Singapore

Precision diabetes management with AI-driven nudging

The Empower+ app, powered by HealthGen's unified biomarker platform, was tested in a 997-patient RCT across SingHealth Polyclinics to evaluate how AI-enabled behavioral interventions improve glycaemic control.

The intervention

Across 997 patients, the platform combined personalised nudging, structured behavioral logging, and integration of activity and glucose-related biomarkers to strengthen diabetes self-management in primary care settings.

At the heart of the trial was an AI engine capable of profiling daily lifestyle signals and predicting behavior—reaching over 90% median accuracy in predicting whether a participant would achieve 30 minutes of MVPA or complete an activity log on a given day.

Three-layer precision engine

AI profiling, biomarker integration, and closed-loop feedback

Behavioral AI

Precision-timed nudges based on predicted behavior

Models analyze daily lifestyle signals to predict activity completion with >90% accuracy, triggering personalized nudges at optimal times.

Biomarker streams

Continuous activity + CGM for comprehensive profiling

Weekly analysis of steps, MVPA, and CGM data generates automated insights and personalized notifications that reinforce behavior change.

Closed-loop system

App, wearables, CGM, coaches, and clinicians unified

Glucose variability, lifestyle logs, and AI nudges connect patient app, health coaches, and polyclinic clinicians in a single workflow.

Closed-loop engagement architecture

Real-time biomarkers feed AI models that trigger precision nudges

Behavioral AI
>90% prediction accuracy
Activity Data
Steps, MVPA
CGM
Glucose variability
App Logs
Behavior patterns
Precision Nudges
Optimal timing
Coach Alerts
Intervention triggers
Clinician Insights
Automated summaries

Clinical outcomes

997-patient RCT demonstrated significant improvements in glycaemic control and patient activation

HbA1c reduction at 3 months
0.80%
intervention
vs
0.49% control
HbA1c reduction at 12 months
1.15%
intervention
vs
0.67% control

A hybrid implementation study across three SingHealth Polyclinics similarly showed reductions in HbA1c, blood pressure, and improvements in patient activation measures.

The evidence

Precision nudging—driven by behavioral AI models and real-time biomarker integration—provides a scalable, personalized mechanism to reinforce healthy habits, improve diabetes self-management, and deliver measurable metabolic improvements in routine primary care.

Ready to unify your digital biomarkers?

See how HealthGen can help you ship faster, scale further, and get to insights without rebuilding the stack.