Research

Department programs for diabetes AI healthcare

Our themes focus on transforming complex diabetes signals into reliable, interpretable insights that can support prevention, monitoring, and clinical decision pathways.

Multi-modal diabetes modeling

Fuse longitudinal clinical records with lab and biomarker signals for robust stratification.

Longitudinal data

Interpretability

Early risk detection

Identify individuals at elevated risk to support earlier intervention strategies.

Risk scoring

Clinical validation

Biomarker discovery pipelines

Evidence-tracked candidate generation with reproducibility and quality checks.

QC workflows

Reproducibility

Clinical decision support

Design decision-support artifacts aligned to clinician workflows and outcomes.

Decision alignment

Outcome metrics

How we deliver progress

Phase 1

Data governance & cohort definition Clear inclusion criteria, privacy constraints, and measurable endpoints.

Phase 2

Modeling & validation Model development with evaluation aligned to diabetes-relevant outcomes.

Phase 3

Interpretation & governance Transparent behavior checks, documentation, and responsible deployment guidance.

Discuss a research collaboration Review our publication list Read our standards

FAQ

Do you publish research outputs?

Yes. We aim to share method notes, validation learnings, and high-level results when permitted by governance and partnership agreements.

What does “multimodal” mean here?

We combine multiple signals (for example: clinical time series plus laboratory or biomarker measurements) to improve robustness and interpretability.

How do you measure clinical relevance?

We focus on endpoints and evaluation strategies that reflect clinically meaningful outcomes, not only statistical metrics.