axis within the highest 5%. For example, the 10th percentile is the value between 10% of the lowest values and 90% of the highest values, while the 95th percentile is the value between 95% of the lowest values and 5% of the highest values. , an extreme value of a trajectory parameter 120 corresponds to a percentile less than or equal to 15 or to a percentile greater than or equal to 85; more preferably, at a percentile less than or equal to 10 or a percentile greater than or equal to 90 and even more preferably at a percentile less than or equal to 5 or a percentile greater than or equal to 95.

Alternatively, an extreme value of a trajectory parameter 120 may correspond to a percentile greater than or equal to the percentile 5 and less than or equal to the percentile 15 (low extreme values) or to a percentile greater than or equal to the 85 percentile and a percentile less than or equal to percentile 95 (high extreme values).

In addition, beyond the calculation of a quantizer 130, the inventors have shown that the comparison open eyes (O) / eyes closed (F) on certain parameters or quantifiers would differentiate performance in terms of balance . Thus, in particular, the quantization method 1 according to the invention comprises for at least one quantizer, the calculation of a ratio O / F or F / O.

During their development, the inventors have selected a set of quantizers 130 particularly relevant for quantifying the equilibrium. Preferably, in the context of the quantization method 1 according to the invention the quantizer linked to the position of the pressure center is selected from among the following values: the average or median value on the X or Y axis of the pressure center, - the average value of the squares of the X or Y values of the center of pressure, - the extreme values on the X or Y axis of the pressure center, and - the variance on the X or Y axis of the pressure center .

In particular, in the context of the quantization method 1 according to the invention, the quantizer related to the extreme values on the axis X or Y of the pressure center is selected from: - the extreme low values of the position of the center of pressure along the transverse axis (Y), preferably included in percentiles less than or equal to 10, these values correspond to the observed maximum retropropulsion, and - the extreme extreme values of the trajectory of the center of pressure along the transverse axis (Y), preferably included in percentiles greater than or equal to 90.

Preferably, the quantization method according to the invention comprises the calculation of at least one, more preferably at least two quantizers related to the position of the pressure center that can be selected independently from the following quantifiers: the percentile 15, preferably the percentile 10 and even more preferably the percentile 5 of the position of the center of pressure on the X axis, preferably for a displacement of the center of pressure measured while the individual had the eyes closed, - the median value on the Y axis of the center of pressure, preferably for a displacement of the center of pressure measured while the individual had his eyes closed, - the median value on the X axis of the center of pressure, preferably for a displacement of the center of pressure measured while the individual had his eyes open, - the percentile 15, preferably the percentile 10 and even more preferably the percentile 5 of the position of the center of pressure on the Y axis, preferably for a displacement of the center of pressure measured while the individual had his eyes closed, - the variance on the Y axis of the trajectory of the center of pressure, preferably for a displacement of the center of pressure measured while the individual had his eyes closed, - the variance on the X axis of the trajectory of the center of pressure, preferably for a displacement of the measured center of pressure. while the individual had his eyes closed, - the mean or median values on the Y axis and more particularly the ratio between these values for a statokinesigram recorded while the individual had eyes open and these values for a statokinesigram recorded while the individual had his eyes closed, - the extreme values of the radius in polar coordinates as for example: o the percentile 15, preferably the percentile 10 and so More preferred is the percentile 5 of the radius in polar coordinates of the center of pressure, preferably for a displacement of the measured center of pressure while the individual had the eyes closed, where the percentile 85, preferably the 90th percentile, and so on. more preferably the 95th percentile of the polar coordinate radius of the center of pressure, preferably for a displacement of the measured center of pressure while the individual had the eyes open, - the median of the radius in polar coordinates of the center of pressure, and - the average value of the squares of the radius in polar coordinates of the center of pressure.

Preferably, the quantization method 1 according to the invention comprises the calculation of at least one, more preferably at least two quantizers related to the stability of the pressure center that can be selected independently from the following quantifiers. the percentile of the radial equilibrium for a duration of 0.5 seconds, preferably for a displacement of the center of pressure measured while the individual had his eyes closed, the 90th percentile of the radial equilibrium for a duration of 0.1 second, preferably for displacement of the measured center of pressure while the individual had his eyes open, - the variance of the temporal equilibrium for a distance of between 1 to 10 millimeters, preferably from 1 to 5 millimeters, preferably the ratio of the value obtained for a displacement of the center of pressure measured while the individual had his eyes open on the value obtained for a displacement. of the center of pressure measured while the individual had his eyes closed, - the average value of the average value of the squares or the median value, preferably the mean value of the ballistic interval values, preferably for a displacement of the center of pressure measured while the individual had his eyes closed, and - the extreme high values of the ballistic range, for example the 85th percentile, preferably the 90th percentile and even more preferably the 95th percentile of the interval ballistic.

Preferably, in the context of the quantization method according to the invention, the quantizer related to the dynamics of the pressure center is selected from among the mean, the median, the average of the squares or the variance of the power, preferably for a displacement of the center of pressure measured while the individual had his eyes open.

Preferably, the quantization method according to the invention comprises the calculation of at least one, more preferably at least two quantizers related to the dynamics of the pressure center that can be selected independently from the following quantifiers: the extreme low values of the acceleration of the displacement of the center of pressure such as, for example, the percentile 15, preferably the percentile 10, the extreme low values of the deviation of the displacement of the center of pressure, for example the percentile 15, of preferably the percentile 10, even more preferably the percentile 5, preferably for a displacement of the measured center of pressure while the individual had the eyes open, and - the extreme high values of the pressure center displacement speed as for example the 85th percentile, preferably the 90th percentile, preferably for a displacement of the pressure center it was then measured that the individual's eyes were open.

The inventors propose for the first time a multiparametric method based on the combination of a plurality of parameters and / or quantizers. Thus, in addition to extracting at least one parameter related to the stability 122 of the center of pressure and at least one parameter related to the position 121 of the center of pressure, they have determined that to obtain a more reliable value in the process of Quantization 1 according to the invention, it is preferable to calculate and then combine several quantizers, for example several quantizers relating to the same trajectory parameter. Thus, in order to better quantify the equilibrium and more particularly to allow the prediction of a risk of falling, it is preferable to calculate, from trajectory parameters 120, at least five quantizers 130, even more preferably at least eight quantifiers 130.

In addition, the quantization method 1 according to the invention comprises a step of comparing 40 of several quantizer values 130 with the values of the same quantizers obtained from reference statokinesigrams 111.

This comparison step can be implemented by means of a data processing module 220 and can be performed via known statistical methods. Preferably, this comparison step 40 can be carried out from comparisons, classifications or learning models such as: perceptron, Kernel, Multiple kernel learning, vector support machine, decision trees, random drills, bagging, AdaBoost, k-nearest neighbor and / or penalized linear fisher discriminant.

Among the statistical methods, the linear regression models are simple and easy to implement. However, these linear models are limited because of the linearity assumption and they are not optimal for the quantization method 1 according to the invention. Thus, the comparison step 40 involves the use of a nonlinear statistical model.

More preferably, the comparison step 40 is based on a model, driven on a data set and configured to predict the tag of a new population observation. For example, for the purposes of calibration, it is possible to use a dataset from a set of individuals representative of a population, characterized by several reference statokinesigrams 111 and their associated reference parameters and / or quantizers. and by a binary label or class, for example of the form "good balance" / "bad balance". The data set may also include multiple tags. In the context of the present invention, the quantification method 1 may be based on at least twenty-five reference statokinesigrams 111, preferably at least fifty, and even more preferably at least one hundred.

The comparison step 40 includes the use of a supervised statistical learning model. Indeed, the inventors have determined that, in the context of the quantification method 1 according to the invention, the best quantification results are obtained from methods based on principles of statistical learning preferably supervised methods adapted to multivariate data. .

In supervised learning methods, the inventors have identified some more effective methods of treatment. This could be related to the fact that the quantizers from the method according to the invention come from data sets having a large imbalance and generally non-uniform metrics. Thus, preferably, this comparison step 40 is performed via various nonlinear supervised statistical learning methods. For example, it can be cited: kernel methods (eg Large Vector Separators - Vector Vector SVM Support, Kernel Ridge Regression) described for example in Burges, 1998. (Data Mining and Knowledge Discovery.) A Tutorial on Vector Support for Pattern Recognition ). set methods (eg Bagging, Boosting, decision trees, Random Forest) described for example in Brieman, 2001 (Machine Learning, Random Forests), or the neural networks described for example in Rosenblatt, 1958 (The perceptron: a probabilistic model for information storage and organization in the brain).

As discussed above, the method according to the invention aims, beyond the prediction of belonging to a group of individuals, to quantify an equilibrium, that is to say to associate a numerical value or a quality level at the equilibrium of an individual. Thus, the quantization method 1 according to the invention comprises a step of determining a representative value 160 of the equilibrium of the individual on the basis of the results of the comparison step 40.

Preferably, the representative value 160 may be based on the definition of an order relationship on the population in order to prioritize the individuals that compose it. This step can be referred to as binary scheduling. In practice, the problem of binary scheduling consists in learning a model, from a sample Sn = {(Xi, Yi), 1 <i <n} of copy of the pair (X, Y) in order to order the minus a new observation Xo of the random variable X, for which the label Yo is not known. Such a model makes it possible to order the new observations by ranking at the top of the ranking the observations with the highest probability of being positive and at the bottom of the ranking the most likely negative observations.

Alternatively, the allocation of this representative value 160 may be based on several statistical methods such as soft-margin SVM or Gaussian Mixture Models.

Preferably, the quantization method 1 of the equilibrium according to the invention includes the use of a scoring algorithm 500 that can be integrated into a data processing module 220 and configured to implement: the step of comparing the values of several quantizers 130 to the values of the same quantizers obtained from reference statokinesigrams 111, and the step of determining a representative value 160 of the equilibrium of the individual to the quantizer. 'result of the comparison.

This scoring algorithm 500 may have been constructed from different models of supervised learning. Even more preferably, this scoring algorithm 500 relies on a supervised statistical learning model configured so as to minimize a risk of the scheduling rule and thus making it possible to obtain more efficient prediction rules, such as the methods: Rankboost, Adarank, SVM rank, Lambda rank, Forest Ranking or neural network.

Even more preferably, the scoring algorithm 500 has been constructed from a combination of several nonlinear supervised statistical learning models. Thus, preferably, the creation of the scoring algorithm 500a includes a "Bagging" step and / or a Boosting step. For example, the scoring algorithm 500 may include several Ranking Tree trees constructed using a Bagging step. Indeed, the notation algorithms developed by the inventors in the context of the present invention are very effective if they incorporate a Bagging step.

Even more preferably, the previously calibrated scoring algorithm 500 was obtained by implementing a Ranking Forest or Random Forest method.

[0096] Bagging, or Bootstrap Aggregating is a method of training a learning algorithm on different subsets of the initial learning set. In practice, B subsets are generated by randomly drawing N times each in the initial training set. The subsets are called bootstrap samples. Bagging and its implementation are described in detail in Galar et al 2011 (A Review on Sets for the Class Imbalance Problem: Bagging, Boosting, and Hybrid-Based Approaches). Bagging is very effective in combination with learning algorithms such as decision trees and the inventors have further discovered that the underbagging method is particularly well suited to the method according to the invention.

The boosting gathers a set of algorithms such as: Adaboost, LPBoost, TotalBoost, BrownBoost, xgboost, MadaBoost, LogitBoost. Boosting refers to a method of training a learning algorithm to produce accurate decisions by combining "weak" decision rules (ie, capable of recognizing two classes at least as well as chance would). Boosting allows successive iterations of weak rules, the generation of a rule of classification, or decision, strong. Boosting is therefore a sequential method and each sample is drawn based on the performance of the basic rule on the previous sample. Boosting and its implementation are described in detail in

Freund &amp; Schapire 1999 (Machine Learning Large Margin Classification Using the Perceptron Algorithm).

The Random Forest or Random Forest analysis is part of the supervised statistical learning methods. It is based on the principle of Bagging, the originality of the analysis is the aggregation of K trees built with a small number of variables. Each node is built with a small number of variables but always constant and randomly chosen. For example, several independent models are built to predict the same Y variable and then the predictions of these models are aggregated. This aggregation of independent models makes it possible to reduce the variance and thus reduce the prediction error.

Preferably, the quantification method 1 according to the invention makes it possible to obtain a quantification of the equilibrium in the form of a score or a value between zero and one hundred, proportional to the quality of the equilibrium. . For example, a value less than 30 indicates a low balance. Preferably, the step of determining an equilibrium value based on the results of the comparison is followed by a recording step 60 of the representative value 160 of the equilibrium obtained and possibly the association of said value with a unique identifier linked to said individual.

[00101] Thus, this allows the individual to compare his representative value of the equilibrium 160 over time. Thus, preferably, the quantification method 1 according to the invention can be implemented in the same individual at different dates so as to follow the evolution of its value representative of the equilibrium 160 and therefore the quality of the his balance.

As shown in FIG. 1, the quantization method 1 of the equilibrium according to the invention may also comprise a step of graphical representation 70 of said representative value of the equilibrium 160. The value may be the object of a display via a display module 340a. This display can be a simple display indicating a value or a graphical representation.

This graphical representation can represent the evolution of the representative value of the equilibrium 160 over time or the placement of this value in one of a group of individuals.

[00104] Said representative value of the balance 160 can also be transmitted on remote systems such as tablets, servers or personal computers. Thus, the quantization method 1 according to the invention may comprise a step of transmitting the value representing the equilibrium 160, calculated quantizers and / or calculated parameters to at least one communicating system such as a tablet, a server or a computer, via at least one communication network.

[00105] Preferably, the invention relates to a quantization method 1 of the equilibrium comprising the quantization of the static equilibrium and the dynamic equilibrium. Even more preferably, the invention relates to the quantification of static equilibrium.

As part of the development of this new quantization method 1 of the equilibrium, the inventors have verified the relevance of value representative of the equilibrium 160 obtained via the statistical models developed and in particular the notation algorithms used by the intermediate of ROC curves. The statistical models developed and in particular the notation algorithms 500 used by the inventors make it possible to obtain AUCs greater than 0.7, preferably greater than 0.8.

[00107] Figure 6 shows a particular embodiment of the invention. From this embodiment, the inventors analyzed the statokinesigrams of 49 individuals as presented in the examples.

[00108] Figure 7 shows a decision tree according to the invention as constructed by a Ranking Forest model.

F1 and F2 correspond, within the meaning of the invention, to quantizers relating to a parameter related to the position 121 of the center of pressure. In addition, the 90 percentile of the ballistic range is a quantizer particularly preferred in the sense of the invention. F3 corresponds to a quantizer relative to the stability 122 of the center of pressure. The combination of these quantifiers by means of a scoring algorithm 500 makes it possible to generate a representative value 160 of the equilibrium (e.g., 0, 0.33, 0.66, 1).

According to one aspect, the invention relates to a quantization device 2 of the equilibrium capable of implementing the quantization method 1 of the equilibrium according to the invention.

More particularly, the quantization device 2 of the equilibrium according to the invention comprises: - a communication module 210, able to receive a statokinesigram 110 of said individual, - a storage means 280, and - a processing module 220.

A quantization device of the equilibrium 2 according to the invention is shown schematically in FIG. 8. These modules are distinct in FIG. 8, but the invention can provide various types of arrangement, for example a single cumulative module. all the functions described here. These modules can be divided into several electronic cards or collated on a single electronic card.

The communication module 210 is configured to receive and transmit information to remote systems such as platforms, tablets, telephones, computers or servers. The communication module makes it possible to transmit the data on at least one communication network and may comprise wired or wireless communication. Preferably, the communication is operated via a wireless protocol such as wifi, 3G, 4G, and / or Bluetooth.

The communication module 210 allows for example to receive raw data displacement of the pressure center or statokinesigrams. It is also configured to send the data relating to the calculated parameters, the computed quantizers and the representative value of the balance 160. These data exchanges can take the form of sending and receiving files containing the raw values of the pressure sensors. , files containing the coordinates of the trajectory of the center of pressure, and files comprising the parameters, the quantizers and the values representative of the equilibrium 160 determined from the statokinesigram 110.

The exchanged data can preferably be transferred in encrypted manner and associated with a specific key of the individual studied.

The data processing module 220 is configured to: • extract, from a statokinesigram 110 of an individual transmitted by the communication module 210, values of at least one trajectory parameter relating to the position 121 of the center of pressure and the values of at least one trajectory parameter related to the stability 122 of the center of pressure, • Determining several quantizers 130, from the values of the trajectory parameters 121, 122 extracted, • Comparing the values of said quantifiers 130 to the values of the same quantizers obtained from reference statokinesigrams 111, and • determining a value 160 representative of the equilibrium of the individual on the basis of said comparison.

The data processing module 220 advantageously comprises a processor and is able to connect to a storage means 280.

The storage means 280 may comprise a transient memory and / or a non-transitory memory. It is able to record, for example in the form of files, the raw values of the pressure sensors, the coordinates of the trajectory of the center of pressure, the parameters 120, the quantizers 130 and the values representative of the equilibrium 160 determined in FIG. from the statokinesigramme (s) 110. The non-transitory memory allows for example to record the configuration of the data processing module while the non-transitory memory allows for example to record the statokinesigram 110. The non-transitory memory can be a medium such as a CDrom, a memory card, or a hard disk hosted by a remote server.

[00121] Preferably, the data processing module 220 is configured to implement the different steps of the quantization method 1 according to the invention. Thus, the preferred steps of the quantization method 1 of the equilibrium according to the invention are also preferred configurations for the data processing module 220 according to the invention.

The quantization device 2 according to the invention may also comprise a generation module 230 of statokinesigram 110. This module is configured to generate the data relating to a statokinesigram 110 (eg position along the x and y axes as a function of time ) from raw pressure center displacement data as they can be generated by force or pressure sensors.

The quantization device 2 according to the invention may also comprise a resampling module 240. Indeed, the devices capable of generating raw data displacement of the pressure center or statokinesigram 110 do not all propose a frequency controlled sampling. Thus, certain devices can cause the generation of statokinesigram 110 having a first random frequency, the frequency of which can not be predicted because it varies constantly during acquisition, for example, for the same statokinesigram 110, between 10 and 1000 Hz. However, such a variation in frequency can lead to decreases in the performance of the quantization method 1 of the equilibrium according to the invention. Thus, preferably, the resampling module is configured to process the raw data or the statokinesigram 110 at a first frequency so as to generate statokinesigrams 110 resampled at a second frequency and having a substantially constant frequency. By substantially constant frequency, it is necessary to understand a frequency varying from less than 10% within the statokinesigram 110, preferably varying from less than 5%, even more preferably varying from less than 1%.

The statokinesigram 110 at a second frequency, generated by the resampling module 240, has a sampling frequency of at least 25 Hz. Preferably, the second frequency is substantially identical to the frequency of the reference statokinesigrams. 111.

The quantization device of the equilibrium according to the invention may comprise a denoising module 290 configured to filter the raw data generated by the pressure or force sensors so as to reduce or eliminate the spurious signals. Denoising can be based on various methods such as wavelet denoising, thresholding, Wiener filter and deconvolution.

The device may also include a control interface 260. This control interface is configured to allow the interaction of a user with the equilibrium quantization device. It may comprise, for example, manual actuators (e.g. buttons) or a touch screen adapted to receive the commands of the user.

The device may also include a display module 270. This display module may comprise a liquid crystal display. It makes it possible to display various information such as the results of the quantization, the representative value of the equilibrium 160, the progression over time of said value and its positioning with respect to the values representative of the equilibrium 160 within a group of people.

According to another aspect, the invention relates to a quantization system of the equilibrium 3 shown in FIG. 9, adapted to implement the quantization method 1 according to the invention. Preferably, the equilibrium quantification system 3 of an individual comprises: a platform 310, said platform 310 being adapted to receive an individual and comprising pressure and / or force sensors 312 configured to generate data 313, at a first frequency, a function of a pressure exerted by the feet of the individual on the platform 310, a treatment unit 320 of the raw data, arranged to obtain at least one statokinesigram 110 of the individual from raw data 313 generated by the platform 310, and - an equilibrium quantization device 2 described above, able to communicate with the processing unit 320.

As shown in FIG. 9, the platform 310 according to the invention is a support intended to receive an individual and able to measure the displacement of a pressure center by means of force and / or pressure sensors. . Any sensor system for measuring the center of pressure may be used. The only prerequisite is that platform 310 be able to produce raw data to position the center of pressure. This support may, for example, be a pair of sole, a "smart" floor equipped with a sensor or a balance. Preferably platform 310 comprises a plate 311. In general, the dimensions of one side of the plate 311 may be between 15 and 70 cm, preferably of the order of 25 to 40 cm. This plate 311 may for example comprise a template allowing reproducible positioning of the feet, between individuals and over time for the same individual. The platform 310 is configured to measure the pressure or forces applied to the plate at a given instant and comprises sensors 312 for this purpose. The sensors will transform the applied force into an electrical, optical or magnetic signal corresponding to the raw data. . These raw data can be combined and processed in order to specify the coordinates of the pressure center and to follow its variations over time. These sensors can be pressure or force sensors. A force sensor measures the resultant of the bearing forces of a standing subject. The measurement of the forces and moments exerted on the platform makes it possible to specify the coordinates of the center of pressure and to follow its variations over time. A pressure sensor may comprise, for example, a pressure cell configured to measure or detect the pressure induced by the weight of the individual placed on the platform or the pressure exerted by the feet of the individual on the platform. The data from these sensors are the raw data. The platform 310 may also comprise a plurality of resistive or piezoelectric sensors (for example between 1000 and 6000 sensors). The sensors are preferably 4 in number, and located at the ends of the platform for example separated from 20 to 50 cm for the sensors right and left or up and down. For example, as shown in Figure 9, the platform 310 has four sensors 312 located at the four corners of the board (Top Left, Top Right, Bottom Left, Bottom Right).

The platform 310 advantageously comprises a time counting module and can be configured to measure the values of its various sensors 312 at random intervals, at a frequency that may vary for example from 10 Hz to 1000 Hz. Preferably, the platform 310 is configured to measure the values of its various sensors 312 at a frequency greater than or equal to 25 Hz, more preferably greater than or equal to 50 Hz.

Even more preferably, the platform 310 is configured to measure, during the acquisition of a statokinesigram 110, the values of its different sensors 312 at a frequency greater than or equal to 25 Hz and substantially constant. Indeed, if the sampling frequency is too low, or too random, the quantization of the equilibrium will not be sufficiently precise. If the frequency is not constant then preferably the average acquisition frequency is greater than or equal to 60 Hz, more preferably greater than or equal to 75 Hz.

The platform 310 may comprise a display device, preferably positioned so that the individual, standing on the plate 311 can see the display device.

The platform 310 may also include a loudspeaker device that can give instructions to the individual (e.g. up or down the board 311). These instructions may also be given by the display device.

The platform 310 may also include a module for measuring the weight of the individual, its fat, water, bone, muscle, heart rate and / or body mass index.

The equilibrium quantization system 3 also includes a raw data processing unit 320 generated by the platform. This raw data processing unit 320 is arranged and / or configured to generate at least one statokinesigram 110 of the individual from the raw data generated by the sensors 312. This raw data processing unit 320 can for example be integrated into the data processing unit 320. the platform 310 as shown in FIG. 9. Nevertheless, it can also be integrated in a remote server 330, in the quantization device 2 (eg it integrates therein the generation module 230 of the statokinesigram 110) or in a device order 340.

The equilibrium quantization system 3 may comprise a remote server 330 as shown in FIG. 9. It is for example possible to access this remote server 330 via a web interface or directly via the appropriate functions. directly implemented on a controller 340. All communications between the control device (s) 340 and the remote server 330 can be secured for example by HTTPS protocols and AES encryption 512.

This remote server can host the quantization device 2. Thus, a single quantization device 2 can track a plurality of individuals.

The quantization system according to the invention may comprise a control device 340 of the system configured to interact with the platform 310 and the equilibrium quantization device 2. This control device 340 of the system makes it possible, for example, to control acquiring the data from the platform 310 and displaying the results from the quantization device 2.

This control device 340 of the system is of presence a mobile device such as a tablet 340a, a laptop, or a watch.

According to another aspect, the invention relates to a computer program product 4 configured to implement the quantization method 1 of the equilibrium according to the invention. The computer program product 4 is recorded on a non-transient memory medium and is capable of being executed on a computer, a tablet or a server; said computer program comprising at least: an algorithm adapted to extract, from a statokinesigram 110 of an individual, values of at least one trajectory parameter relative to the position 121 of the center of pressure and the values of at least one trajectory parameter related to the stability of the center of pressure - an algorithm adapted to determine several quantizers 130, from the values of the trajectory parameters 121, 122 extracted, an algorithm adapted to compare the values of said quantizers the values of the same quantizers obtained from reference statokinesigrams 111, and an algorithm adapted to determine a value representative of the equilibrium of the individual on the basis of said comparison.

More preferably, the computer program product 4 is recorded on a non-transitory memory medium and is capable of being executed on a computer, a tablet or a server; said computer program comprising at least: an algorithm adapted to extract, from a statokinesigram 110 of an individual, values of at least one trajectory parameter relating to the position 121 of the pressure center, values of at least one trajectory parameter related to the stability 122 of the center of pressure and the values of at least one trajectory parameter related to the dynamics 123 of the pressure center - an algorithm adapted to determine several quantizers 130, from the values extracted trajectory parameters 121, 122, 123, and - a notation algorithm adapted to compare the values of several quantizers 130 and determine a value representative of the balance of the individual on the basis of said comparison [00143]. method, device, system and computer program product according to the invention allow quantification of the balance of an individual and can have many applications.

Indeed, the invention makes it possible to provide a measurement tool, namely a method, the device for implementing the method and the system integrating the device, which makes it possible to obtain an encrypted and objective value of the device. 'balance of an individual in order to answer three main questions relating to the balance of an individual: - A] the evolution of the balance, natural or under treatment, - B] the quality of the balance and therefore , its corollary, the severity of a possible equilibrium disorder (eg what is the risk of falling?), and - C] the cause (s) of the eventual balance disorder.

Indeed, the generation of a representative value of the balance 160 of an individual and indicative of the quality of his equilibrium enables the individual, or others, to assign a numerical value and objective to this balance.

These values or these scores can be used as part of a follow-up in time to identify deviations from the reference learned.

Similarly, the invention can be used to highlight the effects of different treatments and the recovery rate could be followed by the quantification of the balance according to the invention. Thus, the invention can be implemented in the context of the evaluation of the performance of sport programs, prostheses, sports shoes, compensation soles, rehabilitation protocols, treatment of neurological disorders and / or of surgical techniques. The method according to the invention is particularly suitable for the elderly [00148] In addition, the invention can be used to compare the quality of the balance of an individual with the quality of the balance of other individuals and to determine by example if this individual presents a risk of falling. Thus, the invention can be implemented as part of the measurement of a risk of falling for example to 6 months. In this context, the representative value of the equilibrium 160 determined by the quantification method 1 according to the invention, a value indicative of a risk of falling to 6 months. In particular, the individual within the meaning of the invention is a person over 60 years, preferably over 70 years.

In addition, the comparison of the values of the quantizers 130 obtained in an individual to the values of the quantifiers obtained in different categories of persons can make it possible to target the management of the individuals and to direct them towards the adapted services (eg traumatology, rheumatology, neurology). Thus, the invention can be implemented in the context of the determination of the origin of the disturbance of equilibrium.

Examples [00150] Individuals Studied [00151] The results presented below were made according to a protocol approved by the National Agency for the Safety of Medicines and Health Products and a written consent was obtained for all the participants.

The invention has been implemented on a group of 49 individuals having the following characteristics: - over 70 years old; - able to stand on the platform, BMI less than 21 at the time of consultation, and gave informed consent.

[00153] Among the 49 individuals included in the study: - 27 reported a fall during the 6 months preceding the consultation, - 7 falls led to trauma and hospitalization, 16 led to trauma and no hospitalization, and - 4 were classified as benign (no trauma, no hospitalization).

[00154] Measurement of the displacement of the center of pressure [00155] During the consultation, the movements of the center of pressure of the individuals were followed by a Wii Balance Board (registered trademark) and recorded thanks to a custom application specially developed in the framework of the invention. The feet were positioned in the most comfortable position for the patient, without exceeding the width of the shoulders. The trajectory of the center of pressure was recorded for 20 seconds with eyes open, then 20 seconds with eyes closed. A drop questionnaire was completed for each individual to record reported falls that occurred within the last 6 months.

[00156] Pretreatment Before calculating the statokinesigrams, the output of the raw signals by the WBB were de-noiseed and resampled.

[00158] Statistical analysis.

For each statokinesigram, fifteen quantifiers were calculated from the statokinesigrams.

The quantizers obtained from the reference statokinesigrams make it possible to parameterize a notation algorithm based on a Ranking Forest algorithm such as the subpart of a decision tree represented in FIG. 7. F1 corresponds to the closed eyes / ratio. open eyes for the 90th percentile of the ballistic range, F2 is the ratio of closed eyes / open eyes for the variance of the position on the Y axis and F3 is the variance of the temporal equilibrium with the eyes closed. The quantizers obtained from a statokinesigram to be analyzed serve as an input for this notation algorithm which calculates a value, between 0 and 1, indicative of the equilibrium and more particularly of the risk of falling of the individual. A low score corresponds to a high risk of falling, while a high score is characteristic of a low risk of falling.

Thus, unlike the prior art, the invention provides a measurement tool for obtaining a numerical value and objective balance of an individual.

Claims (10)

claims A method of quantizing the equilibrium of an individual to obtain a representative value (160) of the balance of said individual, said method being implemented by a device comprising at least one connected data processing module (220) to a storage means (280) and said method comprising: a) a step of recording (10), on the storage means (280), at least one statokinesigram of the individual (110) obtained from a platform (310) comprising pressure and / or force sensors (312), b) an extraction step (20), by the data processing module (220) and from the statokinesigram (s) ) of the individual (110) recorded on the storage means (280), values of at least one trajectory parameter relative to the position (121) of the pressure center selected from: the position of the center of pressure along the X axis, the position of the center of pressure along the Y axis and the radius in polar coordinates and values of at least one trajectory parameter relative to the stability (122) of the pressure center selected from: the radial equilibrium, the temporal equilibrium and the ballistic interval, c) a step determining (30), by the data processing module (220), the value of a plurality of quantizers (130), from the values of the trajectory parameters (121, 122) extracted in step b), d) a step of comparing (40), by the data processing module (220), said values of several quantizers (130) with the values of the same quantizers obtained from reference statokinesigrams (111), and e) a determining step (50), by the data processing module (220), said representative value (160) of the balance of the individual after the comparison (40). 2. Quantization method according to claim 1, characterized in that step b) further comprises extracting (21) the values of at least one trajectory parameter relative to the dynamic (123) of the selected pressure center. among: the speed of displacement of the center of pressure, the acceleration of the displacement of the center of pressure, the power and the deviation. 3. A method of quantification according to one of claims 1 or 2, characterized in that step b) is performed from two statokinesigrams (110) generated during a Romberg test. 4. Quantization method according to any one of claims 1 to 3, characterized in that at least one of the quantizers (130) determined in step c) is an extreme value of said path parameter (121, 122, 123 ), said extreme value corresponding to a percentile between a percentile greater than or equal to 5 and a percentile less than or equal to 15, or a percentile between a percentile greater than or equal to 85 and between a percentile less than or equal to 95, values of the trajectory parameter (121, 122, 123). 5. Quantization method according to any one of claims 1 to 4, characterized in that steps d) and e) are performed by implementing the values of the quantizers (130) determined in step c) in an algorithm notation system (500) previously calibrated on the basis of the values of the same quantizers obtained from the reference statokinesigrams (111). 6. Quantization method according to claim 5, wherein the previously calibrated scoring algorithm (500) was obtained by the implementation of a supervised learning statistical method, preferably a RANKING FOREST or RANDOM method. FOREST. 7. A method of quantification according to one of claims 5 or 6, wherein the previously calibrated scoring algorithm (500) has been obtained by the implementation of a statistical method of supervised learning including the implementation of a step of Bagging. 8. Quantization method according to any one of claims 1 to 7, characterized in that it comprises in step c) the determination of the value of at least five quantizers (130). 9. Device for quantifying (2) the equilibrium of an individual, said device comprising: - a communication module (210) adapted to receive a statokinesigram (110) of said individual, - a storage means (280) adapted to storing the statokinesigram (110), and - at least one data processing module (220), adapted to connect to the storage means (280), said data processing module (220) being configured to: o Extract, to from a statokinesigram (110) of said individual transmitted by the communication module (210), values of at least one trajectory parameter relating to the position (121) of the pressure center selected from: the position of the center of pressure along the axis X, the position of the center of pressure along the Y axis and the radius in polar coordinates and values of at least one trajectory parameter related to the stability (122) of the pressure center selected from: 1 radial balance, the time equilibrium and ballistic interval, o Determine several quantizers (130), from the values of trajectory parameters (121, 122) extracted, o Compare the values of said quantizers (130) with the values of the same quantizers obtained from statokinesigrams reference numeral (111), and o determining a value (160) representative of the equilibrium of the individual on the basis of said comparison. 10. Quantification system (3) of the balance of an individual, comprising: - a platform (310), said platform (310) being adapted to receive an individual and comprising pressure sensors (312) and / or force configured to generate raw data (313), at a first frequency, a function of pressure exerted by the feet of the individual on the platform (310), - a processing unit (320) of the raw data, arranged to obtaining at least one statokinesigram (110) of the individual from the raw data (313) generated by the platform (310), and - an equilibrium quantizing device according to claim 9, able to communicate with the unit treatment (320).