A digital twin framework for predicting and simulating type 2 diabetes onset using retrospective lifestyle data

Area of Science:

• Digital health
• Epidemiology
• Preventive medicine

Background:

• Type 2 Diabetes Mellitus (T2DM) is a growing global health issue.
• Current predictive tools often require real-time data, limiting use in resource-limited settings.
• Lifestyle and psychosocial factors significantly influence T2DM development.

Purpose of the Study:

• To develop a novel digital twin (DT) framework for forecasting T2DM onset using retrospective data.
• To simulate the impact of preventive interventions on T2DM risk.
• To explore the influence of lifestyle, behavioral, and psychosocial factors on T2DM prediction.

Main Methods:

• Utilized UK Biobank data from 19,774 participants followed for up to 17 years.
• Employed a penalized Cox proportional hazards model with 14 selected predictors.
• Applied causal inference techniques (DAGs, counterfactual simulations) to analyze intervention effects.

Main Results:

• The DT model achieved strong predictive performance (C-index = 0.90).
• Psychosocial stressors (loneliness, insomnia, poor mental health) substantially increased T2DM risk.
• Dietary factors and ethnic disparities also influenced T2DM risk, with significant variations observed.

Conclusions:

• The DT framework offers a transparent, simulation-enabled approach for T2DM risk assessment and intervention planning.
• This method is scalable for public health, especially in low-resource environments.
• Integrating psychosocial and lifestyle data advances equitable and behaviorally informed digital health solutions.