Sint Multiverse: Prototype Simulator for Generating Synthetic Health Determinant Data for Explainable AI Training

Integrating Explainable Artificial Intelligence (XAI) into healthcare requires reliable and representative datasets to support transparent, interpretable, and ethical models. However, privacy concerns, limited data availability, and population heterogeneity often hinder the training and validation of Artificial Intelligence (AI) systems in clinical contexts. This study proposes a simulation-based approach for the holistic construction of structured synthetic health datasets, grounded in Health Determinants (HDs) and Health Events (HEs), to support the development and continuous training of XAI in health applications. A hybrid methodology was employed, combining system development, case study analysis, and experimental research. The resulting cross-platform simulator models a wide range of human behaviors, clinical conditions, lifestyle patterns, and medication effects over time. Its minimalist data structure currently supports 139 health markers across biological, psychological, demographic, and socioeconomic domains. Interactions between HDs and HEs enable the simulation of dynamic health trajectories, including disease onset, therapeutic interventions, and preventive actions. A dedicated entity system simulates human behavior across extended timeframes, and longitudinal synthetic datasets are generated to capture evolving health profiles. The simulator was validated through multiple scenarios, including a Type 2 Diabetes Mellitus model. It demonstrates the ability to replicate clinically meaningful patterns and support predictive analyses. This framework contributes to ethical AI research by offering a secure, scalable, and customizable environment for developing and testing intelligent systems. It also promotes interdisciplinary collaboration and serves as a valuable resource for training, hypothesis testing, and the exploration of complex health dynamics in a controlled setting.