The problem
Respiratory disease is under-detected where stethoscope expertise and imaging are scarce. A low-cost screening aid could prompt earlier referral.
What to build
Develop a research prototype that records chest audio and flags recordings that deserve clinical review, with honest reporting of limits and error rates.
Success looks like
Sensitivity and specificity reported on a public dataset with a bias analysis, and the tool framed as a referral aid rather than a diagnosis.
Suggested technologies
- Audio signal processing
- CNN
- Spectrograms
- Model cards
- Ethical review
An original, independent statement. It is not an official Smart India Hackathon problem statement, and Arventiq Labs is not affiliated with or endorsed by that programme. The scenario is illustrative and does not describe any specific organisation. Free to use with credit.