NICE opens the door to Generative AI in health economic evaluations

NICE has recently completed a series of stakeholder workshops exploring the potential applications, opportunities, and challenges of Generative AI (GenAI) in health economic evaluation (HEE). These discussions build on NICE’s 2024 Position Statement on the Use of AI in Evidence Generation, which acknowledges that AI methods are likely to play an increasing role in future HTA submissions — provided they are used transparently, responsibly, and give clear added value.

The overarching NICE guidance is clear:

• GenAI should only be used where there is demonstrable value from doing so.
• The submitting organisation remains fully accountable for its evidence and conclusions.
• All uses of AI must comply with relevant legislation (data protection, copyright, licensing).
• Tools that enhance explainability and transparency are encouraged.
• AI should augment, not replace, human expertise.

‍

What can pharmaceutical companies and other organisations learn from this?

Health economic evaluation is one of the most resource-intensive elements of HTA submissions — and therefore one of the areas where GenAI could bring the greatest benefit. The GenAI technology offers clear opportunities to improve efficiency and quality by automating repetitive tasks, accelerating evidence identification and supporting the conceptualisation and implementation of economic models.

GenAI can rapidly extract input parameters from the literature, generate or adapt code for model construction, and even draft technical documentation. The potential time savings are substantial, but so are the risks — particularly around the consistency and robustness of AI-generated outputs.

NICE’s position signals both openness and caution. The agency recognises the potential of AI but expects full human oversight, justification of use, and transparency. This marks a step in the right direction: a green light to explore AI in modelling, balanced with clear expectations for accountability and rigour.

Pharmaceutical companies should therefore start exploring AI-supported evidence generation, piloting tools that improve model development, validation, and reporting — always with human validation at the core.

‍

‍

How can Generative AI support model development and validation?

From a technical perspective, GenAI can add value at nearly every stage of health economic model development:

• Model design: Supporting literature reviews to identify suitable model structures and comparators.
• Parameter sourcing: Automating targeted evidence review, extraction and data summarisation.
• Model construction: Assisting in generating R, VBA, or Python code/ Excel formulas, and model templates.
• Validation: Running automated plausibility and consistency checks, comparing AI-replicated versions of models against the original.
• Adaptation and reporting: Updating country-specific inputs and drafting technical reports efficiently.

However, human oversight remains indispensable. AI can generate, suggest, or verify, but cannot judge context, clinical plausibility, or policy relevance. Economists and technical experts must interpret AI outputs, resolve ambiguities, and ensure final models reflect methodological best practice.

NICE’s recent workshops and position statements mark an important first step towards mainstreaming AI in HTA. By embracing AI thoughtfully — as an accelerator rather than a substitute for expertise — the industry can achieve faster, more consistent, and higher-quality evaluations while maintaining the scientific integrity on which HTA decisions depend.

‍

‍

References:

1) NICE Generative AI in health economic evaluation - https://www.nice.org.uk/what-nice-does/our-research-work/hta-lab/hta-lab-projects#generative

2) Use of AI in evidence generation: NICE position statement - https://www.nice.org.uk/position-statements/use-of-ai-in-evidence-generation-nice-position-statement

‍