NICE’s Position on AI in Evidence Generation

The National Institute for Health and Care Excellence (NICE) has issued a position statement on the use of artificial intelligence (AI) in evidence generation. This statement clarifies what NICE expects when incorporating AI methods for evidence generation and reporting. The guidance emphasizes the necessity for transparency, rigor, and trust throughout the evidence generation process.

Key Objectives and Expectations

The position statement outlines NICE’s expectations regarding AI methods in evidence generation. It highlights the importance of adhering to existing regulations, good practices, and guidelines when utilizing AI methods.

Benefits of AI in Evidence Generation

AI methods, such as machine learning and generative AI, can efficiently handle large datasets, reveal patterns, and produce novel outputs beneficial for evidence synthesis. This capability can significantly enhance clinical evidence, trial design, systematic reviews, real-world data analysis, and cost-effectiveness studies.

Transparency and Justification

When AI is employed, it is essential for reporting to be transparent. NICE advises using established checklists like PALISADE to justify AI application and TRIPOD+AI to outline AI model development. Any AI approach must be treated as part of the clinical trial, with comprehensive details provided in the submission.

Human Involvement and Oversight

NICE emphasizes that AI should enhance human involvement, not replace it. Human oversight is vital to validate AI-generated outputs and to mitigate risks like algorithmic bias.

Specific Applications of AI

AI can streamline conventional literature search and review processes, covering search strategies, study classification, primary and full-text screening, and visualizing search results. Also, large language models (LLMs) can automate data extraction and generate code for synthesizing gathered data.

In real-world data analysis, AI supports various evidence generation stages, including updating economic models with newly acquired information.

Engagement with NICE

Organizations looking to leverage AI methods should engage with NICE to review their plans. Initial engagement can occur through NICE Advice, with further discussions involving the appropriate NICE technical teams taking place in later stages.

Future Considerations

NICE recognizes that other institutions, like Cochrane and the Guidelines International Network, are also developing guidelines for responsible AI use in evidence synthesis. This trend indicates greater integration of AI in health technology assessments (HTA).

In conclusion, NICE’s position on AI in evidence generation emphasizes its potential benefits while stressing the importance of transparency, human oversight, and strict adherence to existing standards and regulations to ensure the effectiveness and trustworthiness of AI methods.