Athlete total factor evaluation system based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent exercise evaluation, and provides an artificial intelligence-based athlete total element evaluation system, which comprises a data acquisition device for acquiring multi-element data of an athlete; the data preprocessing and analysis module is used for preprocessing the collected multi-element data, extracting key features for different types of data and performing integration processing to form a fusion feature vector containing multiple elements; the AI comprehensive evaluation module is based on the fusion feature vector, and an artificial intelligence algorithm is used for integrating all elements to obtain an evaluation result of the athlete; and the result output and feedback module provides the result to a coach, an athlete or a manager in a visual form for decision support. The overall architecture of the system adopts a layered modular design, comprehensive, objective and scientific comprehensive evaluation of athletes is realized through deep fusion of an artificial intelligence technology, and accurate data support and decision basis are provided for coaches, athletes and managers.

Description

Technical Field

[0001] This invention relates to the field of intelligent sports evaluation technology, and in particular to an artificial intelligence-based comprehensive athlete evaluation system. Background Technology

[0002] In the field of sports, how to objectively and comprehensively evaluate athletes has always been a key concern for coaches and researchers. Traditional methods often rely on a few physical fitness tests or competition results for horizontal comparisons, which are insufficient to cover all aspects of an athlete's abilities. With the development of information technology, attempts to use data analysis for athlete evaluation have begun both domestically and internationally. However, existing technological solutions still have many shortcomings. Most focus on a single dimension or specific scenario, lacking a comprehensive consideration of daily training data, technical performance, and psychological factors. Their comparison models are mostly used for horizontal comparisons between multiple athletes, and are not suitable for longitudinally tracking the comprehensive development of a single athlete. A comprehensive intelligent evaluation system integrating multiple factors has not yet been formed. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide an artificial intelligence-based comprehensive evaluation system for athletes.

[0004] The technical solution of this invention is implemented as follows: This invention provides an artificial intelligence-based comprehensive athlete evaluation system, including data acquisition equipment, a central server with at least a data preprocessing and analysis module, an AI comprehensive evaluation module, a result output and feedback module, and a central control module, and a user terminal, wherein: Data acquisition equipment is used to collect multi-element data of athletes, including physical and physiological data, technical movement data, psychological data, and environmental and situational data, and uploads them to the central server in real time through a unified interface protocol. The data preprocessing and analysis module is used to preprocess the collected multi-element data and extract key features for different types of data. It also uses multimodal data fusion technology to integrate key features from different channels and types to form a fusion feature vector containing multiple elements. The AI ​​comprehensive evaluation module includes: a model building unit, used to train a comprehensive evaluation model on historical athlete data using supervised learning regression or classification models to learn the nonlinear relationship between various elements and athletic performance; and an analysis and evaluation unit, used to input the fused feature vector into the comprehensive evaluation model for calculation and evaluation, output the athlete's comprehensive ability score and determine the level. The results output and feedback module includes: an evaluation report generation unit, which automatically generates an evaluation report based on the evaluation results, including the athlete's overall ability score, scores of each sub-indicator, and strengths and weaknesses diagnosis; and a personalized training suggestion unit, which generates personalized training plans and recovery arrangements based on the athlete's specific situation and sends the results to the user terminal. The central control module is used to control and coordinate the normal operation of each module.

[0005] In some embodiments, the central server is further deployed with a data storage module for classified storage management according to data type: continuous physiological sensor data is stored in a time-series database, video and image data are stored in a media file library, and psychological test results are stored in a relational database.

[0006] In some embodiments, the result output and feedback module further includes: The intelligent report interpretation and interaction unit is used to present the complex data and information in the evaluation report to users in an intuitive form using natural language processing and multimodal interaction technology, and supports users to interact with the athlete's full-element evaluation system through voice, text, or images.

[0007] In some embodiments, the result output and feedback module further includes: The psychological counseling service unit is used to build digital human virtual mentors. Combining psychological theories and clinical practice experience, it uses knowledge graph technology to build a psychological knowledge base to provide intelligent knowledge support and decision-making assistance for digital human virtual mentors. It enables interactive communication with athletes through natural language processing and affective computing, and provides personalized psychological guidance and intervention.

[0008] In some embodiments, the multi-factor data includes at least heart rate, blood oxygen, movement speed, acceleration, strength, psychological state information, climate conditions, altitude, venue conditions, opponent level, and match and training videos.

[0009] In some embodiments, the data preprocessing and analysis module includes: The preprocessing unit is used to clean, denoise, and normalize all raw data. The physical and physiological analysis unit is used to calculate the athlete's maximum heart rate, resting heart rate, and heart rate recovery index based on heart rate data; to calculate sprint speed, acceleration capacity, and fatigue index based on exercise speed and acceleration data; and to evaluate explosive power and strength endurance based on force sensor data. The technical motion analysis unit is used to extract the motion features of athletes in videos through posture estimation algorithms, and compare the motion features with a standard model to evaluate the standardization and proficiency of the movements; the motion features include at least joint trajectories and posture angles; The psychological state analysis unit is used to indirectly infer psychological state by combining the fluctuation characteristics of athletes' physiological data, and to analyze stress and fatigue levels through heart rate variability indicators. The environmental impact analysis unit is used to extract temperature, humidity, and wind speed indicators from meteorological data; The feature fusion unit is used to integrate the key features extracted from the four dimensions of physical fitness, technical movements, psychology, and environment to form a fused feature vector.

[0010] In some embodiments, the analysis and evaluation unit uses minimum-maximum (Min-Max) normalization to map each indicator in the fused feature vector to the 0-1 interval, and sets reasonable upper and lower limits for indicators with clear good or bad directions. After quantification and normalization, all indicators in the fused feature vector are transformed into dimensionless standardized indicators. For different sports and evaluation purposes, the following linear weighted model is used to calculate the comprehensive ability score S: Where n is the number of index items, and i is a natural number from 1 to n. w is the score of the i-th standardized indicator output by the comprehensive evaluation model. i It is the weight of the i-th standardized indicator, and .

[0011] In some embodiments, in the linear weighted model, four categories are set with weights of 0.4, 0.4, 0.1, and 0.1 respectively for physical fitness score, technical score, psychological score, and environmental score. Within each category, each indicator factor item is further subdivided and assigned a certain secondary weight. The final comprehensive ability score is obtained by hierarchical weighted summation.

[0012] In some embodiments, the evaluation report generation unit utilizes an expert knowledge base to construct a sports-related knowledge graph, and combines formative evaluation based on dynamic tracking with summative evaluation based on phase summaries to provide athletes with personalized growth path planning.

[0013] In some embodiments, the personalized training suggestion unit classifies athletes into beginner, intermediate, or advanced levels based on the assessment results, and formulates personalized training plans and recovery arrangements based on the sports science knowledge base, combined with the athlete's level and specific weaknesses; at the same time, it introduces decision rules based on domain knowledge, dynamically adjusts the training plan according to the athlete's real-time status, and judges and handles specific situations.

[0014] The beneficial effects of the technical solutions provided in this application include at least the following: This invention incorporates all key athlete-related factors into its evaluation scope, including real-time physiological and athletic data, technical details, psychological state assessments, and environmental influences. Through a data fusion platform, the system can aggregate and correlate these diverse data sources, providing a comprehensive, one-stop evaluation perspective. Furthermore, the system fully integrates AI technology into the evaluation process, going beyond mere data collection or simple statistics. Especially in the comprehensive evaluation stage, as data accumulates, the accuracy and precision of the system's evaluation continuously improve, enabling the identification of correlation patterns that are difficult to discover using traditional methods. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 A functional block diagram of an AI-based athlete comprehensive evaluation system provided in an embodiment of the present invention; Figure 2 A schematic diagram of the system architecture of an AI-based athlete comprehensive evaluation system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the workflow of an AI-based athlete comprehensive evaluation system provided in an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0018] It should be noted that the terms "first, second, and third" used in the embodiments of the present invention are only used to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, and third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of the present invention described herein can be implemented in an order other than that illustrated or described herein.

[0019] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which these embodiments of the invention pertain. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0020] Figure 1 A functional module block diagram of an artificial intelligence-based athlete comprehensive evaluation system provided in this embodiment of the invention is shown below. Figure 1 As shown, it includes: a data acquisition device 10, a central server 20 including at least a data preprocessing and analysis module 201, an AI comprehensive evaluation module 202, a result output and feedback module 203, and a central control module 204, and a user terminal 30, wherein: The data acquisition device 10 is used to collect multi-element data of athletes, including physical and physiological data, technical movement data, psychological data, and environmental and situational data.

[0021] The data preprocessing and analysis module 201 is used to preprocess the collected multi-element data and extract key features for different types of data. Using multimodal data fusion technology, key features from different channels and different types are integrated and processed to form a fusion feature vector containing multiple elements.

[0022] AI comprehensive evaluation module 202 includes: a model building unit, used to train a comprehensive evaluation model on historical athlete data using supervised learning regression or classification models to learn the nonlinear relationship between various elements and athletic performance; and an analysis and evaluation unit, used to input the fused feature vector into the comprehensive evaluation model for calculation and evaluation, output the athlete's comprehensive ability score and determine the level.

[0023] The results output and feedback module 203 includes: an evaluation report generation unit, which automatically generates an evaluation report containing the athlete's comprehensive ability score, scores of each sub-indicator, and strengths and weaknesses diagnosis based on the evaluation results; and a personalized training suggestion unit, which generates personalized training plans and recovery arrangements based on the athlete's specific situation and sends the results to the user terminal 30; wherein, the evaluation results include the comprehensive ability score and its level output by the analysis and evaluation unit.

[0024] The central control module 204 is used to control and coordinate the normal operation of each module. The central control module 204 is connected to the data acquisition device 10, data storage module, data preprocessing and analysis module 201, AI comprehensive evaluation module 202, and result output and feedback module 203. It utilizes a microcontroller or controller to coordinate the operation of each module, achieving automation and intelligence in data acquisition, transmission, processing, and analysis, improving data acquisition efficiency and accuracy, and providing comprehensive data support for athlete evaluation.

[0025] In implementation, data acquisition devices 10 are deployed at the sports field to collect multi-element data from athletes in real time. Through a unified interface protocol, this data is uploaded to a central server in real time or periodically, ensuring data integrity and timeliness. This includes all key athlete-related factors in the evaluation, such as physical and physiological data (heart rate, blood oxygen, etc.), technical movement data (speed, acceleration, etc.), psychological data (psychological assessment results, etc.), and environmental and situational data (climate conditions, field conditions, etc.). Data acquisition devices include, but are not limited to, wearable sensors, video acquisition devices, motion capture systems, psychological testing applications, and environmental monitoring equipment.

[0026] Possible ways to implement multi-source data acquisition are: This invention acquires athletes' physical and physiological data, including heart rate and blood oxygen saturation, using wearable sensors. Simultaneously, it incorporates miniature biosensors to monitor blood lactate concentration, muscle electrical signals, and joint range of motion. Multi-angle cameras deployed at the sports field, combined with inertial measurement units, capture athletes' technical movement data, including posture, joint angles, speed, acceleration, and force. Physiological sensors monitor stress responses, while psychological assessment scales collect information on athletes' mental state. Environmental sensors collect environmental and contextual data, including at least climate conditions, altitude, field conditions, and opponent skill level. This embodiment of the invention comprehensively considers various aspects of athlete data, including physical, technical, psychological, and environmental factors, enabling a more holistic evaluation of athlete performance.

[0027] The original multi-element data is aggregated and uploaded to a central server or cloud platform using a high-throughput data bus, and the data is stored using a distributed database or time-series database.

[0028] The data preprocessing and analysis module 201, connected to the central control module, first performs preprocessing operations on the raw data, including cleaning, noise reduction, and normalization, to improve data quality. Then, it conducts in-depth analysis and feature extraction on different types of preprocessed data, such as physical fitness and physiological analysis, technical movement analysis, psychological state analysis, and environmental impact analysis. Finally, it integrates multi-dimensional features from physical fitness, technique, psychology, and other aspects to form a fused feature vector containing dozens of elements. This fused feature vector can comprehensively reflect the athlete's overall ability.

[0029] The AI ​​Comprehensive Evaluation Module 202 uses machine learning and artificial intelligence algorithms to comprehensively evaluate the features and indicators extracted and integrated from each analysis submodule, calculating evaluation results such as the athlete's comprehensive ability score. The comprehensive evaluation model learns from historical athlete data and then infers from new data. It uses a large amount of existing athlete data (including their multiple indicators and known comprehensive evaluations) to train the model, enabling it to learn the non-linear relationship between various factors and athletic performance.

[0030] The AI ​​comprehensive evaluation module 202 is configured with a multi-model collaborative architecture, including evaluation models and prediction models, covering different task types and complexity requirements. Through modular layering and dynamic model collaboration, it achieves adaptive evaluation of different sports. In the evaluation model, multi-layer neural network (MLNN) technology is selected. By simulating the hierarchical information processing mechanism of neurons in the human brain, it can extract high-order features from complex and multi-dimensional data and achieve accurate evaluation. Its training process is based on the backpropagation algorithm, continuously adjusting the weight parameters to minimize the error between the predicted output and the true label. In the prediction model, it learns from the data and then infers from the new data. Its essence is to find a function F such that the function value can approximate the expected evaluation result y*. In the linear case, F can be the above linear weighted sum, and in the nonlinear case, F is implemented by a complex network.

[0031] Furthermore, when using a multi-layer neural network model for machine learning, the concept of weights is implicit in the model parameters. The model will automatically learn the importance of each feature based on the training data, ensuring that the contribution of each element is reasonably set and that no single indicator dominates the result.

[0032] The results output and feedback module 203 visually displays the athlete's comprehensive ability score, level classification, strengths and weaknesses diagnosis, etc., in the form of charts and text reports, making it easy for users to quickly understand the athlete's overall performance. On the one hand, it interprets the output of the comprehensive evaluation model, assessing the athlete's various abilities and current status; on the other hand, it generates targeted training plans and recovery arrangements based on the evaluation results and built-in knowledge; finally, it sends the evaluation report and training suggestions to the user terminal for display to coaches and athletes.

[0033] In some embodiments, the central server 20 is also equipped with a data storage module for classified storage management according to data type: continuous physiological sensor data is stored in a time-series database, video and image data are stored in a media file library, and psychological test results are stored in a relational database.

[0034] To efficiently manage and retrieve large amounts of data, this module employs a distributed database or time-series database for storage management. These database systems provide high-performance, highly available, and scalable data storage services.

[0035] Continuous physiological sensor data, such as continuously changing physiological signals like heart rate, blood oxygen saturation, exercise speed, and acceleration, are stored in time-series databases. Time-series databases are specifically designed for storing time-related data and can efficiently handle the insertion, querying, and analysis of time-series data.

[0036] Video and image data, such as video recordings of matches and training sessions, and images of technical movements, are stored in a media file library. The media file library provides ample storage space and supports efficient file retrieval and access operations, allowing users to easily view and analyze video and image data at any time.

[0037] Psychological assessment results, such as data on athletes' mental state obtained through questionnaires, tests, or interviews, are stored in a relational database. Relational databases store data in tabular format, support complex data queries and operations, and facilitate the management and analysis of structured data such as psychological assessment results.

[0038] This categorized storage method enables the data storage module to efficiently manage and retrieve various types of data, providing strong support for subsequent data analysis and evaluation. At the same time, this storage method also helps improve data security and reliability, ensuring the safety and privacy of athlete data.

[0039] In some embodiments, the result output and feedback module 203 further includes: The intelligent report interpretation and interaction unit is used to present the complex data and information in the evaluation report to users in an intuitive form using natural language processing and multimodal interaction technology, and supports users to interact with the athlete's full-element evaluation system through voice, text, or images.

[0040] Here, this unit uses chart libraries (such as ECharts, D3.js, etc.) to display the evaluation results in the form of bar charts, line charts, radar charts, etc., so that users can intuitively understand the performance of athletes in various indicators.

[0041] By utilizing natural language generation technology, the evaluation results are transformed into easily understandable text reports. Interactive Q&A functionality is also provided, allowing users to query scores for specific sub-indicators, historical data comparisons, and other information based on their needs, satisfying users' demands for in-depth analysis and enhancing the user experience. The user interface is built using front-end frameworks such as React and Vue, providing a smooth user experience and rich interactive functions, enabling users to easily interact with the system.

[0042] In this implementation, an intelligent report interpretation and interaction unit was introduced into the results output and feedback module. This unit not only presents the system's evaluation results to users in an intuitive and easy-to-understand format but also provides an interactive experience, enabling users to deeply understand the evaluation results. Through this unit, users can ask questions via voice or images to understand the correct posture or training method for a particular technical action. The system understands the user's intent and provides accurate answers and suggestions. Simultaneously, it can collect user feedback on the system's output results, continuously optimizing the system's performance and inference results based on user feedback, thereby improving the system's accuracy and user experience.

[0043] In some embodiments, the result output and feedback module 203 further includes: The psychological counseling service unit is used to build digital human virtual mentors; combining psychological theories and clinical practice experience, it uses knowledge graph technology to build a psychological knowledge base to provide intelligent knowledge support and decision-making assistance for digital human virtual mentors; and it realizes interactive communication with athletes through natural language processing and affective computing, and provides personalized psychological guidance and intervention.

[0044] Here, the psychological counseling service unit is primarily responsible for providing athletes with mental health assessments, psychological support, and counseling services. By combining artificial intelligence technology and psychological theories, this unit can help athletes better manage their emotions and stress, thereby improving their competitive performance. Specific functions include: Mental health assessment: Using indicators such as heart rate variability (HRV) to assess athletes' psychological stress and fatigue levels, providing quantitative assessment results of mental health status, and providing a basis for developing personalized psychological counseling programs.

[0045] Psychological counseling: Based on the results of mental health assessments, provide athletes with personalized psychological counseling programs, such as cognitive restructuring and behavioral interventions, to help athletes adjust their mindset and relieve stress.

[0046] Online consultation: Provides a digital virtual mentor function, allowing athletes to communicate with virtual mentors, i.e. psychological counselors, in real time through text, voice, etc., to obtain professional psychological counseling services and provide timely psychological support for athletes.

[0047] In terms of technical implementation, signal processing technology and machine learning algorithms are used to analyze heart rate variability data, extract key features to assess athletes' mental health, and provide a scientific basis for psychological counseling. Combining cognitive behavioral therapy, mindfulness meditation, and other psychological theories and techniques, personalized psychological counseling plans are provided to athletes, ensuring the scientific validity and effectiveness of these plans.

[0048] This project utilizes technologies such as natural language processing, speech recognition and synthesis, and computer vision to construct a digital virtual mentor / psychological counselor. This enables real-time interaction and consultation services with athletes, providing more convenient and efficient psychological counseling. Simultaneously, a psychology knowledge base is built using knowledge graph technology to provide intelligent knowledge support and decision-making assistance to the digital virtual mentor, enhancing the professionalism and accuracy of psychological counseling services. Preferably, the constructed digital virtual mentor can accurately identify various psychological states of athletes and provide guidance and intervention using professional techniques such as cognitive restructuring and behavioral intervention. Based on each athlete's specific situation, a personalized image, tone, and interaction style are tailored for the virtual mentor, enabling it to act as a "personal coach," guiding the athlete's self-management and growth.

[0049] In this embodiment, the psychological counseling service unit can provide athletes with comprehensive and personalized feedback on evaluation results and mental health services, helping them better understand their performance, adjust their mindset, and improve their competitive state. Meanwhile, the digital human virtual mentor, designed as a purely digital intelligent agent, ensures maximum privacy and information security during communication, eliminating the psychological pressure of face-to-face interaction and establishing a trusting partnership.

[0050] In some embodiments, the multi-factor data includes at least heart rate, blood oxygen, movement speed, acceleration, strength, psychological state information, climate conditions, altitude, venue conditions, opponent level, and match and training videos.

[0051] In some embodiments, the data preprocessing and analysis module 201 includes: The preprocessing unit is used to clean, denoise, and normalize all raw data. The physical and physiological analysis unit is used to calculate the athlete's maximum heart rate, resting heart rate, and heart rate recovery index based on heart rate data; to calculate sprint speed, acceleration capacity, and fatigue index based on exercise speed and acceleration data; and to evaluate explosive power and strength endurance based on force sensor data. The technical motion analysis unit is used to extract the motion features of athletes in videos through posture estimation algorithms, and compare the motion features with a standard model to evaluate the standardization and proficiency of the movements; the motion features include at least joint trajectories and posture angles; The psychological state analysis unit is used to indirectly infer psychological state by combining the fluctuation characteristics of athletes' physiological data, and to analyze stress and fatigue levels through heart rate variability indicators. The environmental impact analysis unit is used to extract temperature, humidity, and wind speed indicators from meteorological data; The feature fusion unit is used to integrate the key features extracted from the four dimensions of physical fitness, technical movements, psychology, and environment to form a fused feature vector.

[0052] Here, data preprocessing can remove duplicate, invalid, or erroneous data, ensuring the accuracy and consistency of the data.

[0053] For the physical fitness and physiological analysis unit, the following feature analysis was performed: 1. Heart rate characteristic calculation. Dynamic estimation of maximum heart rate: Based on the Karvonen formula (220 - age), combined with adaptive correction for exercise intensity: ,in Let Y represent the change in exercise load, and Y represent age. Maximum heart rate. Resting heart rate detection: HRV feature clustering analysis of nighttime sleep data (22:00-6:00). Heart rate recovery index: jointly assessed by the heart rate drop in the first minute after exercise cessation (HRR1) and the 30-second slope (HRR30).

[0054] 2. Athletic Performance Assessment. Sprint Speed: A sliding window (3-second) peak detection algorithm combined with GPS trajectory smoothing is used. Acceleration Capacity: Third-order derivative feature extraction of acceleration is used to construct a GG map (acceleration-velocity phase space). Fatigue Index: A correlation model is established between the attenuation coefficient of the exercise power curve (using exponential fitting) and the lactate threshold.

[0055] 3. Strength Feature Extraction. Explosive Power Assessment: Calculated using peak power (P=F·v) from the force-velocity curve. Strength Endurance: Fitted to the power decay curve of repeated maximum strength (RM) tests (double exponential model).

[0056] For the technical motion analysis unit, the following feature analysis is performed: 1. 3D Pose Reconstruction: A hybrid architecture of MediaPipe and OpenPose is used, fusing joint heatmaps with a spatiotemporal attention mechanism. Joint Angle Calculation: Spatial angle calculation is achieved using the rotation matrix decomposition method (Rodrigues' formula). 2. Motion Quality Assessment: Improved Dynamic Time Warping (DTW) Algorithm: DTW-CD (Cost with Derivatives) proficiency scoring with velocity constraints: Joint modeling of sequence prediction error and motion entropy based on LSTM. Keyframe Detection: Employing the Change Point Detection algorithm to locate motion inflection points. For track and field sprinters, analyzing parameters such as stride length, stride frequency, and body tilt angle; for ball sports athletes, analyzing the trajectory, speed, and success rate of passing and shooting.

[0057] For the psychological state analysis unit, the following characteristics are mainly analyzed: 1. HRV feature engineering: Time domain features: SDNN, RMSSD, pNN50; Frequency domain features: LF / HF power spectrum ratio (implemented by wavelet packet decomposition); Nonlinear features: Poincaré cross section SD1 / SD2, sample entropy calculation.

[0058] 2. Psychological state inference: Stress level classification: A Bi-LSTM+Attention network was constructed, and the input sequence length was optimized to 120 seconds. Fatigue accumulation model: A NARX neural network was used, fusing HRV time-varying features and training load data. Ultimately, this achieves quantitative assessment of athletes' psychological preparation and stability.

[0059] For the environmental impact analysis unit, the following meteorological data processing and environmental adaptability assessment were mainly carried out. Specifically: the thermal stress index was calculated using a real-time WGBT (wet-bulb black-bulb temperature) monitoring system; air resistance compensation was calculated using DragEquation to determine the impact of wind speed on kinetic energy consumption.

[0060] The feature fusion unit first aligns multimodal features. On one hand, timestamp synchronization is achieved using interpolation and dynamic time warping to align cross-modal data. On the other hand, spatial alignment is established by creating a transformation matrix between the athlete's body coordinate system and the field coordinate system. Then, the ReliefF algorithm is used to assign cross-modal feature weights.

[0061] This embodiment enables holographic perception of an athlete's condition by constructing a four-dimensional analysis framework encompassing physiology, technology, psychology, and environment.

[0062] In some embodiments, the analysis and evaluation unit uses Min-Max normalization to map each indicator in the fused feature vector to the 0-1 interval, and sets reasonable upper and lower limits for indicators with clear good or bad directions; after quantification and normalization, all indicators in the fused feature vector are transformed into dimensionless standardized indicators; for different sports and evaluation purposes, the following linear weighted model is used to calculate the comprehensive ability score S: Where n is the number of index items, and i is a natural number from 1 to n. w is the score of the i-th standardized indicator. i These are the corresponding weights, and .

[0063] Here, due to differences in various indicator standards, the analysis and evaluation unit first normalizes the features, using Min-Max normalization to map the indicator values ​​to the 0-1 range. For indicators with a clear good or bad direction, reasonable upper and lower limits are set during normalization, such as the scoring benchmark for the 100m sprint performance. ;in, Let i be the score of the i-th standardized indicator. The maximum score is T, where T represents the athlete's actual performance in seconds. The time corresponding to a perfect score The time interval for each deduction (e.g., 0.2 seconds beyond the maximum score time is the deduction time interval). The deduction is the score for each instance (e.g., 0.5 points are deducted every 0.2 seconds).

[0064] Taking the 100-meter sprint as an example, the evaluation aims to select sprinters based on their explosive power and endurance. The indicators include: physical fitness indicators (weight 0.6, with starting reaction time, stride frequency, and maximum speed maintenance time accounting for 0.3, 0.2, and 0.1 respectively); technical indicators (weight 0.3, with arm swing coordination and stride stability accounting for 0.1 and 0.2 respectively); and psychological indicators (weight 0.1, including stress resistance). Example calculation: An athlete's standardized scores are: starting reaction time 85, stride frequency 90, maximum speed maintenance time 80, arm swing coordination 75, stride stability 88, and stress resistance 70. Therefore, the comprehensive ability score S = 85 × 0.3 + 90 × 0.2 + 80 × 0.1 + 75 × 0.1 + 88 × 0.2 + 70 × 0.1 = 83.4.

[0065] This embodiment provides a flexible weight configuration mechanism, which can adjust the weight of each indicator according to different evaluation purposes and sports events, making the evaluation more targeted.

[0066] In some embodiments, in the linear weighted model, four categories are set with weights of 0.4, 0.4, 0.1, and 0.1 respectively for physical fitness score, technical score, psychological score, and environmental score. Within each category, each indicator factor item is further subdivided and assigned a certain secondary weight. The final comprehensive ability score is obtained by hierarchical weighted summation.

[0067] Here, the specific calculation process for the comprehensive ability score is as follows: Data preprocessing: Clean the original indicators (outlier handling, missing value imputation), and select a normalization method (linear / nonlinear standardization) according to the characteristics of the indicators.

[0068] Secondary weighted calculation: Indicator score within each dimension = Σ(Indicator value × Secondary weight) Example: Fitness Score = 0.25 × Maximum Heart Rate Reserve + 0.20 × Sprint Power Index + ... Dimensional fusion calculation: Comprehensive ability score = 0.4 × physical fitness score + 0.4 × technical score + 0.1 × psychological score + 0.1 × environmental score.

[0069] This embodiment is based on the design of an athlete comprehensive ability scoring system constructed using a linear weighted model. It adopts a hierarchical weight allocation structure to ensure the scientific integration of the four dimensions and their sub-indicators.

[0070] In some embodiments, the evaluation report generation unit utilizes an expert knowledge base to construct a sports-related knowledge graph, and combines formative evaluation based on dynamic tracking with summative evaluation based on phase summaries to provide athletes with personalized growth path planning.

[0071] Here, in the core functional module of the evaluation report generation unit, the system relies on the authoritative data source of the expert knowledge base to deeply mine and integrate the professional knowledge system in the sports field, constructing a comprehensive and detailed sports knowledge graph. This knowledge graph not only covers multi-dimensional information such as sports training, competitive strategies, and physical recovery, but also realizes the association and reasoning of knowledge points through intelligent algorithms, providing solid knowledge support for subsequent evaluation work.

[0072] In terms of evaluation mechanisms, the system innovatively integrates two evaluation modes: one is formative evaluation based on dynamic tracking, which continuously collects and analyzes data by monitoring athletes' training performance, physiological indicators, and psychological state in real time, providing athletes with immediate feedback and adjustment suggestions; the other is summative evaluation based on stage summaries, which focuses on a comprehensive review and summary of an athlete's training cycle or competition stage, assessing their overall progress and areas for improvement. Simultaneously, the system analyzes the athlete's strengths and weaknesses and current state, and will flag athletes as fatigued or under excessive psychological pressure during the evaluation process.

[0073] Ultimately, the system combines the results of the two evaluation models with professional knowledge from the sports knowledge graph to create a personalized growth path plan for each athlete. This plan not only considers the athlete's current level and potential but also incorporates long-term development goals and phased tasks, aiming to help athletes achieve a dual leap in skill improvement and competitive performance under scientific guidance.

[0074] In some embodiments, the personalized training suggestion unit classifies athletes into beginner, intermediate, or advanced levels based on the assessment results, and formulates personalized training plans and recovery arrangements based on the sports science knowledge base, combined with the athlete's level and specific weaknesses; at the same time, it introduces decision rules based on domain knowledge, dynamically adjusts the training plan according to the athlete's real-time status, and judges and handles specific situations.

[0075] Here, athletes are categorized into beginner, intermediate, or advanced levels based on the assessment results, with different training program generation logics corresponding to different levels. The system combines the athlete's level and specific weaknesses to develop personalized training plans and recovery schedules, outputting assessment reports and training suggestions for coaches and athletes to reference.

[0076] In the decision-making rules, taking sprint events (100m / 200m) as an example, the manually set scoring points include: At the start, the crouching posture is used to determine the distance between the feet, the angle of the forward lean, and the position of the hands for support (excellent runners should have an angle ≤45° and reasonable spacing). During the race, arm swing coordination is used to determine the relaxation of the shoulder joint and the range of forward and backward arm swing (excellent runners should not swing their arms past their noses or backward past their hips); body posture is used to determine the angle of the torso leaning forward (excellent runners should have an angle of 5° to 10°). In the sprint, the finish line action is used to determine whether the runner actively leans forward to touch the finish line (excellent runners should not show significant deceleration upon touching the finish line). Technical deduction details: insufficient push-off at the start results in a deduction of 2-5 points; arm swing above the shoulder or lateral swaying results in a deduction of 1-3 points; premature deceleration in the sprint results in a deduction of 3-5 points.

[0077] The personalized training suggestion unit defines a rule list, encompassing machine-judged rules covering all aspects, including physical, psychological, nutritional, and motor aspects. Simultaneously, a suggestion function is defined, where suggestions are generated based on a comprehensive ability score S and other relevant features, meeting specific conditions. Input parameters include the original comprehensive ability score S, the trigger rule list, and the athlete's individual characteristics. The output of the suggestion function is structured as: urgency level + main suggestion + sub-suggestion items. Suggestions employ a priority ranking mechanism to ensure high-risk issues are addressed first: injury risk > physiological load > psychological problems > technical deficiencies > long-term development. Urgency levels are categorized into high-risk, medium-risk, and normal states.

[0078] The following is the rule set for the athlete ability evaluation system in sprint events: The first item is physiological load and fatigue management, including the following rules: Rule 1, if the fatigue index (F) > 0.8 and the recovery index (R) < 0.5 (sleep duration < 6 hours / day or stretching duration < 5 minutes / session), it is marked "Current recovery is insufficient, it is recommended to reduce training intensity and increase recovery measures"; Rule 2, if the weekly training volume increase > 10% or the single intensity > 150% of daily training, it is marked "Training load exceeds the safety threshold, there is a risk of rhabdomyolysis, it is recommended to adjust the plan to a gradual increase"; Rule 3, if the resting heart rate is > 10% of the baseline value for 3 consecutive days or the blood oxygen saturation is < 90%, it is marked "Cardiovascular system overload, high-intensity training needs to be suspended and cardiopulmonary function checked".

[0079] The second item is the standardization of technical movements, including the following rules: Rule 4, if the head lift timing during the starting phase is <25 meters (standard is 25-30 meters) and the push-off angle is <47° (standard is 47-51°), it is noted that "starting technique needs to be optimized, and it is recommended to strengthen the training of the heart muscle group and the center of gravity control practice"; Rule 5, if the ground contact time is >0.12 seconds or the stride length is <height × 1.2 times, it is noted that "stride frequency and stride length are mismatched, and targeted ground-pushing action training and hip joint flexibility improvement are required"; Rule 6, if the trunk forward lean angle is >10° or the arm swing amplitude is asymmetrical >15% during the middle of the race, it is noted that "core stability is insufficient, and it is recommended to increase plank support and anti-rotation training".

[0080] The third item is psychological and stress management, including the following rules: Rule 7, if the psychological stress index (P) > 0.7 and the training motivation score < 60 points, it is marked "excessive psychological load, it is recommended to introduce psychological counseling or adjust the training rhythm (such as interspersing relaxation runs)"; Rule 8, if the pre-race anxiety index > 0.6 and the race error rate > 20%, it is marked "need to strengthen simulated race scenario training and use breathing relaxation methods to reduce stress response". The fourth item is injury risk warning. Rule 9, if the joint load index > the threshold for 3 consecutive times, it is marked "high risk of patellar tendinitis, it is recommended to reduce the frequency of sprinting and increase swimming or cycling cross-training"; Rule 10, if the plantar pressure distribution is abnormal (such as medial pressure > 60%), it is marked "insufficient arch support, it is necessary to change running shoes or customize orthotic insoles".

[0081] The fourth item is recovery and nutritional monitoring. Rule 11: If the blood lactate clearance rate is <50% / hour or the protein intake is <1.2g / kg body weight after exercise, it is noted that "metabolic recovery capacity is insufficient, and it is recommended to supplement branched-chain amino acids and increase the intake of high-quality protein". Rule 12: If the water intake is <30ml / kg body weight for 3 consecutive days, it is noted that "there is a risk of dehydration, and urine specific gravity needs to be monitored and a hydration plan needs to be developed".

[0082] Example: The athlete's file records the athlete's personal information, including name: Li Ming (14-year-old junior high school student), training years: 2 years, weekly training volume: 48 kilometers, biological characteristics: height 1.72m, slight arch collapse, weakness: starting reaction time 0.28 seconds (standard <0.25 seconds).

[0083] Input parameters are: overall ability score S = 58; trigger rule list is ['Rule 9', 0.85, 'Injury Risk'], ['Rule 5', 0.75, 'Technical'], ['Rule 7', 0.65, 'Psychological'], ['Rule 11', 0.6, 'Recovery']; athlete personal characteristics are: "Age: 14, Weekly Training Volume: 48, Weakness: Starting Reaction Time, Body Fat Percentage: 16.5%".

[0084] The output recommendation is: "Urgency level: High risk; Main recommendation: Immediately stop the impact training and switch to swimming cross-training for 9 days; Sub-recommendation items: [Use a resistance umbrella for cadence and rhythm training, arrange for Teacher Zhang to provide guidance twice a week, add yoga relaxation, increase protein intake by 500ml of milk and 2 eggs daily, basic ability needs to be strengthened, it is recommended to add starting reaction training × 3 times / week]; Basis explanation: Comprehensive ability score 58+ triggering rules 4".

[0085] The above-mentioned AI-based athlete comprehensive evaluation system will be described below with reference to a specific embodiment. However, it is worth noting that this specific embodiment is only for better illustrating the present invention and does not constitute an improper limitation of the present invention.

[0086] like Figure 2As shown, the hardware architecture of the AI-based athlete comprehensive evaluation system provided by this invention is as follows: The left side consists of data acquisition devices, including wearable sensors (e.g., recording movement and heart rate), video acquisition devices, psychological testing applications, and environmental monitoring devices. These front-end devices transmit the collected data, including body and physiological data, technical movement data, psychological data, and environmental and situational data, to a central server via a wireless network. The central server, as the core processing unit, includes the aforementioned data preprocessing and analysis module, AI comprehensive evaluation module, and data storage module. After receiving the multi-element data collected by the data acquisition devices, the central server first performs fusion and preprocessing in the data processing module. The data is processed and then fed into a built-in comprehensive evaluation model for comprehensive scoring and analysis. The analysis results and athletes' historical data are stored together in a database (including a distributed database or a time-series database). On the right are user terminals, including coach terminals (PC or tablet) and athlete terminals (mobile app). The central server distributes the evaluation results and training plans to the coach terminals or user terminals via wired or wireless networks for display. At the same time, the athlete terminals support manual input to obtain feedback information, which is then returned to the AI ​​comprehensive evaluation module. The entire system forms a closed-loop data flow: collected multi-element data → server analysis → user terminal feedback → training adjustment based on feedback → new data generation → re-analysis, continuously cycling.

[0087] like Figure 3 As shown, the workflow of the AI-based athlete comprehensive evaluation system provided by this invention is as follows: S1 collects multi-factor data on athletes.

[0088] First, in the data collection phase, multi-element data of athletes are collected, including physical and physiological data, technical movement data, psychological data, and environmental and situational data.

[0089] S2, Data Preprocessing: Cleaning / Normalization / Merging.

[0090] Next, the data preprocessing stage is entered, where the collected raw data is cleaned, normalized, and the data from different sources are merged and aligned.

[0091] S3 extracts key performance indicator features.

[0092] In the feature extraction stage, key performance indicator features are extracted from the preprocessed data to form a fusion feature vector for model evaluation.

[0093] S4, the AI ​​comprehensive evaluation module calculates the comprehensive ability score.

[0094] The AI ​​comprehensive evaluation module uses a trained comprehensive evaluation model to process the input fusion feature vector and calculate the athlete's comprehensive ability score.

[0095] S5, Analyze the results and output an evaluation report: assess ability / status / score.

[0096] The evaluation report generation unit in the results output and feedback module interprets the model output, analyzes the athlete's strengths and weaknesses and current state, and outputs an evaluation report. If the evaluation process detects that an athlete is fatigued or under excessive psychological pressure, it will also flag this. Based on the evaluation results, athletes are categorized into beginner, intermediate, or advanced levels, with different training plan generation logics corresponding to different levels.

[0097] S6 generates training plans and recovery schedules.

[0098] The Personalized Training Suggestion unit in the Results Output and Feedback module develops personalized training plans and recovery arrangements based on assessment results and built-in knowledge, combined with the athlete's level and specific weaknesses.

[0099] S7 is sent to the user terminal.

[0100] The assessment report and training recommendations are sent to the user terminal for display to coaches and athletes.

[0101] It's important to note that this process is cyclical. After an athlete completes a training period, new data is input into the system for evaluation, creating a closed-loop improvement cycle. If the athlete continues training, new data will be generated in the next training cycle, and the process returns to the starting point via the dotted loop arrow for a new round of evaluation.

[0102] The embodiments of this invention can be applied to various sports fields, such as track and field and ball games, to help coaches and athletes better understand the athletes' condition and ability, formulate scientific training plans, and improve the athletes' competitive level.

[0103] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of the invention, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the invention. The sequence numbers of the above-described embodiments of the invention are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0104] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0105] The above description is merely an embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An artificial intelligence-based comprehensive athlete evaluation system, characterized in that, include: The data acquisition equipment shall include at least a central server with a data preprocessing and analysis module, an AI comprehensive evaluation module, a result output and feedback module, a central control module, and user terminals, wherein: Data acquisition equipment is used to collect multi-element data of athletes, including physical and physiological data, technical movement data, psychological data, and environmental and situational data, and uploads them to the central server in real time through a unified interface protocol. The data preprocessing and analysis module is used to preprocess the collected multi-element data and extract key features for different types of data. It also uses multimodal data fusion technology to integrate key features from different channels and types to form a fusion feature vector containing multiple elements. The AI ​​comprehensive evaluation module includes: a model building unit, used to train a comprehensive evaluation model on historical athlete data using supervised learning regression or classification models to learn the nonlinear relationship between various elements and athletic performance; and an analysis and evaluation unit, used to input the fused feature vector into the comprehensive evaluation model for calculation and evaluation, output the athlete's comprehensive ability score and determine the level. The results output and feedback module includes: an evaluation report generation unit, which automatically generates an evaluation report based on the evaluation results, including the athlete's overall ability score, scores of each sub-indicator, and strengths and weaknesses diagnosis; and a personalized training suggestion unit, which generates personalized training plans and recovery arrangements based on the athlete's specific situation and sends the results to the user terminal. The central control module is used to control and coordinate the normal operation of each module.

2. The athlete comprehensive evaluation system according to claim 1, characterized in that, The central server also deploys: The data storage module is used to classify and manage storage according to data type: continuous physiological sensor data is stored in a time-series database, video and image data is stored in a media file library, and psychological test results are stored in a relational database.

3. The athlete comprehensive evaluation system according to claim 1, characterized in that, The result output and feedback module also includes: The intelligent report interpretation and interaction unit is used to present the complex data and information in the evaluation report to users in an intuitive form using natural language processing and multimodal interaction technology, and supports users to interact with the athlete's full-element evaluation system through voice, text, or images.

4. The athlete comprehensive evaluation system according to claim 1, characterized in that, The result output and feedback module also includes: The psychological counseling service unit is used to build digital human virtual mentors; combining psychological theories and clinical practice experience, it uses knowledge graph technology to build a psychological knowledge base to provide intelligent knowledge support and decision-making assistance for digital human virtual mentors; and it realizes interactive communication with athletes through natural language processing and affective computing, and provides personalized psychological guidance and intervention.

5. The athlete comprehensive evaluation system according to claim 1, characterized in that, The multi-factor data includes at least heart rate, blood oxygen, exercise speed, acceleration, strength, psychological state information, climate conditions, altitude, venue conditions, opponent level, and match and training videos.

6. The athlete comprehensive evaluation system according to claim 5, characterized in that, The data preprocessing and analysis module includes: The preprocessing unit is used to clean, denoise, and normalize all raw data. The physical and physiological analysis unit is used to calculate the athlete's maximum heart rate, resting heart rate, and heart rate recovery index based on heart rate data; to calculate sprint speed, acceleration capacity, and fatigue index based on exercise speed and acceleration data; and to evaluate explosive power and strength endurance based on force sensor data. The technical motion analysis unit is used to extract the motion features of athletes in videos through posture estimation algorithms, and compare the motion features with a standard model to evaluate the standardization and proficiency of the movements; the motion features include at least joint trajectories and posture angles; The psychological state analysis unit is used to indirectly infer psychological state by combining the fluctuation characteristics of athletes' physiological data, and to analyze stress and fatigue levels through heart rate variability indicators. The environmental impact analysis unit is used to extract temperature, humidity, and wind speed indicators from meteorological data; The feature fusion unit is used to integrate the key features extracted from the four dimensions of physical fitness, technical movements, psychology, and environment to form a fused feature vector.

7. The athlete comprehensive evaluation system according to any one of claims 1 to 6, characterized in that, The analysis and evaluation unit uses minimum and maximum value normalization to map each indicator in the fused feature vector to the 0-1 interval, and sets reasonable upper and lower limits for indicators with clear good or bad directions. After quantification and normalization, all indicators in the fused feature vector are transformed into dimensionless standardized indicators. For different sports and evaluation purposes, the following linear weighted model is used to calculate the comprehensive ability score S: Where n is the number of index items, and i is a natural number from 1 to n. w is the score of the i-th standardized indicator output by the comprehensive evaluation model. i It is the weight of the i-th standardized indicator, and .

8. The athlete comprehensive evaluation system according to claim 7, characterized in that, In the linear weighted model, four categories are set up: physical fitness score, technical score, psychological score, and environmental score, with weights of 0.4, 0.4, 0.1, and 0.1 respectively. Within each category, the indicators are further subdivided and assigned secondary weights. The final comprehensive ability score is obtained by hierarchical weighted summation.

9. The athlete comprehensive evaluation system according to any one of claims 1 to 6, characterized in that, In the evaluation report generation unit, a knowledge graph in the sports field is constructed using an expert knowledge base. Combined with formative evaluation based on dynamic tracking and summative evaluation based on stage summary, personalized growth path planning is provided for athletes.

10. The athlete comprehensive evaluation system according to any one of claims 1 to 6, characterized in that, The personalized training suggestion unit categorizes athletes into beginner, intermediate, or advanced levels based on assessment results. It then develops personalized training plans and recovery arrangements based on a sports science knowledge base, taking into account the athlete's level and specific weaknesses. Simultaneously, it introduces decision rules based on domain knowledge to dynamically adjust the training plan according to the athlete's real-time status and to make judgments and handle specific situations.

Images