A XAI-based Framework for Frequency Subband Characterization of Cough Spectrograms in Chronic Respiratory Disease
A new AI model analyzes cough spectrograms, identifying key frequency bands to distinguish chronic respiratory diseases.
A research team led by Pablo Casaseca-de-la-Higuera developed an explainable AI (XAI) framework for diagnosing chronic respiratory diseases like COPD. It uses a Convolutional Neural Network (CNN) trained on cough sound spectrograms and occlusion maps to highlight diagnostically relevant regions. The system decomposes these into five frequency subbands, extracting interpretable spectral markers that successfully distinguish COPD from other conditions and chronic from non-chronic patient groups.
Why It Matters
This enables non-invasive, AI-powered diagnostic tools for respiratory diseases, moving beyond 'black box' models to provide doctors with clear, frequency-based evidence.