Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?

Currently accepted at: Journal of Medical Internet Research

Date Submitted: Nov 26, 2025
Date Accepted: May 20, 2026

This paper has been accepted and is currently in production.

It will appear shortly on 10.2196/88517

The final accepted version (not copyedited yet) is in this tab.

Machine and Deep Learning Techniques for Adventitious Respiratory Sound Classification: A Scoping Review

  • Khadija Manzoor Hussain Cheema; 
  • Khaled AlThelaya; 
  • Jens Schneider

ABSTRACT

Background:

Respiratory disorders pose a major threat to human health, making prompt and precise diagnosis essential. Traditional methods for identifying and monitoring respiratory issues are often slow, costly, and require specialized medical expertise. Increasingly, artificial intelligence (AI) technologies are being leveraged to assist in diagnosing respiratory conditions through machine learning and deep learning techniques. These systems employ sophisticated algorithms and neural network models to automatically detect and classify both normal and abnormal sounds.

Objective:

The purpose of this research is to examine and classify AI-driven techniques for detecting and categorizing respiratory sounds. Furthermore, this study analyzes the dependability of chosen studies by assessing the relationships among dataset size, the categories of respiratory sounds, and the performance metrics employed to evaluate these models.

Methods:

We carried out a systematic search for studies utilizing the SCOPUS, PubMed, Association for Computing Machinery Digital Library (ACM DL), and Institute of Electrical and Electronics Engineers (IEEE) Xplore databases, following the guidelines set by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR). The studies that were included in this review concentrated on the analysis of respiratory sounds, particularly in identifying or classifying abnormal and normal lung sounds through AI technologies. The selection of studies and extraction of data were conducted independently by two reviewers, and the data were synthesized narratively, categorizing studies according to the diagnostic AI techniques and evaluation metrics used.

Results:

We retrieved 1474 papers from the 4 databases, of which 175 were eligible for this review. Shallow AI-based techniques were used in 58 studies, and deep AI-based techniques were used in 117 studies. The studies used up to 11 evaluation metrics to assess the proposed models, where 110 studies used accuracy as the primary evaluation metric. Overall, studies that used smaller data sets reported higher accuracy.

Conclusions:

This study evaluates several AI-driven models for analyzing respiratory sounds but finds direct comparisons difficult due to diverse evaluation of metrics and sound data types. The performance is influenced by dataset size, sound class variety, and modeling techniques. Thus, the reliability of shallow and deep models with high accuracy scores was uncertain, given their training on small, less diverse datasets.


 Citation

Please cite as:

Cheema KMH, AlThelaya K, Schneider J

Machine and Deep Learning Techniques for Adventitious Respiratory Sound Classification: A Scoping Review

Journal of Medical Internet Research. 20/05/2026:88517 (forthcoming/in press)

DOI: 10.2196/88517

URL: https://preprints.jmir.org/preprint/88517

Download PDF


Request queued. Please wait while the file is being generated. It may take some time.

© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.