Generative AI in HTA is rapidly becoming a cornerstone in modern healthcare, offering innovative solutions to enhance HTA processes. By automating systematic literature reviews and analysing real-world evidence, generative AI significantly improves efficiency and accuracy. In this Directors Digest, I highlight the applications, benefits, and challenges of integrating generative AI into HTA, and reflect on the insights published in the ISPOR Working Group Report (in the resources listed below).

Generative AI, particularly large language models like GPT-4, has dramatically improved data processing capabilities. These models, trained on extensive datasets, can perform complex tasks such as language generation and data analysis. In HTA, generative AI aids in automating literature reviews, enhancing real-world evidence analysis, and supporting health economic modelling.

Notes from the Report

• Systematic Literature Reviews: Generative AI can automate the identification of relevant literature, propose search terms, and screen abstracts, reducing the time and effort required for systematic reviews. Studies indicate that models like GPT-4 achieve high accuracy in data extraction and meta-analysis code generation, though human oversight remains crucial.
• Real-World Evidence (RWE): By processing unstructured data from electronic health records, generative AI improves the accuracy and consistency of RWE, vital for informed decision-making in HTA. Techniques such as few-shot learning enable efficient variable extraction, despite challenges related to data privacy and bias.
• Health Economic Modelling: Generative AI supports economic model development by aiding in conceptualisation, parameterisation, and validation, leading to more efficient resource allocation and policy decisions. Expert guidance is essential to mitigate risks of inaccuracies and ensure model validity.

Strategic Insights and Ethical Considerations

The integration of generative AI in HTA promises to enhance healthcare decision-making by improving efficiency and accuracy. However, this advancement must be balanced with ethical considerations.

Governance models, informed by global standards such as those from the UN and WHO, are essential to ensure AI’s ethical deployment. These models should focus on inclusivity, equity, and public interest, ensuring that AI tools do not exacerbate existing health disparities. The lessons from Big Data governance highlight the importance of robust frameworks that address data privacy, security, and algorithmic biases.

As AI continues to evolve, adaptive governance models will be crucial in maintaining the balance between innovation and ethical responsibility. Stakeholders must engage in continuous dialogue to refine these frameworks, ensuring that AI serves as a catalyst for improving patient outcomes and bridging health disparities.

Future Directions and Final Thoughts

Generative AI holds immense promise for transforming HTA, offering opportunities to enhance efficiency, accuracy, and equity in healthcare. While the technology is still in its nascent stages, its potential to reshape evidence generation and policy formulation is undeniable. As the field evolves, continuous evaluation and human oversight will be essential to harness its full potential responsibly. Healthcare professionals and policymakers are encouraged to engage with ongoing research initiatives to stay at the forefront of this transformative wave in healthcare.

Resources