Model for predicting repeated exercise time of female football player after lower limb exercise injury
By constructing a prediction model for the return-to-sport time after lower limb sports injury in female football players, using LASSO regression and SHAP algorithm to screen important factors, the problem of lack of risk prediction of lower limb sports injuries in the existing technology is solved, and female football players are realized Effective prediction and recovery time prediction improve game performance.
Patent Information
- Application Number
- CN202410941332.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-12-13
AI Technical Summary
There is currently a lack of prediction models for lower limb sports injuries in professional athletes, especially for local professional athletes. The application value is low. The existing models mostly use foreign athletes as samples. The research lacks continuity and the variables are single, so it is impossible to effectively predict lower limbs of female football players. Return to exercise time after exercise injury.
A prediction model for female football players' return to exercise time after lower limb sports injury was constructed. By selecting provincial and municipal female football players, basic information, lower limb sports injury data, training competition information and psychological diet information were collected, and LASSO regression algorithm was used to screen 33 risk factors are identified, 'height', 'daily training frequency' and 'physical training duration' as important factors, and the influencing factors are identified through the SHAP algorithm to establish an early warning model.
It effectively predicts the return time of female football players after lower limb sports injuries, determines influencing factors, helps athletes recover faster and improve their game performance, and provides targeted injury prevention training plans and rehabilitation education.
Smart Images
Figure CN119132588A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of sports medicine, and in particular relates to a model for predicting the time for female football players to return to sports after lower limb sports injuries. Background Art
[0002] In recent years, artificial intelligence technology supported by big data has developed rapidly in the medical field, playing a vital role in disease diagnosis, screening, and treatment selection guidance. [1] Compared with traditional statistical methods, artificial intelligence technology can abstract low-dimensional features layer by layer into higher-dimensional features, thereby better integrating various influencing factors to obtain a higher level of prediction accuracy. [2] Through computer models that automatically evaluate various types of medical data (such as medical imaging data, medical history information, test results, etc.) and strategies such as variable conversion and fusion to improve model performance, artificial intelligence technology can fully realize information complementarity between data and get rid of the time and space limitations of single data, so as to more comprehensively understand the risk of disease occurrence or the prognosis of disease treatment, and provide objective basis for clinicians' decision-making. [2,1] . At present, big data platforms and artificial intelligence algorithms have been widely valued and applied in clinical diagnosis and treatment and scientific research activities in institutions including Huashan Hospital Affiliated to Fudan University, and have a good application foundation in disease model establishment and data analysis, such as intelligent identification of imaging lesions and risk assessment of postoperative complications of patients. Therefore, we believe that with the help of big data and artificial intelligence technology, the risk of outcome events in patients can be calculated by incorporating risk factors or protective factor data related to lower limb sports injuries, making it possible to predict the occurrence and development of sports injuries, thereby achieving the purpose of early detection and early intervention, and establishing an artificial intelligence-assisted prevention and treatment system for lower limb sports injuries.
[0003] However, there is no prediction model for lower limb sports injury risk in professional athletes in the medical and sports communities. The few initial models that have been published mostly use foreign athletes as samples, and the variables included are single (discipline limitations), and the research lacks continuity, which makes them of low application value to local professional athletes. [3–5] . Summary of the invention
[0004] The purpose of the present invention is to provide a model for predicting the time of returning to sport after lower limb sports injuries for female football players, so as to help them perform better in the game.
[0005] The prediction model of the time for female football players to return to sports after lower limb sports injuries provided by the present invention is constructed by the following steps:
[0006] The first step is the selection of female soccer players required for training.
[0007] The specific selection is members of the provincial and municipal women's football teams (more than 150). Selection criteria: All-Olympic alternate members of the women's football team who voluntarily joined the study. Exclusion criteria: Athletes who refused to participate before the study, athletes who were unable to participate in competition training for a long time (more than half a year) due to major illnesses, etc. Subject withdrawal criteria: Subjects are unwilling or unable to continue to participate in the study; After the subject is enrolled, it is found that he does not meet the inclusion criteria, or meets any of the exclusion criteria; The study cannot be completed as planned due to major illness, retirement, pregnancy, etc. during the study.
[0008] The second step is to collect information on female football players required for training and assign them to cohorts.
[0009] The following information of athletes was retrospectively collected by the software of the lower limb injury data management platform for elite athletes: (1) basic information data, such as name, height, age, weight, gender, medication and supplement use, vaccination history, family medical history, menstrual history and its impact on sports performance, etc.; (2) medical standard data of lower limb sports injuries, such as the name of the patient, disease name, disease location, diagnosis record, diagnosis time, etc.; (3) information related to athlete training and competition, such as physical training before, during and after the season, and injury prevention training, etc.; (4) athlete psychological diet information, such as sleep conditions, expectations for performance, and diet, etc.
[0010] In addition, the following information about athletes is collected prospectively: (1) medical standard data on lower limb sports injuries, such as the name of the patient, disease name, disease location, diagnosis record, diagnosis time and other related fields; (2) information related to athlete training and competition, such as physical training before, during and after the season and injury prevention training; (3) athlete psychological and dietary information, such as sleep conditions, expectations for performance and dietary conditions; and (4) changes in basic information data.
[0011] The outcome variable was determined to be the time it took female soccer players to return to sport after lower limb sports injuries. Lower limb sports injuries specifically included common anterior cruciate ligament injuries of the knee, meniscus injuries of the knee, patellar tendon tendinopathy, ankle instability, Achilles tendon tendinopathy, and tibial fatigue fractures.
[0012] According to the above conditions, the number of female football players to be analyzed was determined; then, a portion (most) of the female football players were randomly assigned to the model training cohort, and the remaining portion (a small portion) was randomly assigned to the model validation cohort. There was no significant difference in the lower limb injury-related information, training and competition information, psychological diet information, and other information between the training and validation cohorts.
[0013] The third step is to establish an early warning model.
[0014] The least absolute shrinkage and selection operator (LASSO) regression algorithm was used to screen variables and model 33 risk factors from x1 to x33. Specifically, the Lasso library was imported using the from sklearn.linear_model import Lasso command, sklearn version 1.4.0. The model lasso_model = Lasso (alpha = 8.35) was first defined, where the regularization parameter was 8.35 (corresponding to a smaller MSE and a larger R^2), and then the lasso_model.fit() method was used for fitting, and finally the plot() method was used for plotting. The directly related factors were obtained: x2: height, x3: weight, x12: daily training frequency, x17: physical training duration;
[0015] The specific expression of the early warning model is:
[0016] y=a2x 2 -a3x 3 -a12x 12 -a17x 17 -a0;
[0017] a2, a3, a12, a17, a0 are coefficients. y is the return time (rounded, in days).
[0018] Variable explanation: x1: age, x2: height, x3: weight, x4: BMI index, x5: body fat percentage, x6: dominant foot, x7: team, x8: position on the field, x9: whether the lower limb pain is a contact injury, x10: whether the lower limb pain is injured during training or in the game, x11: whether it is a secondary injury, x12: daily training frequency, x13: daily training duration, x14: whether warm-up training is performed, x15: whether physical training is performed, x16: physical training frequency, x17: physical training duration, x18: I slept well last night, x19: I am looking forward to today's training or competition, x20: I am optimistic about my future athletic performance, x21: I feel energetic, x22: I have a good appetite, x23: I have very little muscle discomfort, x24: History of allergies, x25: Whether the new crown vaccine has been administered, x26: Do any of the following diseases exist in myself or my immediate family members, x27: Menarche, x28: Agility, x29: Balance, x30: Running ability, x31: Mobility, x32: Muscle strength, x33: Explosive ability;
[0019] The fourth step is to identify important risk factors.
[0020] The SHAP algorithm was used to identify the important risk factors that affect the time it takes for female football players to return to sport after lower limb sports injuries. First, the shap library was imported, and then the shap.Explainer() method was used to create an explainer. The explainer.shap_values() method was used to calculate the shap value, and finally the shap.summary_plot() method was used to draw the graph. Specifically, "height", "daily training frequency", "physical training duration", and "weight" were screened as the five risk factors with the greatest contribution, suggesting that these risk factors may lead to the time it takes for female football players to return to sport after lower limb sports injuries.
[0021] Technical Effects of the Invention
[0022] The present invention establishes a prediction model for the time it takes for female football players to return to sports after lower limb sports injuries. After screening 33 risk factors including lower limb injury-related information, training and competition information, psychological and dietary information, and other information using a screening model strategy, 4 characteristic variables are selected to effectively predict the time it takes for female football players to return to sports after lower limb sports injuries. In addition, the present invention also identifies the important risk factors that affect the time it takes for female football players to return to sports after lower limb sports injuries, suggesting that "height", "daily training frequency", "physical training duration", and "weight" may significantly affect the time it takes for female football players to return to sports after lower limb sports injuries. The present invention can assist in predicting the time it takes for female football athletes to return to sports after lower limb sports injuries and accelerate the time to return to sports based on high-level risk factors, thereby facilitating better game performance. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 This is a nomogram for predicting the time to return to sport after lower limb sports injuries in female soccer players. The model prediction factors include "height", "daily training frequency", "physical training duration" and "weight".
[0024] Figure 2 This is a bar chart of the important risk factors identified by the SHAP algorithm. The larger the SHAP value, the greater the contribution of the risk factor to the outcome. "Height", "Daily Training Frequency", "Physical Training Duration", and "Weight" were selected as the risk factors with the greatest contribution.
[0025] Figure 3 This is a waterfall chart that uses the SHAP algorithm to identify important risk factors in the model, showing the precise contribution of different features. Gray is positively correlated, black is negatively correlated, and the wider the color area, the greater the contribution of the risk factor. DETAILED DESCRIPTION
[0026] The present invention takes a women's football team in a certain region as an example and further introduces the present invention in conjunction with the accompanying drawings.
[0027] The specific construction process of the prediction model for the time of return to sports after lower limb sports injuries in female football players is as follows:
[0028] The first step is the selection of female soccer players required for training.
[0029] All-Olympic alternates for a women's soccer team in a certain region. Selection criteria: All-Olympic alternates for a women's soccer team who are active during the study period and voluntarily join. Exclusion criteria: Athletes who refuse to participate before the study, athletes who are unable to participate in competitions and training for a long time (more than half a year) due to major illnesses, etc. Subject withdrawal criteria: Subjects who are unwilling or unable to continue to participate in the study; Subjects who are found to not meet the inclusion criteria after enrollment, or meet any of the exclusion criteria; Subjects who are unable to complete the study as planned due to major illnesses, retirement, pregnancy, etc. during the study.
[0030] The second step is to collect information on female football players required for training and assign them to cohorts.
[0031] The software designed by the team was used to retrospectively collect the athletes' (1) basic information data, such as name, height, age, weight, gender, medication and supplement use, vaccination history, family disease history, menstrual history and its impact on sports performance; (2) medical standard data of lower limb sports injuries, such as the name of the doctor, disease name, disease location, diagnosis record, diagnosis time and other related fields; (3) information related to the athlete's training and competition, such as physical training before, during and after the season and injury prevention training; (4) athlete's psychological diet information, such as sleep status, expectations for performance and diet, etc. And prospectively collect the athletes' (1) medical standard data of lower limb sports injuries, such as the name of the doctor, disease name, disease location, diagnosis record, diagnosis time and other related fields; (2) information related to the athlete's training and competition, such as physical training before, during and after the season and injury prevention training, etc.; (3) athlete's psychological diet information, such as sleep status, expectations for performance and diet, etc.; (4) changes in basic information data. The outcome variable was determined to be whether the female football players returned to sports after lower limb sports injuries, including common lower limb sports injuries such as anterior cruciate ligament injury of the knee, meniscus injury of the knee, patellar tendinopathy, ankle instability, Achilles tendinopathy, and tibial fatigue fracture.
[0032] Specifically, according to the above conditions, a total of 150 female football players were included; a portion of them (120 cases) were randomly assigned to the model training cohort, and the remaining part (30 cases) were randomly assigned to the model validation cohort. There was no significant difference in lower limb injury-related information, training and competition information, psychological diet information, and other information between the training and validation cohorts.
[0033] The third step is to establish an early warning model.
[0034] The least absolute shrinkage and selection operator (LASSO) regression algorithm was used to screen variables and model 33 risk factors. Specifically, the Lasso library was imported using the from sklearn.linear_model import Lasso command, sklearn version 1.4.0, and the model lasso_model = Lasso (alpha = 8.35) was first defined. Here, the regularization parameter was selected as 8.35 (corresponding to a smaller MSE and a larger R^2), and then the lasso_model.fit() method was used for fitting, and finally the plot() method was used for plotting.
[0035] The specific expression of the early warning model is:
[0036] y=2.881955·x 2 -0.016774 x 3 -3.606269 x 12 -0.562465 x 17 -432.6826;
[0037] Variable explanation: x1: age, x2: height, x3: weight, x4: BMI index, x5: body fat percentage, x6: dominant foot, x7: team, x8: position on the field, x9: whether the lower limb pain is a contact injury, x10: whether the lower limb pain is injured during training or in the game, x11: whether it is a secondary injury, x12: daily training frequency, x13: daily training duration, x14: whether warm-up training is performed, x15: whether physical training is performed, x16: physical training frequency, x17: physical training duration, x18: I slept well last night, x19: I am looking forward to today's training or competition, x20: I am optimistic about my future athletic performance, x21: I feel energetic, x22: I have a good appetite, x23: I have very little muscle discomfort, x24: History of allergies, x25: Whether the new crown vaccine has been administered, x26: Do any of the following diseases exist in myself or my immediate family members, x27: Menarche, x28: Agility, x29: Balance, x30: Running ability, x31: Mobility, x32: Muscle strength, x33: Explosive ability;
[0038] The fourth step is to identify important risk factors.
[0039] The SHAP algorithm was used to identify the important risk factors that affect the time it takes for female football players to return to sport after lower limb sports injuries. First, the shap library was imported, and then the shap.Explainer() method was used to create an explainer. The explainer.shap_values() method was used to calculate the shap value, and finally the shap.summary_plot() method was used to draw the graph. Specifically, "height", "daily training frequency", "physical training duration", and "weight" were screened as the five risk factors with the greatest contribution, suggesting that these risk factors may lead to the time it takes for female football players to return to sport after lower limb sports injuries.
[0040] Prediction of return to sport time after lower limb sports injuries in female soccer athletes
[0041] This model is used to predict the time it takes for female football players to return to sports after lower limb sports injuries based on their comprehensive conditions such as injury, training, psychology and diet. For example, a 20-year-old, 170cm tall, 70kg in weight, BMI24, 14% body fat, right foot, a female adult professional football team, a central defender, has had 1 lower limb injury and the injury was a contact injury during a game, not a secondary injury, has 5 daily training sessions, 2 hours per week, warm-up training before training, 2 daily physical training sessions, 2 hours per week, no special tendency in injury prevention training on average [agility, balance, running, movement, muscle, burst: each injury training There is a 50% probability of conducting special training for each training item], daily psychology, diet, sleep and muscle condition are moderate [I slept well last night; I look forward to today's training or competition; I am optimistic about future athletic performance; I feel energetic; I have a good appetite; I have very little muscle discomfort; 1-5 points scale score 5 is very good, all scores are 3 points], no history of allergies, received the new crown vaccine, and the individual and immediate family members have no hypertension, diabetes or other hereditary diseases. The female football player who had her first menstruation at the age of 16 will need 37 days to return to sports in this model.
[0042] Development of a targeted lower limb injury prevention training program for female football players
[0043] The present invention also identifies important risk factors that affect the time it takes female football players to return to sport after lower limb sports injuries. "Height", "daily training frequency", "physical training duration" and "weight" are screened as the risk factors that contribute the most, indicating that these risk factors may significantly affect the time it takes female football players to return to sport after lower limb sports injuries. Clinical rehabilitation therapists can formulate targeted injury prevention training plans. For example, increasing the frequency of daily training, the duration of physical training, and appropriate weight loss can all help shorten the time it takes to return to sport.
[0044] Rehabilitation education for lower limb injuries in female football players
[0045] The prediction model of the present invention shows that female football players with high height, high body weight, low daily training frequency and short physical training time need longer time to return to sports, and such players should be given special protection and psychological counseling education. The clinical prediction model can be used to assist in the education of female football players, so that they can be aware of the occurrence of the disease and not be too negative or too anxious.
[0046] References
[0047] [1]KUMAR S, SINGH M. Big data analytics for healthcare industry: Impact,
[0048] applications,and tools[J / OL].Big Data Mining and Analytics,2019,2(1).
[0049] DOI:10.26599 / BDMA.2018.9020031.
[0050] [2]POLSON NG,SOKOLOV V O.Deep Learning-Nature Review[J].Nature, 2018, 521(7553).
[0052] [3]LIU F, ZHOU Z, SAMSONOV A, etc. Deep learning approach for evaluatingknee MR
[0053] images:Achieving high diagnostic performance for cartilage lesion
[0054] detection[J / OL].Radiology,2018,289(1).DOI:10.1148 / radiol.2018172986.
[0055] [4]DALLINGA JM,BENJAMINSE A,LEMMINK KAP M.Which Screening ToolsCan
[0056] Predict Injury to the Lower Extremities in Team Sports?:A Systematic
[0057] Review[M / OL] / / Sports Medicine.(2012).DOI:10.1007 / BF03262295.
[0058] [5]WHITTAKER J L,BOOYSEN N,DE LA MOTTE S etc.Predicting sport and
[0059] occupational lower extremity injury risk through movement quality screening:
[0060] A systematic review[M / OL] / / British Journal of Sports Medicine.(2017).
[0061] DOI:10.1136 / bjsports-2016-096760。
Claims
1. A model for predicting the time of return to sports for female football players after lower limb sports injuries, characterized in that: It is constructed by the following steps: The first step is to select female football players required for training; The specific selection is for members of the provincial and municipal women's football teams, more than 150; selection criteria: All-Olympic alternate members of the women's football team who are in service during the study period and voluntarily join; exclusion criteria: athletes who refuse to participate before the study, athletes who are unable to participate in competitions and training for more than half a year due to major illnesses, etc.; subject withdrawal criteria: subjects are unwilling or unable to continue to participate in the study; after the subject is enrolled, it is found that they do not meet the inclusion criteria, or meet any of the exclusion criteria; unable to complete the study as planned due to major illness, retirement, or pregnancy during the study; The second step is to collect the information of female football players required for training and allocate them to cohorts; The following information of athletes was collected retrospectively by the software of the lower limb injury data management platform for elite athletes: (1) basic information data, including name, height, age, weight, gender, medication and supplement use, vaccination history, family medical history, menstrual history and its impact on sports performance; (2) medical standard data of lower limb sports injuries, including the name of the doctor, disease name, disease location, diagnosis record, and diagnosis time; (3) information related to athlete training and competition, including physical training before, during and after the season and injury prevention training; (4) psychological diet information of athletes, including sleep status, expectations for performance and diet; In addition, the following information of athletes was collected prospectively: (1) medical standard data of lower limb sports injuries, including the name of the doctor, the name of the disease, the location of the disease, the diagnosis record, and the diagnosis time related fields; (2) information related to the athlete's training and competition, including physical training before, during and after the season and injury prevention training; (3) psychological and dietary information of the athlete, including sleep conditions, expectations for performance, and diet; (4) changes in basic information data; The outcome variable was determined as the time it took female soccer players to return to sports after lower limb sports injuries. Lower limb sports injuries specifically included common anterior cruciate ligament injuries of the knee, meniscus injuries of the knee, patellar tendinopathy, ankle instability, Achilles tendinopathy, and tibial fatigue fractures. According to the above conditions, the number of female football players to be analyzed was determined; then, most of the female football players were randomly assigned to the model training cohort, and the remaining part, i.e., a small part, was randomly assigned to the model validation cohort; there was no significant difference in lower limb injury-related information, training and competition information, psychological diet information, and other information between the training and validation cohorts; The third step is to establish the prediction model; The least absolute shrinkage and selection operator (LASSO) regression algorithm was used to screen and model 33 risk factors from x1 to x33. Specifically, the Lasso library was imported using the from sklearn.linear_model import Lasso command, sklearn version 1.4.
0. The model lasso_model = Lasso (alpha = 8.35) was first defined, where the regularization parameter was 8.35, and then the lasso_model.fit() method was used for fitting, and finally the plot() method was used for plotting. The directly related factors were obtained: x2: height, x3: weight, x12: daily training frequency, x17: physical training duration; The specific expression of the early warning model is: <h2 style=";text-align:left;direction:ltr">y = a2x2 - a3x3 - a12x<h2 style=";text-align:left;direction:ltr"> 12 <h2 style=";text-align:left;direction:ltr"> -a17x<h2 style=";text-align:left;direction:ltr"> 17 <h2 style=";text-align:left;direction:ltr"> -a0; a2, a3, a12, a17, a0 are coefficients; y is the reentry time in days; Variable explanation: x1: age, x2: height, x3: weight, x4: BMI index, x5: body fat percentage, x6: dominant foot, x7: team, x8: position on the field, x9: whether the lower limb pain is a contact injury, x10: whether the lower limb pain is injured during training or in the game, x11: whether it is a secondary injury, x12: daily training frequency, x13: daily training duration, x14: whether warm-up training is performed, x15: whether physical training is performed, x16: physical training frequency, x17: physical training duration, x18: I slept well last night, x19: I am looking forward to today's training or competition, x20: I am optimistic about my future athletic performance, x21: I feel energetic, x22: I have a good appetite, x23: I have very little muscle discomfort, x24: History of allergies, x25: Whether the new crown vaccine has been administered, x26: Do any of the following diseases exist in myself or my immediate family members, x27: Menarche, x28: Agility, x29: Balance, x30: Running ability, x31: Mobility, x32: Muscle strength, x33: Explosive ability; Step 4: Identification of important risk factors; The SHAP algorithm was used to identify the important risk factors that affect the time for female football players to return to sports after lower limb sports injuries. First, the shap library was imported, and then the shap.Explainer() method was used to create an explainer, the explainer.shap_values() method was used to calculate the shap value, and finally the shap.summary_plot() method was used to draw the graph. Specifically, "height", "daily training frequency", "physical training duration", and "weight" were screened as the five risk factors with the greatest contribution, suggesting that these risk factors may lead to the time for female football players to return to sports after lower limb sports injuries.
2. The prediction model for the time to return to sports of female football players after lower limb sports injuries according to claim 1 is characterized in that: Specifically: y=2.881955·x2-0.016774·x3-3.606269·x 12 -0.562465·x 17 -432.6826。