A novel framework for COPD management in cyber-physical systems using machine learning

Chronic Obstructive Pulmonary Disease (COPD) exacerbations pose significant challenges to healthcare systems due to their unpredictable nature and severe impact on patients. Current COPD prediction models often lack real-time capabilities and fail to leverage multi-source data for accurate forecasts. This research proposes a Cyber-Physical System-enabled framework that integrates both primary (clinical) and secondary (online) data sources to predict COPD exacerbations in real-time. The framework employs advanced machine learning techniques, specifically Random Forest and Artificial Neural Networks, for feature selection and prediction accuracy. Statistical validation through ANOVA ensures the harmonization of diverse data sources, enhancing the robustness of the prediction models. Experimental results demonstrate the framework's effectiveness, with key metrics such as accuracy, precision, recall, F1-score, and AUC showcasing its potential for early COPD detection. The proposed system offers proactive healthcare solutions by delivering timely alerts, forecasting exacerbations, and supporting clinical decision-making, ultimately improving patient outcomes and reducing healthcare costs.