A real-time biosurveillance mechanism for early-stage disease detection from microblogs: a case study of interconnection between emotional and climatic factors related to migraine disease

For many years, certain climatic factors have been used to predict potential disease outcomes of relevance to humans. This is because early discovery of disease (or its symptoms) would help people or healthcare professionals to take the necessary precautions. Since microblogs can be used to create new connections and maintain existing relationships,disease detection in microblogs is still considered a serious problem for many healthcare systems, especially for establishing a successful epidemic recognition procedure. To tackle this issue, this study proposed a novel tracking approach to diagnose illnesses in microblogs. It is based on the interconnection between certain emotional type and climatic factors associated with a specific disease (e.g., migraine). In this study, detailed migraine data were collected from Twitter. We used K-means and Apriori algorithms to extract migraine-related emotions and investigate the potential associations between migraine symptoms and climatic factors. The results showed that sad emotions were highly interrelated with migraine symptoms. The classification results showed that Sequential Minimal Optimization (SMO) was efficient (95.53% accuracy) in detecting the migraine symptoms from Twitter. The proposed mechanism can be used efficiently in biosurveillance systems due to its capability in identifying the hidden symptoms of a sickness on microblogs. This study paves the way to discover disease-related features using both emotional and climatic factors.