Research Square · Jul 31, 2026

Resting electrocardiographic and pulse-wave deep learningidentifies exaggerated exercise blood pressure response in elite athletes

By Eckerle D, Gumpfer N, Guckert M et al.

An exaggerated systolic blood pressure response to exercise (eBPR) is associated with future arterial hypertension and adverse cardiovascular outcomes, yet it remains an exercise-derived phenotype with heterogeneous definitions. We investigated whether resting cardiovascular biosignals can identify eBPR, defined by a workload-indexed systolic blood pressure-to-metabolic equivalent slope (SBP/MET slope) >6.2 mmHg/MET. In this retrospective study, 197 predominantly male professional athletes contributed 334 pre-season screening examinations comprising resting 12-lead electrocardiograms (ECG), radial pulse waves (RPW) and cycle ergometry. Convolutional neural networks were trained with group-stratified five-fold cross-validation and evaluated on a hold-out test set. The combined RPW-ECG model, via late fusion, achieved an area under the receiver operating characteristic curve of 0.79 (95% CI 0.65-0.90), slightly exceeding single-modality models (ECG: 0.77 [0.62-0.90]; RPW: 0.66 [0.50-0.80]). Explainable AI identified contributing signal regions. As proof of concept, resting biosignals could complement, not replace, exercise testing, pending external validation and prospective outcome studies.

doi.org/10.21203/rs.3.rs-10315317/v1Open accessBiomarkers

AgingResearch.News