MONITORING OF ASTHMA AND REGULATING BREATHING PRACTICE USING MACHINE LEARNING

Asthma is a chronic respiratory disorder that requires continuous monitoring and timely intervention to preventexacerbations and improve patient outcomes. Traditional asthma management approaches are largely reactive,relying on self-reporting and periodic clinical evaluations, which often delay the detection of critical symptoms.This research proposes an intelligent, Python-based machine learning system for real-time asthma monitoring andregulated breathing assistance. The system employs a microphone sensor to capture respiratory audio signals, whichare then analyzed using advanced machine learning algorithms such as XGBoost, Cat Boost, Light GBM, IsolationForest, and Hidden Markov Models (HMM). These models classify breathing patterns, detect anomalies, andpredict potential asthma-related complications. The extracted features including MFCCs, spectral contrast, andbreathing rates are processed using Librosa, NumPy, and Pandas. A gamified mobile application further enhancespatient engagement by guiding users through breathing exercises using biofeedback from their own respiratory data.The system supports cloud integration for remote monitoring by healthcare professionals and is scalable forwidespread deployment. The proposed approach significantly improves early detection, encourages preventive care,and reduces dependence on clinical supervision, offering a low-cost, non-invasive solution for asthma managementin real-world environments.