• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Can Machine Learning Algorithms Detect Acute Respiratory Diseases Based on Cough Sounds?

Article

The predictive ability of cough sound algorithms shows promise in detecting acute respiratory diseases, study finds.

A machine learning algorithm for detecting and classifying acute respiratory disease showed good predictive ability based on cough sounds among pediatric patients, according to a new study published in International Journal of Medical Informatics.

Cough is a common symptom of acute respiratory diseases and one of the most common symptoms in primary care worldwide. However, cough sound assessment is limited due to subjective interpretations by clinical practices, which can lead to misdiagnoses and the need for emergency hospital visits.

“Despite the potential significance of objective cough sound evaluation in clinical decision-making of acute respiratory diseases, no evidence syntheses have been completed on this topic,” wrote the researchers. “Therefore, we conducted a systematic review to determine the ability of machine learning methods to predict acute respiratory diseases in the pediatric population using cough sound.”

In this study, researchers examined the objective use of artificial intelligence (AI) as a potential aid in clinical respiratory disease diagnosis.

The study reviewed 6 articles, sourced from the Scopus, Medline, and Embase databases on January 25, 2023. These articles included cough sound features and AI algorithms in the diagnoses of pediatric patients 18 years or younger. Studies based on noncough sound features, such as demographics and clinical data, in addition to cough sound features were included. Furthermore, quality assessment of these studies was performed using the checklist for the assessment of medical AI (ChAMAI).

As a result, the analysis showed variability when inputting the algorithms, including various cough sound features and the combination of sound and clinical features. Additionally, the machine learning algorithms differed from conventional algorithms.

Furthermore, accuracy for detection of bronchiolitis, croup, pertussis, and pneumonia across 5 articles ranged from 82% to 96%, but a significant drop in accuracy was found in the detection for bronchiolitis and pneumonia in the sixth article. The researchers believe that this drop in accuracy shows how clinical decision-making for diagnosing these 2 respiratory diseases may be subjective.

The researchers also found that cough sound features used for detecting croup resulted in higher accuracy ability compared with combined cough and clinical features, which they believe could be because individuals with croup have a distancing barking cough.

However, due to the limited number of studies, the researchers believe further work needs to be done to better understand how AI in health care can be used for detecting acute respiratory diseases in children.

Despite limitations, the researchers believe that the findings from this review are a good starting point and that “the promising diagnostic accuracy in most of the reviewed studies shows its potential as a respiratory disease assessment tool,” wrote the researchers. “Knowledge gained from this systematic review can be used in future study designs and also be useful to regulatory bodies, technology manufacturers, engineers and data scientists, and clinicians.”

Reference

Sharan RV, Rahimi-Ardabili H. Detecting acute respiratory diseases in the pediatric population using cough sound features and Machine Learning: A Systematic Review. International Journal of Medical Informatics. Published online April 18, 2023:105093. doi:10.1016/j.ijmedinf.2023.105093

Related Videos
Video 6 - "Closing Thoughts on the Current Ovarian Cancer Treatment Landscape"
Video 5 - "Harnessing Biomarkers & Molecular Profiling in Ovarian Cancer Treatment"
Video 7 - "Prior Authorization and Access to Targeted Treatment for Ph+ ALL Patients"
Video 6 - "Current Front-Line Targeted Treatments in Ph+ ALL"
dr amy paller
dr krystyn van vliet
dr amy paller
Video 2 - "Early-stage Type 1 Diabetes and The Psychological Impact of Progressing to Stage 3"
Video 2 - "Exploring Therapies for Platinum Sensitive & Platinum-Resistant Ovarian Cancer"
Video 2 - "Exploring Therapies for Platinum Sensitive & Platinum-Resistant Ovarian Cancer"
Related Content
© 2024 MJH Life Sciences
AJMC®
All rights reserved.