AI-Based Cough Sound Analysis for Early Tuberculosis Screening Using SVM and CNN

This paper presents an intelligent cough sound analysis framework for early tuberculosis screening using machine learning techniques. The proposed system is designed as a non-invasive, portable, low-cost, and rural-friendly solution that analyzes cough audio signals to identify TB and related respiratory abnormalities. The framework consists of cough sound acquisition through a microphone interface, preprocessing stages including noise reduction and normalization, and feature extraction using Mel Frequency Cepstral Coefficients (MFCC), spectrogram-based descriptors, zero-crossing rate, and energy-frequency characteristics. These extracted acoustic features are utilized to train and evaluate machine learning classifiers such as Support Vector Machine (SVM) and Convolutional Neural Network (CNN) for robust cough pattern recognition. The developed model is intended for real-time deployment through a simple user interface, with optional embedded implementation for offline operation in remote healthcare settings. This work demonstrates the potential of artificial intelligence-driven cough analysis as an effective preliminary screening tool for tuberculosis, with future scope for multi-disease respiratory classification and cloud-assisted healthcare integration.