GR20170100068A - A system and method meant for the real-time follow-up and prognosis of the physical and clinical condition of athletes - Google Patents

Abstract

Novelty: an innovative and smart system for the real-time prognosis and rating of the athlete’s and athletic team’s performance is disclosed. Advantages: the invented system is able to evaluate and display, in real time, the performance of an infinite number of athletes on a common internet-based platform. This worldwide classification system provides real-time information on the history files, physical condition, health and performance of athletes. Ti can be utilized, inter alia, for the individualized evaluation analysis, the follow-up and the report giving information on the whole physical condition, the exercising load, the abilities, the nutrition, the wellbeing, the recovery from wounding and the performance of athletes having the same characteristics. This standardized assessment method aims at the classification of individualized results, the follow-up of the efficiency of the training programs, the improvement of the exercising techniques in use and the better performance of the athletes. Purpose: to provide objective and reliable classification of athletes in a present or future time, to evaluate their physical condition in a precise period of time and to continuously boost competition in a worldwide scale.

Description

PHYSICS MONITORING AND FORECASTING METHODOLOGY AND SYSTEM /

CLINICAL STATUS AND SPORTS PERFORMANCE OF REAL USERS

TIME

RECORD

The use of small-sized sensors, controllers and routers has enabled in recent years the remote recording and processing of some key training and fitness / health parameters in order to periodically monitor the performance of athletes. Online systems have been developed to monitor athlete performance for specific sports, training and exercise effectiveness, injury prevention, and improving their training program. In some examples, performance tests are applied to assess the performance of athletes and provide information about the performance of many athletes instead of just one.

So far, there is no common platform/system for many athletes of various sports that can be compared with each other and have an objective ranking on a global basis, based on their sports performance in a certain time, and various performance indicators, especially in predicting the athlete's future performance.

DESCRIPTION

A real-time personalized sports performance and health status monitoring and prediction system based on a real-time sports scoring method by normalizing and more accurately comparing the overall sports performance of athletes with similar characteristics in a single online scoring platform. The results from each test / training program / official event for a given athlete are normalized in real-time by comparing the test results to a database that provides the distribution of test results among a class of similar athletes and then assigning to each test, a grade based on the percentile of the test result among the distribution of those test results. By combining the results from each test, with given weights, a single graded score is produced for each athlete that represents overall athleticism, according to the sport.

There is no system that evaluates athletes on actual performance in current and future time. The described system evaluates and ranks athletes of similar sports and characteristics around the world based on an objective scoring mechanism. Athletes compare themselves to millions of other athletes, and increase their ability and performance. In reality, athletes are compared, not only in terms of their personal performance over time, but also among similar athletes around the world, leading to constant and increasing competition.

The innovation lies not only in the performance evaluation methodology but mainly in the predictive engine that is trained to predict future performance. The global ones

PHYSICS MONITORING AND FORECASTING METHODOLOGY AND SYSTEM /

CLINICAL STATUS AND SPORTS PERFORMANCE OF REAL USERS

TIME

rankings are based on future projected performance built with deep learning techniques and generated by the scoring engine with dynamic predictive updates. The training algorithm uses the tools of Neural Networks, Multilinear Programming and Simulation, combined with Bayes and Logistic Regression.

The system consists of a multi-sensor wireless mobile device, embedded controller, memory, wireless network transceiver, low power wide area gateway, cloud computing server, intelligent processing unit and the web platform.

The methodology of the system is analyzed as follows:

A) The user / athlete provides the personal information necessary to create a personalized profile and personalized test performance pattern through the online platform

B) The personalized test performance standard is created based on sports science, personal requirements / goals and state-of-the-art standardized test suites recognized by the scientific and sports community

C) A small portable device, allows the continuous measurement of specific parameters, in real time, during the daily trainings/competitions where data/signals (geographic, biometric, environmental, psychological and/or movement) are recorded and transmitted, in real time , to the cloud computing server, through the low definition network (WAN) gateway and a protocol that enables machine-to-machine communication.

D) The data aggregator collects data / signals from external interconnected systems and forwards them to the integrated processing unit for performance evaluation, hosted on the cloud computing server.

E) The processing unit, consisting of the intelligent analysis engine and THE predictive model, processes the recorded data together with the personal information, using the integrated intelligent algorithms in the cloud computing server.

F) Indicators are calculated and displayed on the web interfaces, for each category / performance group as described in the performance standard for each sport, which is personally customized for each athlete.

G) The distribution of test result data for athletes of similar characteristics is determined using the empirical cumulative distribution function, followed by meta-analysis to increase the total sample size resulting in the normalization of each athlete's performance and presentation on the online platform of scores for each athlete compared to other athletes with the same characteristics, both in

PHYSICS MONITORING AND FORECASTING METHODOLOGY AND SYSTEM /

CLINICAL STATUS AND SPORTS PERFORMANCE OF REAL USERS

TIME

current and future time, along with many other grouped metrics, by athlete, by sport, by search criteria.

H) A unique score is produced, in real time, for each athlete for each performance group (i.e., biometrics) at a given time ti and a predictive score at a given time ti+n, evaluating the athlete's future performance based on present performance and the individualized "dynamics" as calculated and simulated by the predictive model, based on a multivariate weighted function according to the type of sport.

I) Overall athletic performance results are displayed on the online platform, where athletes are automatically ranked and ranked according to their athletic performance results within the athletic performance of the controlled environment tests.

The online platform allows each athlete to enter personal information, profiles, records, projected work / achievements, periodic data on nutrition, fitness, performance, upcoming events, etc. The sensor data received from the remote devices they cannot be edited, preventing deliberate data falsifications. However, the athlete will be able to upload other information, starting from the time of registration on the platform, or valid published data such as 100m race results in a national championship.

The second part of the online system includes the grading system, where the input data is processed by multivariate and combination algorithms aimed at several measurements, classified into groups. Global rankings are based on deep learning techniques and predictive performance models (scoring engine with dynamic predictive updates). These metrics (performance indicators) added together with their weighting factors create a single Athlete Score, updated in real time. The incoming data is analyzed with wrapper methods and trained with Neural Networks and MLP tools.

The overall score for an athlete is generated based on a multivariate weighted function depending on the type of sport. OL athletes are compared to each other from around the world based on real accurate data estimates and receive realistic predictions for their attribute performance. The performance tests are further analyzed and displayed consisting of more than 40 independent metrics, per search criteria.

The sensors are integrated into the wearable multi-sensor device of small size that can be strapped around the leg and/or chest or attached directly in the form of discrete sensors to the athletes' clothing or the attached accessories consist of at least the following:

■ Accelerometer

PHYSICS MONITORING AND FORECASTING METHODOLOGY AND SYSTEM /

CLINICAL STATUS AND SPORTS PERFORMANCE OF REAL USERS

TIME

• Gyroscope

• Barometer

• RTK GPS

<•>3-12 lead electrocardiogram

<•>EMG sensor

<•>Photowasher

<•>Temperature Sensor

<•>Audio / Video Recorder

• Wireless transmitter

The basic system is structured on standardized sets of state-of-the-art tests recognized by the respective scientific and sports community: Fatigue / Performance Index (Internal and External Load), Plytometric Tests and evaluation of clinical and psychological status. The four main measurement groups (health, fitness, psychology, skills/abilities) consist of more than 40 measurements, which were evaluated with 3 thematic tests and consist of 5 categories of data (geographic, biometric, environmental, psychological and / or movement) . The metrics are collectively analyzed based on pre-defined importance weights and interconnected rules of association, to result in a single sports score. At the same time, these metrics can be displayed separately by search criteria, for more information, by athlete, by team, by sport, for a given period of time, especially useful for athletes, coaches, teams, bettors, sponsors, etc.

Performance tests are further broken down below (but not limited to) into performance indicators:

i. Fatigue related tests. Endurance, strength, duration, exercise frequency and intensity, intensity, training volume, body composition and muscle type, HR to RPE ratio, perception of effort, monitor of consciousness (RPE) score, HR recovery (HRR), etc.

ii. Plyometric Tests. Reverse locomotion / jump, sprint performance, isokinetic and histometric dynamometry, average power, peak velocity, peak force, jump height, time of flight, contact time and rate of force development, etc.

III. Assessment of clinical and psychological status. Heart rate, heart rate variability, oxygen saturation, ECG measurement, breathing rate, blood pressure, fat, happiness, anxiety, mental approach, determination, stimulation, etc.

The intelligent processing engine consists of the intelligent analysis engine and the predictive model that produces metadata for historical data and forecasts for

PHYSICS MONITORING AND FORECASTING METHODOLOGY AND SYSTEM /

CLINICAL STATUS AND SPORTS PERFORMANCE OF REAL USERS

TIME

future performance based on predictors and neural networks with some hidden data layers and/or Bayesian probability models. Multivariate functions are treated as combinatorial problems, with weights of specific importance for each variable as estimated by the binding rules.

The integrated intelligent algorithms within the predictive model are trained to simulate with dynamic and probabilistic models the exact possible performance of an athlete or team / group of athletes in specific conditions. The abundance of past records is very important at this point to train the algorithms accurately, towards building a reliable predictive personalized engine. These records can be obtained, in real time, during or after routine training, sporting events, past events, etc. The ability to dynamically update both personal records and performance metrics of a single athlete in comparison to others simultaneously provides important real-time information about an athlete's current athletic status as well as their estimated performance at a given time ti under specific conditions.

Each athlete's personalized profile is based on deep learning techniques and algorithms that are trained and updated each time new records are entered into the athlete's profile, allowing real-time tracking of personal activities and tracking of personal goals. The intelligent analysis engine of the intelligent processing module, with the help of the integrated simulation module, creates personalized limits for each athlete based on his personal characteristics. performs dynamic simulation and produces warnings of impending injuries, fitness deficiencies, overtraining or the rate of improvement in specific indicators and recovery from injuries.

The personalized profile helps to further develop a recovery and training program and develops a personalized risk detection mechanism for each athlete, based on the sport, previous injuries, existing injury recovery data, effectiveness of past training programs, body response to drugs, etc.

Finally, the data aggregator is designed to collect data from other interconnected systems and automatically input data into each athlete's profile. In this way, an athlete can, not only enrich their history files, but also use other systems / applications with a compatible format (i.e. physical activity, dietary, electronic health records, social applications, etc.) to automatically transfer data within the online platform.

Claims (10)

METHODS AND SYSTEM FOR MONITORING AND PREDICTING USERS' PHYSICAL / CLINICAL STATUS AND SPORTS PERFORMANCE IN REAL TIME CLAIMS 1. A personalized web-based real-time monitoring and prediction system of users' sports and physical performance, supported by a wireless sensor device including: a processor, a memory, a wireless network transceiver for communication and data transfer to the cloud server, and an array of interconnected sensors: • Accelerometer • Gyroscope • Barometer • RTK GPS • ECG • EMG • Photoplothysmograph • Temperature recording • Image Capture • Audio Recording 2. The system of claim 1, wherein the sensor device can be strapped around the leg and/or chest or attached directly to the body, in the form of discrete sensors on the athletes' clothing or wearable accessories. 3. The system of claim 1, wherein the sensor device can receive signals and data from the connected external devices, such as: spirometer, oximeter, blood pressure sensor, pedometer, body weight and body fat meter, biofeedback device and exercise devices / training. 4. The system of claim 1, wherein the sensor array supports audio and video files and enables voice interaction with the user, image and audio/voice detection and recognition for analysis, processing and feedback. 5. The system of claim 1, producing predictive scores for athlete performance wherein the method comprises: a) A small-sized wearable device, which allows the continuous measurement of specific parameters in real time during the daily exercise of athletes or in official competitions b) The collection of data wirelessly, in real time, through an innovative protocol that allows machine-to-machine communication c) The collection of the data by the collector from external interconnected systems and their promotion to an integrated analysis article for personalized performance evaluation (average power, normalized power, speed, accelerations, etc.) and personalized profile built on learning techniques and predictive models METHODOLOGY AND SYSTEM FOR MONITORING AND PREDICTING PHYSICAL / CLINICAL STATUS AND SPORTS PERFORMANCE OF USERS IN REAL TIME d) The determination of the distribution of the test results of the data for athletes with similar characteristics, using the empirical cumulative distribution function, after a meta-analysis to increase the total sample size, e) The overall score for an athlete who determined based on a multivariate weighted function, depending on the type of sport. 6. The system of claim 4, wherein the embedded analysis component builds trend performance prediction models (scoring engine with dynamic predictive updates and predictive models) based on learning techniques and the input data is analyzed and trained with neural networks NN, MLP (multi linear programming) and simulation tools, which allow dynamic updating of scoring with predictive coefficients. 7. The system of claim 4, wherein the integrated analysis component normalizes the results from each test for a given athlete by comparing the test results in a database between a similar class of athletes and between the results of previous tests of the same athlete and then assigns to each result a number / grade based on said value system. 8. The system of claim 4, wherein the integrated analysis module combines the scores from each test, with given weights, and creates a single score for each athlete representing their overall performance. 9. The system of claim 4, wherein the embedded analysis component receives signals from variables such as: duration, frequency and intensity of exercise, training load, body composition and muscle type, heart rate (HR) , the HR / RPE ratio, the perception of effort, Session Rating of Perceived Exertion (RPE), HR Recovery (HRR), average force, maximum speed, maximum force, jump height, flight time, contact time, rate of force development, the performance intensity, performance pressure, pre-stretch/half-seat jump, sprint performance, heart rate variability, stress, anxiety, motivation, environment, psychological state, pace, posture, speed, distance traveled, co-activation EMG and EMG shape, etc. and to assess isokinetic and isoinertial dynamometry, the person's fatigue, the person's fitness level or a combination of these, and not only. 10. The system of claim 4, wherein the embedded analysis component defines the personalized thresholds based on the results of the variables / indicators / and personal information as indicated by the user's profile, and generates appropriate messages and recommended actions to optimize the performance and fitness.