Diagnophone: An Electronic Stethoscope for Respiratory Audio Analysis

Yağ E., İNCE G.

Proceedings of the International Conference on Computer Science and Engineering (UBMK 2019), 1 - 04 January 2019 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/ubmk.2019.8907227
  • Keywords: Machine Learning, Signal Processing, Audio Processing, SVM, CNN, KNN, AdaBoost, Human Computer Interaction
  • Istanbul Technical University Affiliated: Yes

Abstract

Today, pulmonary diseases are one of the major causes of mortality in the world. Even though there are different diagnostic tests available such as X-ray and tomography, the stethoscope is still the first, cheapest and the most frequently used diagnostic device for the physicians. In this paper, a smart electronic stethoscope has been designed to help physicians with the diagnosis of the disease using Machine Learning. In order to create a design that satisfies all the needs of the physicians, 15 doctors and medical students from several hospitals have been contacted and interviewed. The developed system has been tested with different machine learning techniques and its efficiency has been shown by obtaining 84% accuracy while classifying respiratory audio.