ACE-AI

How ACE-AI works:

• Inputs: routine health data captured at the clinic (age, sex, BMI, blood pressure, lifestyle questionnaire, etc.)
• Prediction: probability of developing diabetes / hyperlipidaemia in the next 5–10 years
• Output: clinicians see a risk score plus intervention recommendations
• Closed loop: high-risk individuals are pulled into the Healthier SG proactive health-management workflow

Technically it is not complex — at its core it is a risk-prediction model trained on national health data. But its deployment scale and real-world medical impact far exceed most academic medical AI.

ACE-AI is a representative of the "pragmatic line" in medical AI deployment — not chasing frontier models, but chasing real medical impact.

Its value lies not in algorithmic novelty, but in:

• Data scale: training uses national NEHR data — a sample size larger than any commercial medical AI
• Deployment scale: simultaneous use across 1,400+ clinics nationwide
• Closed-loop value: prediction outputs feed directly into Healthier SG intervention flows, not just "diagnostic assistance"

This kind of "national-grade data + national-grade deployment" medical AI project is rare globally.

ACE-AI is the flagship "national deployment" project of Singapore's medical AI strategy.

In the "seven transmission levers" framework:

• Lever 5 (government adoption): a national-scale AI deployment in the healthcare system
• Lever 3 (industry adoption): a model for medical AI deployment

A take: ACE-AI proves that Singapore's "small country + centralised healthcare system" is a real advantage in medical AI — large countries like the US or Indonesia could not pull off "one nationwide system" even if they wanted to. ACE-AI's success (if it succeeds) will be studied by the global medical AI industry for years.

Worth watching: actual prediction accuracy, clinician adoption (do doctors actually intervene per the AI's recommendation), and improvement in patient outcomes (does incidence among high-risk patients drop) — answers will only become clear in 5–10 years.