Using Audio Spectral Features and Machine Learning for Breathing Pathology Detection

We will use audio waveform signatures to determine whether breathing pathologies can be distinguished. More specifically, we will determine breathing audio features and use machine learning algorithms to develop diagnostics for respiratory diseases. We will be tuning previously developed machine learning algorithms based on neural networks and will provide comparative results in terms of performance, confusion matrices, and computational complexity. Moving forward, we need to identify an open-sourced database that we can utilize to conduct this study.