Page - 336 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

Toward Early Detection and Monitoring of Chronic Heart Failure Using Heart Sounds Martin GJORESKIa,1, Anton GRADIŠEKa, Borut BUDNAa, Matjaž GAMSa and Gregor POGLAJEN b a Department of Intelligent Systems, Jožef Stefan Institute, Ljubljana, Slovenia b Advanced Heart Failure and Transplantation Programme, Department of Cardiology, UMC Ljubljana, Slovenia Abstract. Chronic heart failure (CHF) affects over 26 million of people worldwide and represents a significant societal, logistic and financial burden both for the patients and for the healthcare system, necessitating novel management approaches of this patient population. In this paper, we explore the possibilities of detecting heart failure worsening based on heart sounds using machine-learning methods. First, we developed a method that distinguishes between healthy individuals and those with a decompensated CHF episode. Our method includes filtering, segmentation, feature extraction, and machine learning, and was tested with a leave- one-subject-out evaluation technique on the data from 193 individuals. The method achieved 82% accuracy, outperforming the baseline classifier for 14 percentage points. In the next stage, we explored the differences between decompensated and recompensated states of CHF patients. We identified ten features for which there is statistically significant difference (p<0.001) in the features distributions, when calculated between decomensated and recompensated state of CHF. These features may be the key for developing algorithms for continuous personalized remote monitoring of the CHF patients. Keywords. Chronic heart failure, wearable device, heart sound 1. Introduction Chronic heart failure (CHF) is a chronic progressive condition where the heart is unable to pump enough blood to meet the metabolic needs of tissues and organs at the physiological filling pressures [1]. The incidence of CHF is increasing by 2%. In developed world, CHF affects 1-2 % of total population and 6-10 % of people older than 65 years. Despite the progress in medical- and device-based treatment approaches in the last decades, the overall prognosis of CHF is still dismal as 5-year survival of this population only reaches 50%. In the typical clinical course of CHF we observe alternating episodes of compensated (when the patient feels well) and decompesated phases when symptoms and signs of chronic heart failure (such as dyspnea, orthopnea, pulmonary edema, lived congestion, pulmonary edema etc.) can readily be established. During the latter episodes patients often require hospital admission for treatment with intravenous medications (diuretics, inotropes) to achieve successful recompensation. Early HF worsening detection would allow a treating physician to timely adjust patient’s medical management and thus avoid hospital admission. Currently an experienced physician can detect worsening of HF by examining the patient and through characteristic changes in the changes in heart failure biomarkers (determined from the patient’s blood). Additionally, in some patients, characteristic changes in heart sounds 1 martin.gjoreski@ijs.si Intelligent Environments 2019 A. Muñoz et al. (Eds.) © 2019 The authors and IOS Press. This article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0). doi:10.3233/AISE190061 336