Abstract
Cardiorespiratory fitness expressed as maximal oxygen consumption (V̇O(2max)) is a strong predictor of cardiovascular health, but its measurement through cardiopulmonary exercise (CPX) testing is complex and costly. This study develops and validates an algorithm for non-exercise estimation of V̇O(2max) using seismocardiography (SCG-V̇O(2max)). Data from SCG recordings and CPX tests of 300 subjects were combined into a database, with 83 subjects undergoing repeated sessions. SCG was recorded via a sensitive accelerometer on the lower sternum in a supine position. A machine learning algorithm was trained on data from 221 subjects, with 74 subjects comprising a test set. SCG- V̇O(2max) (44.8 ± 9.4 ml/min/kg) was comparable to CPX V̇O(2max) (44.0 ± 10.2 ml/min/kg), with a correlation of r = 0.873. Day-to-day variation was low for both methods. SCG-based estimation of V̇O(2max) is a novel, easy-to-use, and accurate method for assessing cardiorespiratory fitness, with high reproducibility and potential for integration into health evaluations.