Machine Learning-Based Asthma Diagnosis Prediction Using Lung Function and Demographic Features

Asthma is a prevalent chronic respiratory disease, which poses significant diagnostic challenges because of its multifactorial nature. This study aims to develop a machine-learning approach for predicting asthma diagnosis using key features such as body mass index (BMI), age, lung function parameters (FEV1 and FVC), and demographic information. A dataset containing clinical and demographic records was utilized to train and evaluate models, including Random Forest, Neural Networks, and XGBoost classifiers. The performance of the following models was assessed using metrics such as precision, recall, accuracy, and F1-score, with Random Forest exhibitingthe highest predictive performance. In addition to traditional performance metrics, advanced visualization techniques like SHAP (Shapley Additive ex Planation’s) values were employed to interpret model predictions and assess feature importance. Results demonstrate that age, BMI, and lung function are key predictors of asthma diagnosis, with lung function parameters showing the strongest correlation with diagnosis outcomes. The study also explores various 3D and interactive visualizations to enhance the interpretability of the models. The proposed approach demonstrates that machine learning models when combined with clinical data, can accurately predict asthma diagnosis and potentially aid healthcare professionals in early detection and personalized treatment plans. This research highlights the potential of data-driven models in improving asthma diagnosis and contributing to better clinical decision-making.