Diagnostic hardware, 2025 to 2026
CoughScope
A handheld, low-cost COPD screener that reads airflow and lung sound together on an ESP32, then classifies the signal with lightweight machine-learning models.
- 92.5%
- accuracy separating COPD from non-COPD cases
- 56
- clinical participants
- Gold
- CREST Award, June 2026
My part: Built the prototype and its signal pipeline, compared four classifiers and took the device to two NGOs, which agreed to on-site testing at their health camps.
How it works
A pressure sensor captures forced airflow and a microphone records lung and cough sounds. The signals are cleaned of drift and noise, normalized, then turned into DFT and MFCC features alongside spirometry measures such as FEV1, FVC and peak flow. I compared an SVM, K-nearest neighbors, a decision tree and a shallow neural network. The model separates COPD from non-COPD cases.
