Singapore Researchers Introduce PRIMARY-AI Framework in Nature Medicine to Set New Standards for AI in Primary Care

Breakthrough in Safeguarding Primary Care with PRIMARY-AI Framework

Singapore-based researchers have made headlines with their groundbreaking publication in Nature Medicine, introducing the PRIMARY-AI framework designed specifically to evaluate and standardize artificial intelligence (AI) applications in primary care. This perspective piece, published on February 11, 2026, addresses a critical gap: the rapid deployment of AI tools in primary care—such as diagnostic support, risk stratification, triage, and clinical documentation—without robust outcomes-based evaluation standards. Primary care, defined as the first point of contact for patients with the health system, handles everything from routine check-ups to managing chronic conditions, serving as the foundation for equitable healthcare delivery worldwide.

The framework poses a fundamental question: Does a given AI tool truly improve patient outcomes and strengthen primary care functions like continuity of care (ongoing relationships with patients), coordination (seamless integration across services), comprehensiveness (holistic coverage of needs), and people-centered care (tailored to individual contexts)? Without such standards, AI risks exacerbating existing pressures, including clinician burnout and health inequities.

Singapore's Leadership in AI-Driven Healthcare Innovation

Singapore, with its aging population—projected to have one in four residents over 65 by 2030—faces mounting demands on primary care systems. The Ministry of Health (MOH) has been proactive, launching initiatives like Healthier SG in 2023 to shift focus toward preventive, relationship-based care. AI plays a pivotal role here, with tools automating documentation and accelerating diagnostics, such as AI-powered diabetic retinopathy screening that has screened millions nationwide.

The country's Model AI Governance Framework, first released in 2019 and updated for generative and agentic AI, provides a blueprint for ethical deployment, emphasizing transparency and accountability. This aligns seamlessly with PRIMARY-AI, positioning Singapore universities like the National University of Singapore (NUS) and Nanyang Technological University (NTU) at the forefront. Researchers affiliated with NUS, Duke-NUS Medical School, and Singapore Eye Research Institute co-authored the paper, underscoring higher education's role in global health advancements. For those exploring careers in this space, opportunities abound in research jobs blending AI and medicine.

Key Researchers Behind the PRIMARY-AI Initiative

Leading the effort are prominent figures like Prof. Tien Yin Wong, corresponding author and executive director at Tsinghua Medicine while holding key roles at NUS and Duke-NUS. A President's Scholar from NUS, Prof. Wong specializes in ophthalmology and AI applications in healthcare, with over 600 publications. Co-corresponding author Prof. Josip Car from King's College London brings expertise in digital health, complemented by a multinational team including experts from Imperial College London, Tsinghua University, and Singapore's MOH Office for Healthcare Transformation.

Equal contributors Dian Zeng and Lorainne Tudor Car highlight the collaborative nature, drawing from Singapore's polyclinics and global primary care models. Their work builds on Singapore's AI ecosystem, supported by the National Research Foundation's S$37 billion RIE2025 plan investing heavily in health tech.

Challenges Facing Primary Care Amid AI Adoption

Primary care worldwide grapples with workforce shortages—Singapore's family physicians per capita lag behind needs amid rising multiple long-term conditions (MLTCs), affecting 30% of adults over 60. Algorithmic bias in AI can widen inequities, particularly for underserved groups, while misaligned workflows contribute to clinician burnout, with surveys showing 50% of primary care doctors experiencing high exhaustion levels.

In Singapore, polyclinics handle 1.5-1.9 million chronic disease visits annually, straining resources as the population ages rapidly. AI promises relief but without standards like PRIMARY-AI, it could fragment therapeutic relationships, disrupting the continuity essential for managing MLTCs like diabetes and hypertension prevalent in 11% and 24% of adults, respectively.

• Workforce gaps: Shortage of general practitioners amid 5% annual demand growth.
• Bias risks: AI trained on unrepresentative data disadvantaging minorities.
• Burnout drivers: Administrative overload, where AI could automate 20-30% of tasks if properly evaluated.
• Inequity amplification: Aging population with MLTCs projected to double by 2040.

Core Components of the PRIMARY-AI Framework

PRIMARY-AI shifts focus from technical accuracy to holistic outcomes, evaluating AI across four pillars derived from WHO's primary health care framework: continuity, coordination, comprehensiveness, and people-centeredness. It identifies gaps in existing tools like STARD-AI, which overlook primary care specifics (as shown in the paper's Figure 1).

Implementation involves validated instruments and toolkits for developers, regulators, and clinicians. Step-by-step: (1) Define AI use case; (2) Assess impact on core functions; (3) Measure patient/clinician outcomes; (4) Iterate based on real-world data. This ensures AI enhances rather than undermines care.

Singapore's context is ideal for piloting, with MOH's regulatory sandbox for AI-as-medical devices allowing safe testing in public healthcare institutions.

Real-World Applications and Case Studies in Singapore

Singapore exemplifies PRIMARY-AI's potential. The AI Medical Imaging platform processes scans across institutions, reducing wait times by 30%. In primary care, GenAI tools summarize notes, freeing doctors for patient interactions—a direct alignment with people-centered care.

Project ENTenna at Ng Teng Fong General Hospital uses AI for real-time communication, while Healthier SG integrates AI for personalized preventive plans. A Duke-NUS study showed AI outperforming humans in retinopathy detection, but PRIMARY-AI would validate its primary care integration to avoid silos.Read the full Nature Medicine paper

Stakeholder Perspectives and Global Implications

Clinicians praise PRIMARY-AI for addressing burnout, with Singapore surveys revealing a 54% knowledge-attitude gap in AI readiness. Regulators like Singapore's Health Sciences Authority (HSA) emphasize governance, aligning with MOH's AI in Healthcare Guidelines.

Globally, as low-resource settings adopt AI, PRIMARY-AI prevents pitfalls seen in biased triage tools. For academics, it opens research avenues; explore tips for academic CVs in AI-health fields.

Future Outlook: Implementation and Research Roadmap

The authors plan a modified Delphi consensus in 2026 to refine PRIMARY-AI, with toolkits forthcoming. Singapore's S$1 billion National AI Strategy to 2030 will accelerate adoption, potentially reducing MLTC complications by 15-20% through optimized AI.

Challenges remain: upskilling workforce via NTU and NUS programs, ensuring data privacy under PDPC guidelines. Optimistically, PRIMARY-AI could redefine primary care, making it resilient for aging societies.

Photo by Galen Crout on Unsplash

• 2026: Consensus study and toolkit release.
• 2027+: Pilots in polyclinics, global benchmarks.
• Long-term: Integrated AI ecosystems reducing costs by 10-15%.

Actionable Insights for Healthcare Professionals and Policymakers

To leverage PRIMARY-AI: Start with audits of existing AI tools against the four pillars. Train via Singapore's upskilling initiatives. For job seekers, higher ed jobs in AI-health are booming at Duke-NUS and NUS.

Policymakers should mandate PRIMARY-AI compliance, fostering public-private partnerships like SingHealth-Duke-NUS. Ultimately, this framework ensures AI serves humanity, not supplants it.

Engage further with Rate My Professor for insights on AI-health faculty or career advice. Singapore's innovation cements its status as a global leader.