Climate-Aware Healthcare: Predictive Modeling of Disease Outbreaks using Multimodal Data

The weather and climatic conditions have a tremendous impact on the spread and intensity of a disease, making their integration with patient health data crucial for proactive healthcare management to facilitate proactive healthcare management. This study introduces a predictive framework that uses multimodal data (50 attributes of patient health and weather-related features) in a machine-learning model to achieve classification of disease types with higher accuracy. The model also aims at foreseeing possible outbreaks early and, thus, improves healthcare preparedness. Among multiple machine learning algorithms evaluated, XGBoost demonstrated superior performance, achieving 99.2% classification accuracy on a publicly available dataset. In order to improve the interpretability, this research uses Explainable Artificial Intelligence (XAI) methods such as SHAP (SHapley Additive exPlanations), which allow for finding significant contributing factors in the context of disease prediction. The analysis revealed that certain weather conditions (e.g., high humidity in case of Dengue, extreme temperature in case of Heat Stroke), along with patient medical history, are the deciding factors in a correct categorization of the disease. The results highlight the role of combining environmental and clinical data in neutralizing the occurrence of disease epidemics.