Chief Supervisor

Prof. HE Mingguang

Project Title

Synopsis

Aging is a complex and dynamic process influenced by interactions among multiple biological, physiological, lifestyle, and clinical factors. Current healthcare systems mainly rely on fragmented measurements and retrospective clinical assessments, limiting the ability to continuously monitor individual health trajectories and predict future health risks.
This project aims to develop an artificial intelligence-driven multimodal digital twin framework for modelling longitudinal human health states using heterogeneous biomedical data. The proposed framework will integrate multiple data modalities, including physiological measurements, electronic health records, lifestyle information, omics data, imaging-derived features, and other health-related observations, to construct personalized digital representations of individuals over time.
The research will focus on developing robust multimodal representation learning methods capable of handling incomplete, irregularly sampled, and heterogeneous longitudinal data. Advanced deep learning approaches, including attention-based architectures and temporal modelling techniques, will be investigated to learn unified health-state representations and capture dynamic changes associated with biological aging and disease progression.
Furthermore, the project will explore interpretable digital twin modelling strategies to quantify individual health trajectories, estimate biological aging patterns, and support personalized risk prediction and intervention simulation. The expected outcomes include a scalable AI framework for longitudinal health modelling and provide methodological foundations for future precision medicine and preventive healthcare applications