A method for predicting an individual's physical state, correlated with a target activity event, based on previous activity events.

Abstract

The invention relates, on the one hand, to a method for predicting an individual's fitness level, correlated with a target activity event, based on previous activity events, and on the other hand, to a management system. Abstract figure: Fig. 2

Description

Title of the invention: Method for predicting the physical state of an individual, correlated to a target activity event, based on previous activity events. technical field

[0001] The present invention is in the field of detection and processing of physical, physiological and psychological measurements for the prediction of an individual's fitness level and optimization strategies. State of the art

[0002] Methods and devices for detecting and processing physical measurements for predicting an individual's fitness level have been known for a long time.

[0003] Among them, connected watches or vests or belts equipped with sensors are the most represented.

[0004] These devices provide the individual who equips themselves with various information related to the activity carried out and / or the anomalies detected, such as the distance covered, the speed of execution, the heart rate, the imbalances observed.

[0005] Thus, such devices and the data processing methods they incorporate can be used for the optimization of sports activity programs or for the monitoring of moderate activities, for example for the detection of falls while walking in elderly people or people with disabilities.

[0006] Furthermore, there is no system equipped with a sufficiently diverse panel of sensors to obtain a global view of physical activity and its consequences on an individual.

[0007] Also, although existing systems allow for the monitoring of a specific physical activity throughout its execution and can sometimes record a searchable history, their contribution remains advisory and limited to that physical activity, and the recommendations provided remain generic, when they exist at all. Indeed, they only allow for the visualization of statistics from a single source, making it impossible to cross-reference information.

[0008] A smartwatch, for example, will not be able to take into account the individual's weight variations in order to offer them an updated training program.

[0009] Also, these systems are not capable of controlling the prevention or resumption of physical activity after an injury.

[0010] There is therefore a need to take into account the history of the individual's activities in order to assess their impacts, predict the risks involved and offer personalized support to the individual. Summary of the invention

[0011] The invention aims to improve the situation described above.

[0012] The invention aims to remedy the drawbacks of the prior art, by proposing a method and a minimally invasive system providing short- and medium-term visibility, updatable at any time, for the prediction of fitness status and the adaptation of personalized recommendations for an individual, all in an adaptive manner, preferably in real time or near real time.

[0013] Another object of the invention is to propose an adaptive process and a system which integrates multi-source data such as performance data, medical data, biomechanical data, reflecting all the physical activities relevant to the individual.

[0014] To this end, the invention relates to a method for predicting the physical state of an individual, correlated with a target activity event, based on previous activity events, the method being implemented by computer means and comprising the steps consisting in particular of:

[0015] - obtain a primary dataset of activity events over a period global temporal prior to the target event from at least one primary data source; - define at least one sub-period of the previous overall time period; - determine a secondary dataset representative of the individual's activity over at least a defined sub-period, based on at least a subset of data from the primary dataset; - determine a primary set of indicators representative of the individual's activity over at least one sub-period, preferably over each sub-period, based on at least a subset of data from the primary dataset and / or a subset of data from the secondary dataset; - provide an optimized set of features to a prediction model configured to establish the prediction of the individual's fitness state, the features being representative of the individual's activity (10, 11) over the previous global time period and selected from the set consisting of the primary data set, the secondary data set and the primary set of indicators; - to obtain, through the prediction model, the prediction of the individual's fitness level.

[0016] An activity event is an event during which an individual's physical and regenerative capacities have been, are, or will be mobilized beyond the activity Everyday physical activity. An activity event is, for example: a training session, a match, a performance test, a rehabilitation session, or a medical event such as an operation, an injury, or a rehabilitation session. The activity event can be tracked using dedicated sensors such as motion and / or performance sensors and / or described, for example, using a form.

[0017] Naturally, the individual has a medical history and may be equipped with motion and / or performance sensors during the activities in which they participate. Their administrative profile and their feelings also constitute important data, particularly for the medical and / or technical staff, in order to best support their progress.

[0018] According to the invention, the individual can be an animal such as a horse or a human. Preferably, the individual is an athlete. Most preferably, the individual is a football player.

[0019] The method according to the invention allows a group of individuals or an individual during their individual practice, as well as their medical and technical entourage where it exists, to monitor the activity events carried out and to obtain a short and medium term forecast of the state of fitness of the group of individuals or of the individual, in particular a forecast correlated to a target event that is likely to occur in the days following the previous overall time period, preferably the day following said previous overall time period.

[0020] Successive corrective measures, for example during or after each new activity event, can be implemented to adjust the forecast if it appears too far from a target or even negative if a risk of injury is detected. These corrective measures can thus be implemented in order to: limit the risk of injury, optimize training programs, or improve the individual's techniques related to the targeted activity.

[0021] In order to implement the method, at each activity event at least one individual is equipped with motion and / or performance sensors and / or describes the event in order to collect the primary data related to that event in its entirety. Preferably, the primary data are continuous data.

[0022] Advantageously, the data constituting the primary dataset, as well as all the data obtained in the following stages of the process, are time-stamped and / or labeled in order to easily identify the date, time and nature such as: training, match, performance test, reconditioning session, operation, injury, rehabilitation session.

[0023] Obtaining the primary dataset can be done from at least one primary data source. The primary data source is preferably chosen from: a user input interface, for example, for a descriptive form; a storage device belonging to the system containing recorded data; at least one sensor capable of providing primary data of interest, such as a GPS and / or GNSS sensor; an inertial measurement unit including accelerometers; and an oximeter. GPS / GNSS sensors advantageously allow for the combined monitoring of speed, distance traveled, tilt, and altitude.

[0024] Thus, through its history and then through the activity events it experiences, the individual accumulates data forming the primary set of activity event data, and including, for example: - motion data including: speed, accelerations, decelerations, distances; - performance data including: heart rate, perspiration, lactate level, temperature, humidity, pressure, oximetry; - declarative data including: medical data such as a history of past or current injuries or a feeling of pain or discomfort, medical test results, administrative data such as age, position in the team or on-site status, RPE (Rating of Perceived Exertion) data.

[0025] The primary dataset is obtained over an overall time period prior to the target event, preferably immediately prior to the target event, even more preferably immediately prior to the target event and of a duration of at least 21 days.

[0026] In variants in "degraded" mode of the invention, the overall prior time period may be less than 21 days and more than 7 days, or better yet less than 21 days and more than 11 days.

[0027] The individual and / or the medical and technical environment defines the overall prior time period over which this primary data is collected.

[0028] The method can be iterated as many times as necessary during a chosen sporting period. For this reason, the method is advantageously applied to rolling time periods. Indeed, a target activity event for which a prediction has been made will then occur, and subsequently become a previous activity event that will be taken into account for establishing a prediction for a new future target activity event.

[0029] In order to obtain predictions with improved accuracy compared to prior art processes and systems, the overall prior time period is advantageously divided into time sub-periods, preferably into three sub-periods.

[0030] These sub-periods are advantageously defined by the individual's medical and / or technical support network. They allow for the study of overall activity and its impact on the individual in the short, medium, or long term, either separately or in correlation with one another. These sub-periods can be defined, for example, by input or by selection from a list of choices, via an interface.

[0031] In a preferred embodiment, the first time sub-period corresponds to the seven days prior to the target activity event, the second time sub-period corresponds to the eleven days prior to the target activity event, and the third time sub-period corresponds to the ten days prior to the second time sub-period.

[0032] Subsequently, a secondary dataset will be determined. This secondary dataset is representative not of the activity event in its entirety, but of the individual's activity itself, that is to say, data presenting added value for the study of the activity event in question.

[0033] Secondary data are determined, i.e. identified and / or calculated, over at least one sub-period, on the basis of at least a subset of data from the primary dataset, preferably on the entire dataset from the primary dataset.

[0034] Preferably, the secondary dataset includes, over at least one sub-period, and more preferably over each sub-period: - the sum of explosive efforts; - the number of sprints; - biomechanical data such as deviations, imbalances, asymmetry of ground support; - the total distance travelled including: distance travelled between 6 and 15 km / h, distance travelled between 20 and 25 km / h and distance travelled at more than 25 km / h.

[0035] It is then easy to understand why prior art solutions, which only take into account a limited number of parameters relating exclusively to the physical activity of the individual, do not allow for the effective prediction of fitness level, the risks of injury, or remain limited to the provision of generic recommendations.

[0036] Indeed, the set consisting of movement data, performance data and the secondary dataset are raw or quasi-raw data, often redundant and in all cases not very intuitive in the information they convey regarding the individual's activity and in the detailed evaluation of their impact on him.

[0037] In order to give a deeper analytical dimension to this data, the process according to the invention provides for the integration of multi-source data and in particular takes into account the entire medical and psychological aspect of the individual.

[0038] Thus, the method includes a step of determining a primary set of indicators representative of the individual's activity over at least one sub-period, preferably over each sub-period, based on at least one subset of data from the primary dataset and / or a subset of data from the secondary dataset. The primary indicators are determined, that is, identified and / or calculated. The primary set of indicators reflects the intensity and duration of the physical effort exerted by the individual, based in particular on a specific calculation of the workload based on the distances covered at different speeds as defined by the secondary dataset.

[0039] It is recalled that from a semantic point of view, in the field of sport: - the "internal" training load represents the impact of training on the body, and is calculated from data from any means acceptable according to this approach, such as performance sensors or RPE (English acronym: for "Rating Perception Effort" questionnaires); - The "external" training load reflects the impact of the environment outside the individual's body and is calculated from movement data. GPS is widely used to quantify external load.

[0040] Preferably, the primary set of indicators includes: training workload, workload during at least one encounter such as a match or competition with one or more opponents, as well as Foster's indicators of monotony, constraint and form.

[0041] The overall workload, including the "internal" training load and the "external" training load, is used to calculate Foster's indicators which allow

[0042] to assess the individual's fitness and physical condition. Preferably, according to the method, three main Foster indicators are calculated from workload data: - the monotony indicator, which measures the daily variation in workload; - the stress indicator, which assesses the accumulation of load and associated risks; - the fitness indicator, which results from the difference between workload and stress, and is used to assess the individual's fitness level.

[0043] Advantageously, these main Foster indicators are calculated over the first time sub-period, in order to obtain a continuous view of the individual's performance and to predict the risks of overload or excessive fatigue.

[0044] Subsequently, the method according to the invention advantageously comprises a step of providing an optimized set of features to a predictive model configured to predict the individual's fitness level. The features forming this optimized set are representative of the individual's activity over the preceding overall time period and are selected from the set consisting of the primary data set, the secondary data set, and the primary set of indicators. The optimized set can be provided, for example, by user input or identified, in particular, automatically.

[0045] However, while modernizing calculation methods and distributing indicators from the primary set of indicators across several sub-periods already improves the accuracy and effectiveness of predictions and recommendations, these can be further enhanced by a more rigorous and adaptive screening of the characteristics presented as input to the prediction model, as the Applicant has observed.

[0046] Indeed, reducing the number of features to an optimal set maximizes the relevance of the data while avoiding overfitting at the level of the prediction model.

[0047] An important criterion for this selection is the correlation between the data, in order to ensure that no redundant or insignificant variable is included in the model.

[0048] To this end, the method according to the invention further advantageously includes a step of identifying an optimized set of features representative of the individual's activity over the preceding overall time period from among the set consisting of the primary data set, the secondary data set, and the primary set of indicators. The optimized set of identified features can then be provided to the prediction model.

[0049] Preferably, the step of identifying the optimized set of features includes the application of a correlation matrix to the set consisting of the primary data set, the secondary data set and the primary set of indicators.

[0050] Preferably the step of identifying an optimized set of features is carried out when the training data are modified, most preferably when the training data of the prediction model are strongly modified, for example if the new training data include a new type of primary data, in order to select the best features and / or indicators.

[0051] Advantageously, the criteria related to the correlation matrix and used to determine which features will be removed and will not be part of the primary set of indicators are as follows: - If the correlation is greater than 0.55, that is, if two features have a correlation of 0.55 or more, one of the two features is removed to avoid redundancy in the model; - If the correlation is low with the target prediction, that is, if the correlation between a characteristic and the target prediction, which is the risk of injury, is too low, this characteristic is considered irrelevant and is removed.

[0052] Preferably, the optimized set of features comprises:

[0053] - the sum of explosive forces;

[0054] - the distance travelled between 6 and 15 km / h; - the distance travelled between 20 and 25 km / h;

[0055] - deviations; - imbalances; - the distance covered at more than 25 km / h during the second sub-period; - the workload in the second part, which is an indicator of the accumulated workload

[0056] during the second time sub-period.

[0057] Consequently, the optimized set of features is provided as input to the prediction model configured to establish the prediction of the individual's fitness level, correlated to a target activity event such as the next match, training session or rehabilitation session.

[0058] Advantageously, the prediction model is a gradient amplification type machine learning algorithm. Preferably, the prediction model includes the XGBoost algorithm.

[0059] Preferably, the prediction includes the estimation of at least one element from among: - an injury risk score; - an area at risk of injury; - an individual's fitness score.

[0060] Advantageously, the prediction method according to the invention further includes a step of training the prediction model using training data such as prior data from primary data sets, secondary data sets, primary indicator sets and prior prediction data from the individual and / or individuals practicing an identical or very physiologically similar sporting activity.

[0061] In advanced variants of the prediction model, in order to compensate for possible data imbalances during training, since injuries are rare compared to other events, a balancing step may be included during the training of the prediction model, for example by oversampling minority observations, i.e. by automatically generating synthetic samples.

[0062] Preferably, the SMOTE technique (English acronym for "Synthetic Minority Oversampling Technique") is used.

[0063] Most preferably the SMOTE-ENN technique (English acronym for "Synthetic Minority Oversampling Technique with the Edited Nearest Neighbors") is used in order to combine the generation of new samples in the minority class with the elimination of noisy points, which makes it possible to rebalance the data while improving in the application targeted by the invention the ability of the model to identify in particular potential injuries.

[0064] Also in advanced variants of the prediction model, in order to optimize the model's performance, a development environment or infrastructure (or "framework") for the automatic tuning and optimization of hyperparameters can be used. Preferably, the Optuna environment is used.

[0065] The Optuna environment allows for more efficient optimization by performing an intelligent and adaptive search for the best hyperparameters by evaluating the performance of the model at each step and during previous iterations and adjusting the values ​​to maximize accuracy, i.e. the number of positives correctly predicted ("true positive") divided by the set of positives predicted ("true positive + false positive"), and recall, i.e. the number of positives correctly predicted ("true positive") divided by the set of positives ("true positive + false negative").

[0066] Most preferably, during the optimization of the model, the main hyperparameters optimized by the use of the Optuna environment are: - n_estimators: number of trees in the model; - max_depth: maximum depth of each tree; - learning_rate: learning rate to adjust the contribution of each tree; - subsample: fraction of samples used to build each tree, in order to avoid overfitting; - colsample_bytree: fraction of features used to construct each tree, in order to reduce the correlation between trees.

[0067] Once the prediction has been obtained from the output of the prediction model, the method according to the invention may further advantageously include at least one step of saving the data once the prediction has been obtained at the level of a storage means belonging to the computer means.

[0068] Of course, other safeguard steps may be planned.

[0069] Once the prediction is obtained from the output of the prediction model, the method according to the invention may further advantageously include a step of displaying, upon request and according to a user's access profile, information including at least one prediction of an individual's fitness level correlated with a target activity event.

[0070] Access to a restricted range of information based on the user's profile ensures, in particular, the confidentiality of certain sensitive information such as medical information. Furthermore, such a selection of information guarantees the provision of information that necessarily offers high added value to the user.

[0071] The invention also relates to a data processing system for managing the fitness status of at least one individual, comprising computer means, including at least one data storage means and at least one processor, configured to implement the method described above.

[0072] The computer means advantageously include at least one data storage means. This at least one data storage means may constitute a primary data source for implementing the process, by virtue of its content, which may include, for example, primary datasets. This at least one storage means may allow the data to be saved at least after the prediction has been obtained.

[0073] In a sporting context, it is particularly advantageous for the user to be able to easily access and / or share data. To this end, computing resources can be distributed locally and / or via a cloud architecture according to needs.

[0074] The computer means preferably comprise at least one interface including:

[0075] - a display means for displaying information including at least one prediction of an individual's fitness level correlated with a target activity event, and / or

[0076] - an input means allowing a user to enter primary data and / or to obtain a prediction correlated to a target event, by the method according to the invention.

[0077] According to another aspect, the invention relates to a computer program product comprising instructions which, when the program is executed by the computer means of a system according to the invention, these instructions lead the program to implement the process according to the invention.

[0078] The invention finally relates to a non-transient data carrier readable by the computer means of a system according to the invention, on which the computer program product is recorded.

[0079] It is clear from here that the invention brings a new paradigm for prediction and support in the context of the physical activity of individuals and in particular of athletes.

[0080] By way of example, it does not, for instance, employ some of the most commonly used ratios in the field, namely: - the acute / chronic load ratio or ACWR (in English: "Acute Chronic Workload Ratio" ACWR), which classically allows us to assess the risk of injury and performance capabilities over time; - the internal / external load ratio which estimates, once the external training load is finished, the psychophysiological stress suffered during training.

[0081] On the contrary, the invention allows for innovative and targeted data processing, on a larger quantitative and temporal scale. Figures

[0082] Other features and advantages of the present invention will become apparent from the description below, with reference to the attached Figures 1 and 2, which illustrate an example of an embodiment without being limiting in any way and on which:

[0083] [Fig.1]

[0084] Fig. 1 represents the prediction method according to the invention;

[0085] [Fig.2]

[0086] Figure [Fig. 2] represents a management system according to the invention. Detailed description

[0087] An embodiment of the device according to the invention is presented below with reference to figures 1 and 2.

[0088] In this embodiment, the situation presented is that of athletes 10, 11 evolving in a team such as a football team, and having a technical and medical entourage TM consisting of a coach and a doctor.

[0089] During each activity event for which it is possible, athletes are equipped with motion sensors Cl and / or performance sensors C2. The sensors Cl, C2 are capable of storing and / or transmitting the primary data related to an athlete 10, 11 and to the activity event being monitored, for example a football training session, during the step Ela of obtaining a primary data set.

[0090] Activity events that cannot be tracked by sensors C1, C2, for example because they are too old or unpredictable, such as an accident occurring in the athlete's private life 10, 11, medical test results D1, administrative data D2 such as age, team position or attendance status, and performance data D3, can be declared during the Ela step of obtaining a primary data set by the athlete 10, 11 or his entourage TM, via at least one interface Hl, H2 in order to enrich the primary data source of the management system in the form of at least one storage means Ml, M2.

[0091] The management system thus advantageously includes computer resources comprising, in the illustrated embodiment: - means of display and input H1, H2 such as a tablet or smartphone; - cloud and local storage options M1, M2; - Kl, K2 processors configured to execute the dedicated program and implement the process according to the invention.

[0092] As illustrated in [Fig. 1], the method for predicting an athlete's fitness level, correlated with a target activity event, based on previous activity events, comprises the steps of:

[0093] - obtain at step Ela a primary set of activity event data such as the speed, accelerations and decelerations, distance, stress tests and the athlete's perception 10, 11, over an overall time period prior to the target event from at least one primary data source Cl, C2, Dl, D2, D3, Ml, M2, Hl, H2;

[0094] - define at step Elb at least one sub-period of the overall time period previous;

[0095] - determine in step E2 a secondary dataset such as the sum of the efforts explosives and deviations, representative of the athlete's activity 10, 11 over at least a defined sub-period, based on at least a subset of data from the primary dataset;

[0096] - determine in step E3 a primary set of indicators such as the workload at PI training, match workload P2 and Foster indicators P3, representative of athlete activity 10, 11 over at least one sub-period based on at least one subset of data from the primary dataset and / or a subset of data from the secondary dataset;

[0097] - possibly, identify in step E4 an optimized set of features, on the basis for applying a correlation matrix to the set consisting of the primary data set, the secondary data set and the primary set of indicators;

[0098] - provide in step E5a an optimized set of features to a model of prediction configured to establish the prediction of the fitness state of the athlete 10, 11, the characteristics being representative of the activity of the individual 10, 11 over the previous global time period and selected from the set consisting of the primary data set, the secondary data set and the primary set of indicators;

[0099] - obtain at step E5b by the prediction model the prediction of the state of form of the In sports, the prediction includes, for example, estimating a score and a zone of risk of injury as well as a set of recommendations regarding the timing and nature of future activity sessions.

[0100] Once the prediction has been obtained, the management system allows any authorized user to access, on request and according to their profile, information including at least one prediction of the fitness state of at least one individual correlated to a target activity event, the display of which E6 is carried out using the display and input means H1, H2 such as a tablet or a smartphone.

[0101] Thus, by way of example and according to the associated user rights: - a homepage, accessible to all users;

[0102] - a global team page, accessible to all users; - a page accessible to the technical team; - a page accessible to medical personnel; - a summary page of the staff status, accessible to all users.

[0103] Preferably, the home page does not contain direct medical data. It displays indicators of workload in match, workload in training, risk of injury, fitness status.

[0104] Preferably, the team's global page provides an overview of the current roster, including the club's standings and schedule. It facilitates the management of players ranked 10 and 11 during periods of interest, with the ability to add or remove players ranked 10 and 11.

[0105] Preferably, the page dedicated to the technical entourage TM displays information specific to each athlete 10, 11 including administrative information such as: photo, age, position, percentage of use in match, information regarding data on workload in match, workload in training, risk of injury and fitness status, information relating to his status including his availability or unavailability, his GPS statistics of the last match.

[0106] Preferably, the page dedicated to the medical entourage TM includes the same information as that dedicated to the technical entourage TM and also includes graphs such as a star graph indicating the risk of injury according to the different areas of the body, a model of the human body showing the location of injuries that have occurred in the athlete 10, 11 as well as access to medical documents and injury history.

[0107] Preferably, the team status summary page is accessible to all users and displays lists of athletes 10, 11 who are training normally, athletes 10, 11 in rehabilitation, and athletes 10, 11 who are injured or ill. It allows, in particular, for monitoring athletes at highest risk without disclosing the exact nature of their injuries.

[0108] The technical and medical TM team can thus apply the recommendations received and / or base themselves on them in order to implement corrective measures F, for example during or after each new activity event, in order to adjust the forecast if it appears too far from a target.

[0109] It is easily understood that this detailed description relates to a particular example of the realization and implementation of the present invention, but that in no way does this description have any limiting character to the object of the invention; on the contrary, its objective is to remove any possible imprecision or misinterpretation of the following claims.

[0110] Thus, it will be understood that the example of sporting activity, the number of participants and the nature of the data collected, processed and obtained can vary considerably depending on the needs and the intended application.

[0111] For example, in a more modest embodiment not shown, the invention can be implemented by an athlete individually, for example by a runner, using at least one sensor which he equips himself with during his sports sessions, the management system being integrated into an application on his smartphone.

[0112] Similarly, the reference signs placed in parentheses in the following claims are in no way intended to be limiting; these signs are solely intended to improve the intelligibility and understanding of the following claims and the scope of the protection sought.

Claims

Demands

1. A method for predicting the fitness level of an individual (10, 11), correlated with a target activity event, based on previous activity events, the method (1) implemented by computer means (M1, M2, K1, K2, H1, H2), comprising the steps of: - obtaining (E1a) a primary dataset of activity events over an overall time period prior to the target event from at least one primary data source; - defining (E1b) at least one sub-period of the overall prior time period; - determining (E2) a secondary dataset representative of the activity of the individual (10, 11) over at least one defined sub-period, based on at least one subset of data from the primary dataset;- determine (E3) a primary set of indicators representative of the activity of the individual (10, 11) over at least one sub-period on the basis of at least one subset of data from the primary dataset and / or a subset of data from the secondary dataset; - provide (E5a) an optimized set of features to a prediction model configured to establish the prediction of the fitness state of the individual (10, 11), the features being representative of the activity of the individual (10, 11) over the previous overall time period and selected from the set consisting of the primary dataset, the secondary dataset and the primary set of indicators; - obtain (E5b) by the prediction model the prediction of the fitness state of the individual (10, 11).

2. Method according to claim 1, further comprising a step of saving the data once the prediction has been obtained at the level of a storage means (M1, M2) belonging to the computing means (M1, M2, K1, K2, H1, H2).

3. A method according to the preceding claim, wherein the step of obtaining (Ela) the primary data set is carried out from at least one primary data source among: at least one motion sensor (Cl) capable of transmitting physical data, at least one performance sensor (C2) capable of transmitting physiological data, at least one data accessible at the level of a storage means (M1, M2) belonging to the computer means (M1, M2, K1, K2, H1, H2).

4. A method according to any one of the preceding claims, wherein the prediction model is a gradient amplification model.

5. A method according to any one of the preceding claims, further comprising a prediction model training step in which the primary and secondary data sets and the indicator sets are oversampled and undersampled.

6. A method according to any one of the preceding claims, further comprising an identification step (E4) of the optimized set of features based on the application of a correlation matrix to the set consisting of the primary data set, the secondary data set and the primary set of indicators.

7. A method according to any one of the preceding claims, wherein a first time sub-period corresponds to a time sub-period immediately prior to the target activity event, a second time sub-period corresponds to a time sub-period immediately prior to the target activity event greater than the first time sub-period, and a third time sub-period corresponds to a time sub-period immediately prior to the second time sub-period.

8. A method according to any one of the preceding claims, wherein the first time sub-period corresponds to the seven days prior to the target activity event, the second time sub-period corresponds to the eleven days prior to the target activity event, and the third time sub-period corresponds to the ten days prior to the second time sub-period.

9. A method according to any one of the preceding claims, wherein the prediction includes the estimation of at least one element among: an injury risk score, an injury risk zone, an individual's fitness score.

10. A method according to any one of the preceding claims, further comprising a display step (E6) on request and according to the access profile of a user of information including at least the fitness status of the individual (10, 11).

11. A data processing system for managing the fitness status of at least one individual, comprising computing means (M1, M2, K1, K2, H1, H2), including at least one data storage means (M1, M2) and at least one processor (K1, K2), configured to implement the method according to any one of claims 1 to 1H

12. 1U. Data processing system according to the preceding claim, wherein the computing means (M1, M2, K1, K2, H1, H2) further comprise at least one interface (H1, H2) including: - a display means (H1, H2) for displaying information including at least one prediction of the fitness level of an individual (10, 11) correlated to a target activity event, and / or - an input means (H1, H2) allowing a user to enter primary data and / or obtain a prediction correlated to a target event, by the method according to any one of claims 1 to 10.

13. Product computer program comprising instructions which, when the program is executed by computer means (M1, M2, K1, K2, H1, H2) of a system according to any one of claims 11 to 12, cause the system to implement the method according to any one of claims 1 to 10.

14. Non-transient data carrier readable by computer means (M1, M2, K1, K2, H1, H2) of a system according to any one of claims 11 to 12, on which is recorded the computer program product according to the preceding claim.

Images