Artificial intelligence adoption in Asia-Pacific hospitals is accelerating quickly, but the technology remains anchored to administrative work rather than clinical care, according to a new report from HIMSS.
The survey found that 46% of healthcare organisations in the region began implementing AI within the past 12 months, while only 23% have used it for more than three years. Despite that short history, usage is already frequent: 51% of respondents reported daily or near-daily use, and another 25% said they use AI several times a week.
Administrative tools dominate
Generative AI tools lead the way, used by 81% of respondents, far outpacing large language models (47%) and medical imaging applications (38%). But the use cases skew heavily operational. Workflow optimisation (74%), medical documentation (65%) and administrative efficiency (64%) topped the list of areas where organisations seek value, and 80% cited improved operational efficiency as the most realised benefit.
Clinical outcomes trail well behind. Only 45% reported improved diagnostic accuracy, and just 33% cited enhanced patient engagement.
The pattern holds across the region, suggesting hospitals are comfortable letting AI for Healthcare handle back-office burdens before it touches patient-facing decisions.
Cost and privacy are the brakes
Financial cost is the leading barrier to broader adoption, cited by 67% of respondents, followed by data privacy (58%) and data security (52%). Investment plans vary sharply by market. Singapore (73%), South Korea (68%) and Thailand (67%) reported planned two-year AI spending above US$150,000. By contrast, nearly half of respondents in India and Indonesia (47% each) reported budgets below US$25,000.
Funding sources are split across digital transformation or innovation funds (29%), government grants (24%) and internal operational budgets (19%).
Support gaps for GenAI and LLMs
Generative AI and LLM tools show the largest gap between adoption and organisational support. While they are the most widely used category, 38% of respondents identified them as needing additional backing - more than any other tool type - citing requirements around prompt engineering and ethical governance frameworks.
Workforce anxiety about displacement is low. Nearly half of respondents (49%) said they are "not very concerned" or "not concerned at all" about AI reducing staffing needs, and qualitative feedback consistently framed AI as a "co-pilot" rather than a replacement.
On training, hands-on workshops with live demonstrations were preferred by 60% of respondents, ahead of on-the-job coaching (51%) and case-based simulations (48%). Respondents also identified a user-friendly interface (67%), seamless system integration (57%) and step-by-step usage guidelines (53%) as the most effective adoption enablers.
Governance largely relies on existing national frameworks rather than AI-specific regulation. Some 34% of organisations cite national healthcare data protection laws as their primary governance reference, 25% point to Ministry of Health guidelines, and only 21% reference AI-specific national or regional regulatory frameworks.
Why this matters for healthcare professionals
The gap between administrative and clinical AI use is not a failure - it reflects where the technology is ready. For clinicians, the practical implication is that AI will keep reshaping documentation and workflow tasks first, and diagnostic support later. Professionals who build familiarity with GenAI tools now, particularly through hands-on practice, will be better positioned as those systems expand into clinical decision-making.
Your membership also unlocks:
