AI-based prediction of VO2 max from 24-h Holter ECG recording

Cardiovascular (CV) deconditioning is a consequence of spaceflight, characterized by functional and structural changes in the heart and blood vessels due to prolonged exposure to microgravity. These adaptations lead to a noticeable reduction in exercise capacity, particularly evident in alterations of maximum aerobic power (VO2 max [mL/kg/min]). This work explored the hypothesis that microgravity-induced CV deconditioning could be anticipated through information derived from a non-exercise longitudinal 24-h Holter ECG recording, thereby offering a means for its regular monitoring in space. Pooled data acquired at baseline and at the last day of six Head-Down Tilt bed rest campaigns of different duration were utilized to test this hypothesis. Machine Learning models were developed using multiple features to predict VO2 max after deconditioning. Obtained results showed good prediction accuracy (RMSE < 5 mL/kg/min) demonstrating the potential of this approach to identify VO2 max deterioration during bed rest, with possible applications in real world space environments and clinical settings.