XAI-Driven Spectral Analysis of Cough Sounds for Respiratory Disease Characterization
Patricia Amado-Caballero, Luis M. San-José-Revuelta, María Dolores Aguilar-García, José Ramón Garmendia-Leiza, Carlos Alberola-López, Pablo Casaseca-de-la-Higuera · 2025
Read XAI-Driven Spectral Analysis of Cough Sounds for Respiratory Disease Characterization by Patricia Amado-Caballero, Luis M. San-José-Revuelta, María Dolores Aguilar-García, José Ramón Garmendia-Leiza, Carlos Alberola-López, Pablo Casaseca-de-la-Higuera (2025) 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/arxiv-2508-14949.txt.
Description
Patricia Amado-Caballero and co-authors use explainable AI to find what sets coughs of different lung diseases apart. Their 2025 paper starts from cough recordings turned into spectrograms, images of sound over time. A neural network classifies them, but its reasons are hidden. Occlusion maps show which parts of the image drive each decision. The team then analyzes those highlighted regions closely.
Weighting the spectra by the maps revealed real differences between disease groups. Coughs from chronic obstructive disease varied more in the key regions. Raw spectrograms, by contrast, showed no clear group differences. Several spectral features were extracted from the highlighted zones. The authors argue the method makes cough analysis more interpretable. It could sharpen diagnosis by pointing to disease specific sound signatures rather than overall patterns.
Catalog record
| CREATOR | Patricia Amado-Caballero, Luis M. San-José-Revuelta, María Dolores Aguilar-García, José Ramón Garmendia-Leiza, Carlos Alberola-López, Pablo Casaseca-de-la-Higuera |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-14949