A Case Study: Adopting Artificial Intelligence to Distinguish Chronic Cough in Taiwan

EasyChair Preprint 8779

Abstract

The aim of this study is to provide a home care solution for recognizing the chronic respiratory diseases via cough sound analysis AI model. The proposed model is based on a deep learning architecture, recurrent neural network (RNN), and has been trained and tested with real world chronic cough sounds collected and labelled by professional physicians, and three chronical respiratory diseases, Allergic Rhinitis (AR), Gastroesophageal reflux disease (GERD) and asthma are considered in this work. Furthermore, we also trained the model to classify whether a cough is a dry or a wet one, which is a valuable indicator as part of clinical queries before any medical diagnosis or treatments. The proposed method has shown that the trained model is capable of providing reliable predictions for the target diseases without any intervention from professional medical staff, and this is a potential rescue for the patients who have had a hard time accessing timely medical consultations or clinics do not have sufficient manpower. As long as the trained model is able to execute on mobile devices such as smart phones or mobile tablets, it can be beneficial to enhance the home-based care for patients of chronical respiratory diseases and reduce the consumptions of medical resources.

Keyphrases: Chronic disease care, chronic cough assessment, home care, machine learning

@booklet{EasyChair:8779,
  author    = {Shu Hui Hung and Chia Yung Jui},
  title     = {A Case Study: Adopting Artificial Intelligence to Distinguish Chronic Cough in Taiwan},
  howpublished = {EasyChair Preprint 8779},
  year      = {EasyChair, 2022}}