DESIGN OF WHEEZING SOUND DETECTION WEARABLE DEVICE BASED ON INTERNET OF THINGS

Abstract

Introduction: Wheezing is one of the most common manifestations of airway obstruction.  The use of a stethoscope in the wheezing examination has several disadvantages such as subjective results  and depends on  the auditor's  hearing sensitivity.  So an easy device is needed that helps determine the wheezing  sound precisely.  This study assembled a single tool to detect wheezing  sounds based on the internet of things.

Method: This tool is designed with a microprocessor hardware connected to  an electric stethoscope so that it can be attached to the patient's chest  wall.  Collection of chest breathing voice data  accessed on kaggle.com.  The creation of algorithms with Convolutional Neural Networks (CNN)  was later changed to Mel Frequency Cepstral Coefficients (MFCC). This model  will be implanted in a microprocessor and use python  language to  be able to record  the sound of chest  wall vibrations.  The recorded sound is converted into MFCC  to make it easier to perform   wheezing sound detection.  MFCC image results and  detection results are sent to the database via  the firebase database feature which stores MFCC  photos in real-time as they are detected.  Designing android application software using Flutter   builds communication between android  applications and firebase databases that allows applications to  retrieve MFCC  images as the final result.

Result: The results of  the tool trial on five volunteers, three exacerbation asthma patients and two healthy people  showed the device can detect wheezing  sounds at a frequency of  400Hz with 80%  accuracy through CNN and MFCC  algorithms  Internet of things based.

Conclusion: This tool can help health workers  to accurately determine wheezing   sounds, enforce the diagnosis   faster, the prognosis of the disease to be  better, so as to  reduce the number  morbidity and mortality of diseases with airway abnormalities in Indonesia