Predicting Performance and Functional Reserves of Athletes Based on Their Pulse Indicators in Different Trainings

Predicting capabilities of individual athletes is important for effective allocation of trainers’ effort, equipment time, and other limited resources available within national sports development programs. The existing machine learning techniques used for this purpose generally require considerable amounts of prior data on the sportsmen’s training and performances. In our paper, we develop an approach based on individual pulse indicators in different types of training, which can be collected reasonably easily and quickly. Our pilot study is done with the data from 14 runners and 11 factors. We construct regression model for running time on the distance of 1 km, which explains 89% of its variance. The significant training-related factors include pulse indicators at recovery cross running, at working cross running, at speed training, and at strength training. Also, the athletes who have higher maximum individual pulse demonstrate better performance. We believe our results might be useful for developing decision-support methods in e-government systems for better sports management.