The National Institute for Health and Care Excellence (NICE) has published its AI Delivery Plan, Unlocking Growth by Enabling Safe AI Innovation, setting out how AI will be embedded across guidance development, health technology assessment (HTA), evidence generation, and the evaluation of AI-enabled technologies. The plan was accompanied by a blog introducing it from NICE CEO Prof. Jonathan Benger. While the plan is framed around organizational transformation, its implications extend far beyond NICE itself, affecting manufacturers, HEOR teams, market access leaders, and evidence generation functions across the life sciences sector.

Why does this matter?

AI is rapidly changing how evidence is generated, analyzed, and used in healthcare decision-making. Recognizing this, NICE positions itself not only as a user of AI, but also as a regulator, evaluator, and methodological leader shaping how AI should be used within HTA. The organization views AI as an opportunity to improve health outcomes, increase system productivity, and support innovation across life sciences and health technologies.

Importantly, this work builds on NICE’s existing AI activity, including its AI statement of intent, position statement on AI-generated evidence, and the publication of 17 evaluations of AI-based medical devices since 2023.

For industry stakeholders, the message is clear: AI is moving from the periphery of HTA into mainstream evidence generation and decision-making.

Three strategic priorities

The delivery plan focuses on three major opportunities:

1. Transforming NICE’s internal operations with AI
2. Enabling AI-driven innovation in HTA methods
3. Evaluating AI-based health technologies more effectively

NICE intends to implement AI through a phased “Scan → Assess → Pilot → Scale → Embed” model, balancing innovation with governance and assurance.

1. Transforming guidance development and knowledge management

Perhaps the most ambitious aspect of the plan is NICE’s effort to transform how guidance is created, maintained, and accessed.

NICE is developing a modern knowledge platform built on structured, machine-readable content, semantic data models, and new digital authoring systems. The objective is to make recommendations easier to find, navigate, and apply while enabling AI-assisted workflows across evidence reviews and guidance development.

The headline ambition is striking: NICE aims to halve the staff time required to produce guidance by 2030.

For HTA and market access professionals, this could eventually translate into:

• Faster evidence assessments
• More dynamic guidance updates
• Greater use of structured evidence submissions
• Increased demand for interoperable and AI-ready data assets

Companies that already organize evidence in structured formats may be better positioned to engage with future NICE processes.

2. AI is becoming a legitimate HTA methodology

The most consequential section for HEOR professionals concerns NICE’s plans for AI-enabled HTA methods.

NICE explicitly acknowledges that AI is already being used, or is expected to be used, across numerous HTA activities including evidence generation, evidence synthesis, and economic modelling. The organization sees significant opportunities to improve efficiency while maintaining scientific rigor.

Several initiatives are particularly noteworthy:

Expansion of AI sandboxes

NICE will continue developing its two experimental environments:

• The HTA Innovation Lab (HTA Lab), a regulatory sandbox
• The AI Research Environment (ARE), a technical sandbox for testing AI use cases

These initiatives provide a valuable signal about the types of AI applications likely to become acceptable within future HTA practice.

Best-practice framework for AI-generated evidence

By March 2027, NICE plans to develop a best-practice methods framework covering AI use in:

• Evidence generation
• Evidence synthesis
• Literature review
• Evidence assessment

This could become one of the most influential methodological developments in HTA over the next few years.

For HEOR teams, key questions will include:

• How should AI-assisted systematic reviews be validated?
• What documentation will be required for AI-supported evidence synthesis?
• How much transparency will be expected around AI-generated analyses?
• What standards will apply to AI-assisted economic models?

The forthcoming framework is likely to shape expectations across the broader HTA ecosystem.

AI in literature reviews

NICE is also exploring AI tools to support scientific literature reviews and evidence assessment, subject to copyright considerations.

This is particularly relevant because literature review activities remain among the most resource-intensive components of HTA submissions and evidence generation programs.

3. Continued focus on evaluating AI technologies

The third pillar focuses on NICE’s assessment of AI-enabled health technologies.

NICE highlights that it has already published guidance on 17 AI-based medical devices and has adopted a lifecycle approach designed to ensure evidence requirements remain proportionate to technology maturity.

Examples include recent recommendations on AI tools supporting fracture detection and ongoing work evaluating AI solutions across multiple clinical areas.

Looking ahead, NICE plans to:

• Continue working closely with the Medicines and Healthcare products Regulatory Agency (MHRA)
• Refine approaches for adaptive and evolving AI systems
• Clarify the scope of AI technologies that NICE will evaluate
• Publish four additional pieces of guidance for AI technologies referred by the Secretary of State

This is particularly important because adaptive AI systems challenge traditional HTA assumptions. As algorithms evolve, questions emerge around evidence validity, performance drift, re-evaluation frequency, and value assessment. NICE’s work in this area could establish influential precedents internationally.

Implications for industry

The delivery plan signals several important trends for manufacturers and evidence teams:

AI-assisted evidence generation is moving toward acceptance

NICE is no longer asking whether AI can be used in HTA. Instead, it is focused on determining how AI should be used responsibly and transparently.

Evidence governance will become increasingly important

As AI-supported evidence generation expands, sponsors should expect growing scrutiny of:

• Reproducibility
• Transparency
• Validation
• Human oversight
• Auditability

Real-world evidence and structured data will gain importance

NICE’s emphasis on data foundations and machine-readable content suggests increasing value will be placed on high-quality, interoperable datasets suitable for AI-enabled analysis.

AI developers face a clearer pathway

The commitment to future-relevant evaluation methods, coupled with initiatives such as the National HealthTech Access Programme, should help reduce uncertainty for developers seeking NHS adoption of AI technologies.

Final thoughts

The NICE AI Delivery Plan represents more than an internal digital transformation program. It is a statement that AI will become an integral part of evidence generation, HTA practice, and health technology evaluation in England.

For HTA, HEOR, and market access professionals, the immediate takeaway is not that submission requirements will change overnight. Rather, NICE is laying the foundations for a future in which AI-assisted evidence generation, AI-enabled literature reviews, and AI-supported decision-making become routine components of the HTA process.

Organizations that invest now in transparent AI methods, strong data governance, and AI-ready evidence infrastructures are likely to be best positioned as NICE’s vision moves from experimentation to implementation.

Share Via