Beyond the traditional models: a network reconstruction based model for predicting and analysing individual health status

The network reconstruction approach holds broad significance across various domains due to its ability to unveil intricate relationships within complex systems. By elucidating connections and dependencies, it enhances understanding of diverse datasets, promoting more accurate predictive analytics and contributing to refined predictions. In recent years, the surge in studies focused on human health prediction models underscores its critical importance. Accurate disease prediction empowers healthcare professionals to make decisions, facilitating precise patient care. Timely identification of diseases motivates individuals to ensure timely medical interventions and adopt lifestyle changes when necessary. Despite advancements in health prediction models using machine learning or data mining techniques, existing models primarily predict current health status without effectively identifying underlying causes. A novel health prediction model using a network reconstruction approach is proposed to predict an individual’s health status. In the proposed model, data features are represented as network nodes, and their interrelationships are depicted as edges. Further, a novel methodology is presented for calculating the decision parameter, \(\alpha\) $\alpha$ to predict an individual’s health status. The proposed model identifies key features influencing health and offers corrective actions for improving health status. The proposed predictive model applies to the hepatitis C and diabetic disease datasets, demonstrating significant enhancements in accuracy over the current state-of-the-art models.