Digital Twin Framework Models Cognitive Decline Trajectories
A team of researchers has introduced the Personalized Cognitive Decline Assessment Digital Twin (PCD-DT), an advanced framework designed to model disease trajectories tailored to individual patients using sparse, noisy, and irregular longitudinal data. This framework combines latent state-space models to capture personalized temporal dynamics, integrates various clinical, biomarker, and imaging features, and employs uncertainty-aware validation with adaptive updates. Additionally, conditional generative models are proposed to facilitate data augmentation and stress testing of less common progression patterns. An initial feasibility study examining longitudinal TADPOLE trajectories reveals a distinct differentiation between groups with normal cognition and those with Alzheimer's disease.
Key facts
- The framework is called Personalized Cognitive Decline Assessment Digital Twin (PCD-DT).
- It is multimodal and uncertainty-aware.
- It models patient-specific disease trajectories from sparse, noisy, and irregular longitudinal data.
- It combines latent state-space models, multimodal fusion, and uncertainty-aware validation.
- Conditional generative models are used for data augmentation and stress testing.
- A feasibility study analyzed longitudinal TADPOLE trajectories.
- The study showed clear separation between cognitively normal and Alzheimer's disease cohorts.
- The work is published on arXiv with ID 2604.27217.
Entities
Institutions
- arXiv