Listens for Coughs After Lung Surgery With Near-Perfect Accuracy

A research team in China is testing an AI system that could make postoperative cough monitoring more objective after lung surgery. Published in Biomedical Engineering Letters, the study describes a two-stage deep learning framework that continuously listens to raw audio, detects cough events, and classifies each cough as dry or productive with high reported accuracy. The motivation is clinical: persistent coughing after lung resections like lobectomy or pneumonectomy is linked to issues such as air leaks, sore throat, sleep disruption, and delayed recovery, yet current hospital assessments are intermittent and subjective. Researchers built the dataset at Tongji Hospital in Wuhan, collecting 7,000-plus recordings from 80 patients between June 26 and August 13, 2024. After processing, 1,908 one-second clips were used, and evaluation used a patient-level split.