A framework for automated heart and lung sound analysis using a mobile telemedicine platform

Abstract

Many resource-poor communities across the globe lack access to quality healthcare,due to shortages in medical expertise and poor availability of medical diagnostic devices. In recent years, mobile phones have become increasingly complex and ubiquitous. These devices present a tremendous opportunity to provide low-cost diagnostics to under-served populations and to connect non-experts with experts. This thesis explores the capture of cardiac and respiratory sounds on a mobile phone for analysis, with the long-term aim of developing intelligent algorithms for the detection of heart and respiratory-related problems. Using standard labeled databases, existing and novel algorithms are developed to analyze cardiac and respiratory audio data. In order to assess the algorithms' performance under field conditions, a low-cost stethoscope attachment is constructed and data is collected using a mobile phone. Finally, a telemedicine infrastructure and work-flow is described, in which these algorithms can be deployed and trained in a large-scale deployment.