Prospective field deployment of Swaasa® AI for community-based screening of tuberculosis and other respiratory diseases in Andhra Pradesh: a feasibility and implementation study

Introduction Tuberculosis (TB) remains a major public health problem in India. In the state of Andhra Pradesh (AP), the burden of TB is compounded by a shortage of specialists and the lack of scalable, real-time community screening tools. We prospectively implemented Swaasa ® AI, a software as a medical device (SaMD) product, in a real-world community healthcare setting and evaluated its feasibility and user experience for TB screening, with the additional capability to identify individuals at potential risk of chronic obstructive pulmonary disease (COPD) and asthma. Swaasa ® uses machine learning (ML) algorithms to analyze acoustic cough patterns and gather information through guided questions about symptoms. Unlike conventional TB screening, which primarily targets symptomatic individuals, Swaasa ® supports symptom-agnostic community screening. Methods The platform was deployed in a pilot screening program and integrated into existing healthcare workflows. Trained frontline health workers collected cough recordings using smartphones, and the platform generated automated risk assessments for TB, COPD, and asthma. Over a 6-week deployment, a total of 7,879 individuals from vulnerable populations were screened. Based on risk stratification (Green, Amber, and Red) from Swaasa ® , high-risk individuals were referred for confirmatory testing, specifically using nucleic acid amplification tests (NAATs) for TB and spirometry for COPD and asthma. Results Swaasa ® AI identified 780 individuals (~10%) at risk of TB and referred them for confirmatory testing. TB was confirmed in 11 of these individuals, yielding an observed TB yield of 1.41% (11/780). Among symptomatic Swaasa ® -positive individuals, 7 of 23 were diagnosed with TB, corresponding to an observed yield of 30.4%. Importantly, 36% of the identified TB cases were asymptomatic. In addition to TB, the platform identified that 6.5–9.5% of the population was at risk of COPD and 1.6–1.9% at risk of asthma. Further evaluation using spirometry as part of the clinical assessment for COPD and asthma showed that 95% of high-risk individuals identified by Swaasa ® had abnormal findings. The study also demonstrated an excellent system usability score (80 to 93%) and a very high perceived usefulness rating (4.37–5) for the use of Swaasa ® . Conclusion This SaMD model offers a scalable pathway for early identification of respiratory diseases, particularly in populations with limited access to specialist care.