ATF-MGIAM: Medically-guided interpretable attention mapping for robust pertussis cough sound recognition
Siheng Zhang, Yan Xu, Ying Xiao · 2026
Read ATF-MGIAM: Medically-guided interpretable attention mapping for robust pertussis cough sound recognition by Siheng Zhang, Yan Xu, Ying Xiao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0348508.txt.
Description
Siheng Zhang, Yan Xu, and Ying Xiao trained a deep learning system to recognise whooping cough from recordings of coughing. The disease's distinctive cough is a diagnostic clue, but computer listeners usually cannot explain their verdicts. Their 2026 framework fuses sound patterns across time and frequency, then steers its attention toward the parts of the sound doctors find medically meaningful, making its reasoning visible.
Tested on three small public datasets, the system scored near-perfect accuracy, a couple of points above the best rivals, and barely degraded when noise was added. Removing either main component hurt performance, confirming both contribute. The authors offer the design as a reusable pattern for medical sound analysis.
Catalog record
| CREATOR | Siheng Zhang, Yan Xu, Ying Xiao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 70342 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |