CLINICAL VALIDATION OF SWAASA ARTIFICIAL INTELLIGENCE PLATFORM USING COUGH SOUNDS FOR SCREENING AND DIAGNOSIS OF RESPIRATORY DISEASES
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Introduction
Analysis of cough sounds have the potential to give a clue regarding the underlying respiratory disease. The Swaasa AI platform using artificial intelligence technology can analyze the cough sounds and provides an output using cough as a marker. The Swaasa AI platform comes under the category of Software As a Medical Device (SaMD), the device can detect underlying respiratory condition as normal vs abnormal, the pattern of disease condition (obstructive, restrictive or normal), the severity of pattern (mild, moderate, severe) and presence of disease conditions for Asthma, COPD, ILD and Bronchiectasis.
Methods
Patients (age >= 18 years) presenting to our departments with respiratory symptoms were prospectively recruited into the study. The patient’s cough sound was recorded by a mobile device in which the Swaasa AI platform was installed. Spirometry was done and patients’ clinical diagnosis was taken as reference standard for comparison. Polychotomous variables were analyzed in binary form as normal and abnormal. Multi-class outcomes were defined as obstructive, restrictive, or normal pattern, mild-moderate-severe as pattern severity or disease conditions like Asthma, COPD, ILD and Bronchiectasis.
Results
We recruited 2179 patients with respiratory diseases and 827 normal individuals. Sixty-two percent of them were males. The Swaasa AI platform has 90% accuracy in distinguishing normal versus abnormal (having respiratory condition) as well as distinguishing normal from abnormal spirometry. The performance was consistent across demographic parameters such as age, gender, and BMI. The accuracy for multi-class outcomes were between 79%-84% for spirometry pattern and 76%-84.5% for pattern severity and 83%-93% for disease type classification.
Conclusion
Swaasa commends an elevated level of precision in classifying normal and abnormal conditions across demographic subgroups indicating its promise as a screening tool. While the accuracy measures emphasize further refinement to ensure consistent high performance across disease scenarios and severity levels, the study shows Swaasa’s reasonable potential in identification of specific respiratory diseases and their severity.
