Periodontal disease affects a large share of adults and, left unmanaged, contributes to tooth loss and systemic conditions such as diabetes and cardiovascular disease. Current practice relies on static, periodic assessments, including probing depths, bleeding scores, and radiographs taken at isolated visits, that limit clinicians' ability to anticipate how a patient's periodontal condition will progress or to tailor treatment to their individual trajectory. A digital twin that continuously integrates a patient's periodontal data into one evolving model enables earlier risk detection and more personalized periodontal care.
DT4OralHealth builds a patient-specific digital twin at the level of individual teeth, gum regions, and jawbone, populated with longitudinal periodontal data spanning demographics, medical history, clinical dental findings, oral hygiene behavior, periodontal radiographic imaging, and relevant biomarkers. This data feeds a pretrained ML model that produces a personalized periodontal risk score and identifies its key drivers using explainable AI (e.g., SHAP). The platform will also simulate interventions, such as improved hygiene, glycemic control, and smoking cessation, to estimate their effect on future periodontal risk.
The platform is expected to produce interpretable, individualized periodontal risk scores that align with observed clinical outcomes, along with ranked risk factors per patient to support periodontal treatment decisions. Intervention simulations are anticipated to show measurable projected reductions in periodontal risk, giving clinicians a quantitative basis for patient counseling and treatment prioritization. This work is ongoing, and validation results will be included as they become available.