Improving respiratory disease detection through SSL-enhanced acoustic analysis and exercise-rest measurements

1

Voice analysis is a non-invasive method for monitoring respiratory and systemic health, but subtle changes are hard to detect at rest.
2

The study evaluates a model combining stress-induced acoustic analysis with machine learning to enhance diagnostic sensitivity.
3

Using the DICOPERIA-Voice dataset, recordings were collected at rest and after physical exertion to assess vocal and respiratory signals.
4

Fusion of traditional acoustic features with self-supervised learning embeddings improved classification performance, achieving peak F1-scores of 82.2% and 80.8%.
5

Post-exercise measurements and self-supervised learning representations enhance the robustness of voice-based screening for respiratory disorders.

Frontiers In Medicine

by Álvaro Vera-López, Darío Tilves-Santiago, José Manuel Ramírez-Sánchez, Laura Docío-Fernández, Carmen García-Mateo, María Bustillo-Casado, Alejandro García-Caballero
June 24, 2026

1