Can we predict training performance with shooting heart rate in archers?: A machine learning approach

PurposeHeart rate (HR) values during different phases of shooting can be used for performance analysis. Machine learning (ML) methods are used in predicting performance. We aimed to develop ML model to predict performance scores using shooting HR values and also to predict the importance of these parameters in an archer. Methods32 archers (15 elite & 17 non-elite) shot two sessions of 30 arrows each indoor wearing heartrate chest monitor and were videographed. When each arrow was shot, 11 HR values were identified at different shooting phases. Other parameters with 35 linear variables and second-degree polynomial HR values were used to build ML models in Python V3.11.4. Session 1 and 2 total scores were used to train and test respectively. Root Mean Squared Error (RMSE) was used to evaluate the model performance after fine-tuning. ResultsRMSE of all 12 ML models ranged from 6.262 - 9.612. The Cat Boost model with the lowest RMSE of 6.262 was used to predict the Session 2 score. SHapley Additive exPlanations (SHAP) values showed each variables importance in prediction. ConclusionsSports age, resting systolic blood pressure, previous competition score, right hand grip-strength, age, HR before 2sec of arrow release, waist-to-hip ratio, concentration disruption trait anxiety and HR after 5sec of release are top parameters to predict score. Practical ApplicationsML model with shooting HR provides a better prediction of archery score of an individual archer.