Cough-Based Lung Infection Detection Using Deep Learning

Abstract

Cough sounds are rich in acoustic features that can reveal underlying respiratory issues.This project aims to develop an automated system for detecting lung infections from cough audio recordings.Using deep learning, particularly Convolutional Neural Networks (CNNs), we extract and classify cough patterns.The audio recordings undergo preprocessing steps, including noise reduction and normalization.Mel-spectrogram representations are generated from these recordings for feature-rich input.Our CNN model learns to differentiate between healthy and infection-related cough sounds.The dataset includes both healthy and infected cough samples, labeled accordingly.We evaluate the model’s performance using metrics like accuracy, precision, recall, and F1-score.Our approach demonstrates high accuracy in distinguishing infected coughs from healthy ones.Such a non-invasive, audio-based diagnostic tool can aid early detection and remote health monitoring.The system’s potential lies in screening large populations, especially in resource-limited settings.Future work involves augmenting the dataset and exploring more robust feature extraction methods.Integration into mobile health applications could make this solution widely accessible.Ultimately, this project highlights the power of deep learning in biomedical acoustics for public health.The findings indicate a promising avenue for rapid, affordable, and scalable lung infection detection.