By Evan Eklund, Director & EMR Practice Lead, Temus
AI is everywhere — from boardrooms to headlines — but as the excitement around AI and the advancements of foundation models grows, two truths are emerging clearly. First, AI is only as powerful as its proximity to and integration with the data and operations that drive it. Second, the governance and regulatory frameworks must be in place to enable prolific adoption and continued innovation.
In healthcare, this is no exception, and in some Asian markets, like Singapore, the building blocks are starting to fall into place. From a technology lens, we have the Frontier Labs like Anthropic and Google making domain-specific initiatives in health. Enterprise electronic medical record (EMR) platforms already offer more than 30 predictive AI models, including ambient clinical documentation, AI-assisted decision support, and patient-facing virtual assistants that already trained on and integrated with clinical and patient data and workflows. From a governance lens, Singapore’s Budget 2026 named healthcare as one of its four national AI mission sectors and several regulatory pushes have already put foundations in place for transformation.
The Health Information Bill, passed in January 2026, will require all licensed providers to share key patient data with the National Electronic Health Record (NEHR). And just this month, Health Minister Ong Ye Kung launched AIHGle 2.0 — Singapore’s updated AI in Healthcare Guidelines — while Singapore’s Health Sciences Authority (HSA) became the first regulatory authority in the world to achieve the highest WHO maturity level for medical device regulation. These are transformative starts.
The combination of the technology and emergence of governance frameworks will help clear the path for AI to go from available to deployed.
Based on nearly two decades of implementing health IT, what will accelerate impact will not be just more technology but to address the common barriers.
Two Barriers That Aren’t About Technology
The regulation will need to continue to evolve:
HSA classifies clinical AI software as a medical device under the Health Products Act — and rightly so. But this creates a practical deployment gap. EMR (Electronic Medical Record) vendors with proprietary AI models — for sepsis detection, readmission risk, patient deterioration — must engage with HSA’s regulatory process, including disclosure of algorithmic methodology and lifecycle governance under the guidelines. Until that process completes, health clusters and private providers cannot deploy EMR-native AI, even when the tools are technically available in their systems.
The result: health clusters source clinical AI from third-party vendors who have independently secured HSA approval — often duplicating functionality already within their core EMR. To its credit, HSA is moving proactively. The newly launched AIHGle 2.0 guidelines explicitly feature regulatory sandboxes for testing AI in real-world clinical settings, including an AI-SaMD (Software as Medical Device) Exemption Sandbox allowing public healthcare institutions to deploy low-to-moderate-risk AI devices without full registration. HSA has also updated Clinical Decision Support Systems (CDSS) guidelines, clarifying that decision support based solely on established clinical guidelines falls outside regulatory scope. These are substantial steps. But the broader challenge remains: bringing global EMR vendors into this evolving process so that AI tools already embedded in health clusters’ core systems can be deployed alongside locally developed solutions.
The infrastructure gap:
Clinical AI requires real-time data integration, AI-capable computing environments, and frameworks that deliver outputs into clinical workflows. Singapore’s centralised infrastructure provides a strong foundation, but gaps persist around real-time patient data access and the ability to build custom analytics on the EMR platforms. The Health Information Bill’s mandate for universal NEHR contribution will create one of the most comprehensive health data ecosystems in the region, which is an incredible asset. But connecting that data to point-of-care AI applications still requires investment.
Where Singapore Can Start Building Now
Rather than waiting for every barrier to resolve sequentially, providers can begin laying the groundwork for AI-enabled care today. Here are four practical scenarios.
Scenario 1: Operationalising preventive care with predictive population health
Healthier SG has enrolled 1.3 millions of residents with GPs as of 2025, but identifying who needs early intervention still relies heavily on manual clinical judgement. AI-powered risk stratification — drawing on data flowing into NEHR and EMRs — could flag patients at rising risk for diabetes complications or cardiovascular events, enabling their GP to intervene before an acute episode. The technology exists within EMR platforms and from third-party vendors. The opportunity is pairing these models with clinical workflows that make predictions actionable at the GP level. On tools and technology today, it is already possible to build population health-level dashboards.
Temus recently built for one public health cluster an interactive tool that enabled Population Health teams to understand the flow of patients across different care sites and institutions, broken down by encounter type & demographics, pulling data across platforms (e.g., CRM, EMR). This has enabled the Population Health teams to move from fragmented, institution-level reporting to a unified view of care consumption and health outcomes across the resident population. They are now able to surface care gaps, track clinical indicators, and enable targeted outreach campaigns to reach the right patients at the right time.
Scenario 2: Supporting care coordination for a super-aged society
Ong Ye Kung confirmed recently, Singapore has officially crossed the 21% threshold — we are now a super-aged society. Programmes like Age Well SG and MIC@Home are shifting care into the community, but clinicians managing patients across polyclinic, specialist outpatient, community hospital, and home care settings often lack a real-time, consolidated view of a patient’s trajectory. AI-enabled coordination tools could synthesise data across touchpoints to highlight readmission risk, flag medication conflicts across providers, or prompt timely transitions of care. Clusters that invest now in the governance and change management for these tools will be better prepared to scale efficiently and safely to respond to these demographic pressures.
Scenario 3: Building AI governance capability now
Perhaps the most immediate opportunity requires no new technology. Each health cluster can begin establishing a unified AI governance function — clinicians, informaticists, and operational leaders evaluating, prioritising, and sequencing AI deployments based on clinical impact and organisational readiness. This team would develop a consolidated AI strategy, engage proactively with HSA’s evolving regulatory pathways, and build the change management muscle that turns AI into adopted practice. Clusters that build this capability now will move fastest when the landscape matures
Scenario 4: Investing in the people who make AI work at the bedside
Rather than treating change management as a training session bolted onto the end of an AI implementation, clusters can start building clinical AI adoption capability now. This means embedding clinicians in the design process — co-developing the alert logic, escalation pathways, and response protocols that determine whether a prediction actually changes behaviour. It means role-specific AI fluency and confidence, so a bedside clinician understands why a risk score fired and what to do next. And it means feedback loops where frontline staff report when AI recommendations do not match clinical reality. The clusters that see the greatest return on clinical AI will invest as heavily in adoption as in deployment.
The Advantage Singapore Must Not Waste
Singapore has something most health systems do not: health systems on a shared national EMR strategy, a single regulator that just became the first in the world to achieve WHO’s highest maturity level for medical device regulation, and a National AI Council at the highest level of government. A regulatory pathway cleared once in Singapore benefits every public institution. A governance framework at one cluster can be adapted across all three.
The signals from government are clear. Minister Ong Ye Kung has explicitly invited private sector partners to collaborate with public health institutions on AI and is aligning MOH agencies into coordinated pathways for priority disease areas like cardiovascular disease and diabetes. This is exactly the multi-stakeholder coordination the AI deployment challenge demands.
But policy signals must be met with operational readiness: cluster-level governance that can evaluate and sequence AI deployments, infrastructure connecting the NEHR’s expanding data to point-of-care applications, and sustained investment in the people who will decide whether AI succeeds or fails at the bedside.
In my view, Singapore has the structural conditions and co-ordinated intent that most health systems spend decades trying to build. The next test is whether that same government commitment extends beyond technology investment to the people and processes that determine how AI can change work at the point of care and empower those who have to make it work on the ground.
