Resting body composition, heart rate, and systolic BP strongly predicted exercise fitness (VO2rel R2=0.55, VO2abs R2=0.79) in athletes using AI models.
Can artificial intelligence models accurately predict exercise fitness indicators using resting parameters in athletes and controls?
Artificial intelligence models, particularly Random Forest regression, can accurately predict exercise fitness indicators using resting parameters such as body composition, resting heart rate, and systolic blood pressure.
Absolute Event Rate: 0% vs 0%
Abstract Introduction The complex relationships of parameters characterising sport adaptation are still poorly understood. Aim To analyse large numbers of athlete and control data using artificial intelligence to reveal the relationships between resting and exercise sport adaptation parameters. Methods Linear-, Multi-layer perceptron-, Support vector-, K neighbors-, Gradient boosting- and Random forest regression analysis of the correlations between anamnestic, ECG, BP, body composition, laboratory, cardiac ultrasound and ergospirometry data of a large number of elite athletes, recreational athletes and controls. Results In the learning database, 170 parameters of 983 different screening tests of 765 asymptomatic subjects (male: 566, age: 18.9 14.1 - 41.1 years, elite athlete: 712) were analysed. Random Forest regression analysis proved to be the best estimator of relative and absolute maximal aerobic capacity (VO2rel, VO2abs), maximal ventilation (VEmax) and heart rate reserve (HRR) (R2 VO2rel: 0.42, VO2abs: 0.79, VEmax: 0.62, HRR: 0.50). Based on this, the strongest resting determinants of VO2rel were age, sex, sport, hours of training, body composition (e.g. body fat% and fat-free mass index as negative predictors), pre-exercise heart rate and systolic BP, certain laboratory parameters (e.g. seCK and seHgb as positive predictors) and atrial diameters. Certain body composition parameters, pre-exercise heart rate and systolic BP were also among the strongest predictors of VO2abs and VEmax. Pre-exercise heart rate and age were the strongest predictors of HRR. The results of 112 studies of 96 asymptomatic athletes and controls (male: 38, age: 23.4 14.9-49.0 years, elite athlete: 84) from a prospective validation database confirmed the accuracy of the models (R2 VO2rel: 0.55, VO2abs: 0.79, VEmax: 0.69,HRR:0.45). Conclusion Our prospectively validated results show that the resting parameters that most influence the studied exercise fitness indicators are body composition components, resting heart rate and systolic blood pressure. Using artificial intelligence will enable a deeper understanding of the correlates of sport adaptation processes and their use in the follow-up of athletes.Validation results
Kiss et al. (Sat,) reported a other. Resting body composition, heart rate, and systolic BP strongly predicted exercise fitness (VO2rel R2=0.55, VO2abs R2=0.79) in athletes using AI models.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: