Loading...
Loading...
Contributors:
Dataset URI:
10.3389/fmed.2026.1964851.s001IntroductionCough sound analysis can provide clues to underlying respiratory disease. The Swaasa AI platform, a Software as a Medical Device (SaMD), applies artificial intelligence to cough recordings to generate indications, including overall risk (normal vs. abnormal), spirometric pattern (obstructive, restrictive, or normal), pattern severity (mild, moderate, severe), and the presence of asthma, COPD, ILD, and bronchiectasis.MethodsPatients (≥18 years) presenting to our departments with respiratory symptoms were prospectively recruited. Cough sounds were recorded on a mobile device running the Swaasa AI platform, spirometry was performed. Spirometry served as the reference standard for physiological pattern and severity, while physician-established clinical diagnosis served as the reference for disease categories. Polychotomous variables were dichotomized as normal vs. abnormal, & multi-class output for disease condition, pattern severity, and disease categories (asthma, COPD, ILD, bronchiectasis) were defined.Trial registrationCTRI/2021/04/032742; registered on 12/04/2021; date of first enrollment: 19/04/2021; trial end date: 14/09/2023.ResultsWe recruited 2,179 patients with respiratory diseases and 827 normal individuals; 62% were male. The Swaasa AI platform achieved 90% accuracy in distinguishing normal from abnormal (respiratory condition present) and in classifying normal vs. abnormal spirometry, with descriptive performance subgroup analyses showing broadly comparable performance across age and gender groups For multi-class outcomes, the accuracy ranges were: spirometry pattern: 79%–84%, pattern severity: 76%–84.5%, and disease-type classification: 83%–93%.ConclusionSwaasa® demonstrates promising performance in distinguishing normal from abnormal respiratory conditions across demographic subgroups, indicating its promise as a screening tool. While the accuracy measures emphasize further refinement to ensure consistent high performance across specific respiratory disease scenarios and severity levels, the study shows its reasonable potential in identifying specific respiratory diseases and their severity.
Keyword(s):
airflow parameters; cough sound characteristics; Foetal Development and Medicine; machine learning; respiratory conditions.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
Supplementary Data:
No Supplementary Data Available
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)