Machine learning identified four distinct cardiac chronotropic response clusters in 2,048 patients, validated in 638 patients, revealing unique heart rate trajectory dynamics.
Does a machine learning framework based on heart rate trajectories during exercise testing identify distinct patient clusters with different cardiac chronotropic responses?
An AI-driven method using routine cardiopulmonary exercise testing can successfully stratify patients into four distinct cardiac chronotropic response clusters based on heart rate trajectories.
Absolute Event Rate: 0% vs 0%
Abstract Background The cardiac chronotropic response to exercise is a complex physiological process involving the sinus and atrioventricular (AV) nodes and the autonomic nervous system. During cardiopathies and heart failure, cardiac chronotropic response is often altered contributing markedly to symptoms and prognosis. However, precise identification of mechanism underlying cardiac chronotropic responses in routine clinical settings remains suboptimal. Purpose To develop a quantitative marker and machine learning framework capable of stratifying patients into distinct chronotropic response patterns based on heart rate (HR) trajectories during standardized exercise testing. Methods Exercise test data from 2,048 patients undergoing ramp protocol cardiopulmonary exercise testing (10 watts/min) with gas exchange measurements (VO2) were analyzed. Heart rate signals were processed using two statistical methods: (1) Davies’ test to assess linearity, and (2) a novel metric for HR elasticity to detect inflection points. The first ventilatory threshold (VT1) was estimated from ventilatory equivalents (VE/VO2) and used to anchor trajectory analysis. Three physiological metrics were derived and used in an unsupervised hierarchical clustering to identify distinct chronotropic patterns. Results Four robust chronotropic response clusters were identified, each characterized by unique heart rate trajectory dynamics and associated with differing physiological profiles. The clustering solution and derived marker were externally validated in an independent cohort of 638 patients, where all four patterns and marker distributions were consistently reproduced, confirming generalizability and robustness. Change in heart rate slope around crossing first ventilatory threshold appears as a key determinant of patient clustering. Conclusion We developed an AI-driven method using routine exercise tests to identify subgroup of patients with distinct cardiac chronotropic responses to exercise. During the session, we will present pathophysiological significance and characteristics of the various subgroups of patients.
Ponnaiah et al. (Thu,) reported a other. Machine learning identified four distinct cardiac chronotropic response clusters in 2,048 patients, validated in 638 patients, revealing unique heart rate trajectory dynamics.