The digital twin (DT) paradigm presents a wide array of opportunities for modeling complex systems in biomedical sciences in a realistic manner, allowing researchers and healthcare professionals to explore various “what-if” scenarios. In dental sciences, DTs can serve as virtual replicas of a patient's periodontal tissues and structures, enabling clinicians to address a variety of tasks such as simulating periodontal conditions, forecasting treatment outcomes, and personalizing dental care plans. However, achieving this vision is impossible without building confidence in making DTs in healthcare trustworthy which requires the development of novel mathematical and statistical foundations behind such fundamental questions as verification, validation, and uncertainty quantification (VVUQ) of dental DTs, robustness of dental DTs to uncertainties, and cohesive integration of multi-modal health-related data at disparate scales. This project aims to develop novel mathematical and statistical methodology to establish a foundation of the artificial intelligence (AI)-driven framework for constructing reliable and personalized DTs for periodontal health. By integrating principles from statistical learning, topological data analysis, and generative AI, specifically, probabilistic diffusion models on graphs, the project opens a pathway to build ensembles of individualized dental DTs, termed “periodontal digital siblings.” These DTs will capture patient variability and uncertainty, off