Smart phone-based auscultation platform

Autor: Richard Fletcher, Vivek Miglani, Alpa Dalal, Vandana Das, Daniela Ganelin, Ajay Lanjewar, Daniel Chamberlain, Rahul Kodgule, Tushar Sahasrabudhe, Yogesh Thorat
Rok vydání: 2016
Předmět:
Zdroj: 1.6 General Practice and Primary Care.
Popis: INTRODUCTION: For global health and telemedicine, there is a need for low-cost tools that can automatically identify adventitious lung sounds. METHODS: We developed a low-cost stethoscope and mobile phone software that can record lung sounds and detect adventitious sounds (Chamberlain et al. 2015). The system was used to record lung sounds from 155 patients who visited hospitals and clinics in India. Each patient9s sound files were reviewed by a pulmonologist who indicated the presence of wheezes or crackles. For each sound file, a spectrogram was generated and used to extract features based on variation in both time and frequency. Using these features and previously published features (Aydore et al. 2009), we trained a Support Vector Machine (SVM) to identify wheezes and crackles. RESULTS: Using leave-one-out cross validation, our crackle detector algorithm yielded an Area Under the Curve (AUC) value of 0.87 and the wheeze algorithm produced an AUC value of 0.71. CONCLUSION: These preliminary results from our mobile phone-based tool indicate potential use for screening patients for pulmonary disease and providing diagnostic support for non-specialist doctors and health workers in low-resource settings.
Databáze: OpenAIRE
Externí odkaz: