SoundTrack-HF: Noninvasive prediction of acute events in patients with chronic heart failure
▶Summary
Heart failure (HF) is a chronic condition which tremendously impacts the life quality and expectancy of the patient as well as the healthcare system. The main responsible are acute episodes of decompensation, which are difficult to promptly detect because signs and symptoms occur late in the process. It was previously proved that changes in the hemodynamic parameters anticipate acute episodes by weeks, and their monitoring could be leveraged to predict and prevent the episode itself. Nevertheless, at date, this can only be performed via invasive sensors, which limit the target population to patients with severe HF. SoundTrack-HF targets this gap in the technology for HF monitoring. The goal of SoundTrack-HF is to design, implement and validate an end-to-end system for the non-invasive monitoring of the status of compensation of HF patients using heart sounds (HS).The project grounds on the use of a novel prototype enabling the self-recording of HS by the patient in a homecare setting. Cross-sectional and longitudinal data will be obtained from HF patients at their domicile. Time-related biomarkers related to the hemodynamic functionality will be extracted from the signals and validated against a gold standard (echocardiography). On one side, the patho-physiological spatio-temporal changes in the parameters will be modelled and interpreted in relation to the status of compensation of the patient. On the other side, an explainable AI-based acute episode predictor will be developed and used to extract novel knowledge about the relationship between HS and HF. At the end, signal- and AI-extracted knowledge will be integrated into a final predictor. SoundTrack-HF is a highly interdisciplinary project. Its innovation resides in the development and validation of groundbreaking algorithmic solutions which, by leveraging a completely novel hardware, will disrupt the standard practice in auscultation and move it to the next level for domicile monitoring.