MACHINE LEARNING MODELS FOR TRACINGCOMMUNICABLE LUNG DISEASES

Abstract

Communicable lung diseases such as tuberculosis, pneumonia, and COVID-19 remain major global health concerns due to the high transmission and mortality rates, particularly in developing countries like Nigeria, where healthcare systems face persistent challenges with early detection, diagnosis, and effective control. Traditional disease-tracing and monitoring methods often fail to accurately predict real time infection severity, thereby hindering timely response, appropriate treatment, and the efficient allocation of medical resources. This project developed a single, unified machine learning model capable to distinguishes disease type (communicable and non-communicable across multiple lung conditions),classifies severity (mild/moderate/severe) in real time, and triggers public-health actions for automatic isolation built into the real working system To achieve this, a Machine Learning models was designed, combining K-Modes clustering for grouping patients with similar symptoms from categorical datasets with a Random Forest classifier for categorizing disease severity levels into mild, moderate, or severe. This integration enables both unsupervised grouping and supervised prediction within a single analytical framework. Developed using the Object-Oriented Analysis and Design Methodology (OOADM) and implemented in Python, the system utilizes Excel as the primary data source, Excel functions as a simple storage medium accessed and managed through Python libraries, particularly Pandas, which is employed for statistical computation and detailed data analysis. SQLite database engine was used to store user biodata and profile for system authentication and authorization. The system allows real-time monitoring by analyzing patient symptoms and vital parameters upon data entry, providing immediate diagnostic and severity-level feedback without the need for laboratory tests. This facilitates rapid clinical decision-making by healthcare professionals and supports proactive public health interventions. The model achieved an accuracy of 95.6%, precision of 96.1%, recall of 95.2%, and an F1-score of 92.2%, demonstrating reliability in predicting infection severity. Ultimately, the application enhances disease surveillance, enables early isolation of high-risk patients, and strengthens national healthcare response systems