Stetho-phone: Low-cost digital stethoscope for remote personalized healthcare

• Published in: 2017 IEEE Global Humanitarian Technology Conference (GHTC) pp. 1 - 7
• Main Authors: Fattah, Shaikh Anowarul, Rahman, Nael Mizanur, Maksud, Ahmed, Foysal, Shariful Islam, Chowdhury, Rakibul Islam, Chowdhury, Sayeed Shafayet, Shahanaz, Celia
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2017
• Subjects: Classification algorithms; Digital Stethoscope; Electrocardiography; Heart; Hidden Markov models; Kth Nearest Neighbor; Medical services; Phonocardiography; Stethoscope; Telemedicine
• DOI: 10.1109/GHTC.2017.8239325

Proper healthcare is one of the fundamental needs of human beings. However, the doctor-to-patient ratio in developing and underdeveloped countries, especially in remote rural areas is quite low. A smart digital stethoscope, which can make preliminary health monitoring by itself and communicate with a doctor from remote areas can minimize such problem. Though there are several such devices from various medical equipment vendors, most of them are costly for our target community and lack wireless communication facilities with the doctors. Our research work aims at developing a smartphone-based personalized healthcare scheme at low cost by providing primary health monitoring focusing on cardiac activity and regularly tracking the condition and reporting to a doctor because almost one-third of all global deaths are attributed to cardiovascular diseases. A low-cost digital stethoscope chest piece was therefore developed from a conventional one using piezoelectric element which picks up body sounds and provides a corresponding electrical signal. The signal is then fed to a smartphone which evaluates the health condition based on the analysis of heart sounds using the acquired Phonocardiogram (PCG) signals. In order to improve the computational efficiency and accuracy, heart sound segmentation, probabilistic models such as hidden Markov model, application of the k-nearest neighbor algorithm to train the network for binary classification have been implemented. Based on the results analyzed, the proposed low-cost scheme shows great promise for preliminary screening of healthcare issues using heart sounds, especially in remote areas with scarce medical assistance available.