Healthcare AI's Dual Reality: ONC HTI-5 Removes Transparency Rules While Eli Lilly Signs $2.75B AI Drug Deal

Introduction

Early November 2026 brought two contrasting yet equally significant developments in healthcare AI. On one hand, the US Department of Health and Human Services' Office of the National Coordinator for Health Information Technology (ONC) published the controversial HTI-5 final rule, removing transparency requirements for AI predictive tools. On the other hand, pharmaceutical giant Eli Lilly and Hong Kong-based AI biotech Insilico Medicine advanced their $2.75 billion AI drug discovery collaboration. Together, these events paint a complex picture of healthcare AI's trajectory: regulatory rollback and commercial acceleration proceeding in parallel, with both opportunities and risks in abundance.

ONC HTI-5 Final Rule: Deregulation Sparks Controversy

Rule Overview

On October 28, 2026, the ONC officially published the "Health Data, Technology, and Interoperability: ONC Deregulatory Actions To Unleash Prosperity" (HTI-5) final rule. The rule's central change is the removal of mandatory AI model-card source attribute transparency standards and risk management requirements for Predictive Decision Support Interventions (Predictive DSI).

What Are Predictive DSIs?

Predictive Decision Support Interventions are AI tools integrated into Electronic Health Record (EHR) systems to assist clinical decision-making, including:

• Predicting patient readmission risk
• Identifying early signs of sepsis
• Recommending medication dosages
• Assessing surgical risk

What Was Removed

Under the previous HTI-1 framework, certified health IT developers were required to provide information about their predictive tools including:

• Training data: Dataset sources and characteristics used to train the model
• Performance metrics: Accuracy, false positive rates across different patient populations
• Limitation disclosures: Known biases and scope restrictions

This information was typically provided in "model card" format, helping healthcare organizations evaluate the suitability and risks of AI tools.

The Controversy

The HTI-5 rule's publication drew widespread criticism from the healthcare community:

Opponents' concerns:

• Removing transparency requirements shifts the burden of evaluating and governing AI models from developers to health systems and providers
• Many healthcare organizations lack the capacity to evaluate the underlying data and performance details of AI tools
• Healthcare organizations will need to seek such information through private contracts with vendors, increasing negotiation costs and information asymmetry

Supporters' arguments:

• Reduces administrative burden, accelerating AI tool development and deployment
• Provides greater flexibility for innovation
• Relies on market competition rather than regulatory mandates to drive transparency

Actual Scope of the Rule

Notably, the HTI-5 final rule was narrower than ONC's initial proposal:

• The proposed rule suggested removing 34 certification criteria; the final version removed only 14
• ONC deferred all proposed changes to the "information blocking" rule
• The rule formally takes effect December 28, 2026, with some implementation timelines extending into 2027 and beyond

Implications for Asia-Pacific

US regulatory developments carry significant reference value for Asia-Pacific. Healthcare AI regulators in Hong Kong, Singapore, Japan, and other jurisdictions are closely monitoring US policy directions. HTI-5's deregulatory approach may influence Asia-Pacific regulators' decisions on AI transparency requirements, though given the diverse healthcare systems and regulatory cultures across the region, policy responses are likely to vary significantly.

Eli Lilly and Insilico Medicine: $2.75 Billion AI Drug Discovery Collaboration

Background

On March 29, 2026, Eli Lilly and Hong Kong-based AI biotech Insilico Medicine announced a strategic collaboration valued at up to $2.75 billion. Building on an initial software licensing agreement from 2023, this deal marks a new phase of commercialization for AI-driven drug discovery.

Insilico Medicine's Core Technology

Insilico Medicine is one of the world's leading AI drug discovery companies. Its core technology platform, Pharma.AI, includes:

• PandaOmics: An AI tool for target discovery that identifies potential drug targets by analyzing multi-omics data
• Chemistry42: An AI system for molecular design that generates and optimizes drug molecules with specific properties

Collaboration Structure and Financial Terms

Strategic Significance

For Eli Lilly:

• Accelerates early drug discovery, compressing timelines traditionally spanning several years
• Gains access to Insilico's AI engine and laboratory automation capabilities
• Establishes competitive advantage in AI-driven drug discovery

For Insilico Medicine:

• Secures significant financial resources to support further R&D
• Validates its technology platform through partnership with a top global pharmaceutical company
• Creates an important success case for Hong Kong's AI biotech ecosystem

The Rise of Asia-Pacific AI Drug Discovery

Insilico Medicine's success represents the rapid rise of Asia-Pacific's AI drug discovery ecosystem. Hong Kong, as a bridge between mainland China and international markets, is emerging as an important hub for AI biotech. Beyond Insilico, several other Asia-Pacific AI drug discovery companies are growing rapidly, including:

• XtalPi (China)
• Preferred Networks (Japan)
• Evidentli (Australia)

Healthcare AI's Dual Reality

Technical Progress and Regulatory Challenges Coexist

The ONC HTI-5 rule and the Eli Lilly-Insilico collaboration together reveal healthcare AI's dual reality:

Technical dimension: AI demonstrates enormous potential in drug discovery, diagnostic assistance, and clinical decision support. The $2.75 billion collaboration scale speaks volumes about industry confidence in AI technology.

Regulatory dimension: How to balance promoting innovation with protecting patient safety remains the core challenge for global healthcare AI regulation. The controversy sparked by HTI-5's deregulatory direction shows this balance point has yet to be found.

The Importance of Patient Trust

According to a Pew Research Center survey from June 2026, 60% of American patients are uncomfortable with AI being used for diagnosis and treatment, and 91% expect to be asked for permission every time AI is used in their care. Against this backdrop, removing transparency requirements may further exacerbate patient concerns about healthcare AI.

Conclusion

These two significant November 2026 developments — the ONC HTI-5 rule and the ongoing Eli Lilly-Insilico Medicine collaboration — together paint a complex picture of healthcare AI's trajectory. Changes in the regulatory environment will profoundly affect how AI tools are deployed in healthcare systems, while large-scale commercial collaborations continue to validate AI's enormous potential in medicine. For healthcare institutions, regulators, and AI developers across Asia-Pacific, closely monitoring these developments and formulating appropriate strategies will be a critical priority in the years ahead.

Broader Context: The State of AI in Healthcare in 2026

Clinical Evidence Accumulating

Beyond the regulatory and investment developments, the clinical evidence base for AI in healthcare continues to grow. A retrospective study published in the Japanese Journal of Radiology in November 2026 found that AI models could serve as a non-inferior "first reader" for lung cancer screening in chest radiographs compared to human clinicians, though with a higher rate of false-positive detections. This finding illustrates both the promise and the limitations of current AI diagnostic tools.

Similarly, a multi-center study in NPJ Cardiovascular Health demonstrated that an AI-enhanced stethoscope could outperform general practitioners in screening for clinically significant valvular heart disease. These results suggest that AI is approaching clinical utility in specific, well-defined diagnostic tasks — even if general-purpose AI diagnosis remains a more distant goal.

The Documentation Burden Problem

One of the most consistently validated applications of AI in healthcare is the reduction of clinical documentation burden. Real-world evidence from major academic medical centers indicates that ambient AI scribes significantly reduce documentation time — saving clinicians between 13 and 16 minutes per encounter — and contribute to improved clinician well-being.

This application is particularly significant because it addresses one of the most pressing problems in modern healthcare: physician burnout driven by administrative overhead. Studies have shown that physicians spend nearly as much time on documentation as they do on direct patient care. AI scribes that can automatically generate clinical notes from recorded conversations represent a high-value, relatively low-risk application that is already achieving widespread adoption.

Patient-Facing AI: Early Evidence

Research published in The Lancet in October 2026 reported that a patient-facing primary care AI system conducted 114 interactions with zero safety-critical errors. While this is a small sample, it represents an important proof-of-concept for direct-to-patient AI applications. The study's authors noted that the AI system was able to appropriately triage patients, provide accurate health information, and recognize when to escalate to human clinicians.

The Trust Gap

Despite technical progress, a significant gap exists between clinical potential and patient comfort. According to a June 2026 survey by the Pew Research Center, while Americans are generally open to AI performing administrative tasks (such as scheduling or translation), a solid majority (60%) are uncomfortable with AI being used for diagnosis and treatment. A separate survey by Tebra found that 91% of patients expect to be asked for permission every time AI is used in their care.

This trust gap has important implications for the HTI-5 rule's removal of transparency requirements. If patients are already skeptical of AI in healthcare, reducing the transparency of AI tools used in their care may further erode trust — potentially slowing adoption even as the regulatory environment becomes more permissive.

Industry Adoption and Financial Drivers

Revenue Cycle Management

Healthcare organizations are moving toward large-scale implementation of AI, primarily to combat rising costs and administrative complexity. UnitedHealth Group and HCA Healthcare project millions in savings through AI-driven automation of revenue cycle management and claims processing. Consequently, 64% of healthcare providers have indicated plans to invest in AI for claims and denial management.

This financial motivation is driving rapid adoption of AI tools in the "back office" of healthcare — areas like billing, coding, and claims processing where the risk to patient safety is lower and the efficiency gains are more easily quantifiable.

The $2.75 Billion Signal

The scale of the Eli Lilly-Insilico Medicine collaboration — $2.75 billion — sends a powerful signal about the pharmaceutical industry's confidence in AI-driven drug discovery. This is not a small pilot program or a research grant; it is a major commercial commitment that puts significant financial resources behind the hypothesis that AI can meaningfully accelerate drug development.

For the Asia-Pacific region, this deal is particularly significant because Insilico Medicine is headquartered in Hong Kong. The company's success in attracting a major global pharmaceutical partner validates Hong Kong's emerging role as a hub for AI biotech innovation, and may encourage further investment in the region's life sciences sector.

Regulatory Landscape Beyond the US

While the ONC HTI-5 rule is a US-specific development, its implications extend globally. Many countries look to US regulatory frameworks as reference points when developing their own AI governance approaches. The HTI-5 rule's deregulatory direction may influence regulators in other jurisdictions to adopt similarly permissive approaches — or, conversely, may prompt them to differentiate themselves by maintaining or strengthening transparency requirements.

In the Asia-Pacific region, regulatory approaches to healthcare AI vary significantly:

• Singapore: The Health Sciences Authority (HSA) has developed a relatively progressive framework for AI medical devices, emphasizing post-market surveillance and real-world performance monitoring
• Japan: The Pharmaceuticals and Medical Devices Agency (PMDA) has been developing guidance for AI/ML-based software as medical devices, with a focus on continuous learning systems
• Australia: The Therapeutic Goods Administration (TGA) has published guidance on AI/ML medical devices that emphasizes transparency and explainability — potentially diverging from the US approach under HTI-5
• China: The National Medical Products Administration (NMPA) has been developing comprehensive AI medical device regulations, with a strong emphasis on clinical validation

Conclusion

The dual developments of November 2026 — the ONC HTI-5 rule and the Eli Lilly-Insilico Medicine collaboration — illustrate the complex, multifaceted nature of AI's integration into healthcare. On one hand, regulatory changes are reducing the transparency requirements that help healthcare organizations evaluate AI tools, potentially increasing risk for patients and providers. On the other hand, massive commercial investments are validating AI's potential to transform drug discovery and clinical care.

For healthcare stakeholders across Asia-Pacific, the key takeaway is that the AI healthcare landscape is evolving rapidly and in multiple directions simultaneously. Staying informed about both regulatory developments and clinical evidence will be essential for making sound decisions about AI adoption and governance in the years ahead.