Integrating Biomarkers, Voice Analysis, and Sensors: STRATIFYHF Decision Support Platform

Heart failure (HF) remains a leading cause of morbidity and mortality worldwide, yet early detection in primary care settings is challenging due to non-specific symptoms and limited diagnostic resources. This study proposes an artificial intelligence-based decision support system integrating clinical biomarkers, voice analysis, and wearable sensor data for early HF detection and risk stratification. The system uses data from retrospective and prospective studies from clinical partners under the STRATIFYHF project. Machine learning models, primarily based on random forest and XGBoost algorithms, were trained for early HF detection using primary care parameters, LVEF-based classification (HFpEF, HFmrEF, HFrEF), and future risk prediction. Voice biomarker extraction employed a multi-test recording protocol, capturing time-domain, pitch, loudness, spectral, and perturbation features linked to cardiorespiratory compromise. A custom wearable device integrating ECG and seismocardiogram (SCG) sensors was developed with dual-core architecture, achieving high-fidelity signal acquisition. The platform roadmap demonstrates significant potential for transforming HF management, enabling accessible screening across primary and secondary care settings.

This paper was presented at the 20th International Symposium on Organizational Sciences (SymOrg 2026), held in Kopaonik, Serbia, 10-12 June 2026, and was developed within the framework of the STRATIFYHF project.