To provide an introduction to the uses of generative artificial intelligence (AI) and foundation models, including large language models, in the field of health technology assessment (HTA).
We reviewed applications of generative AI in 3 areas: systematic literature reviews, real-world evidence, and health economic modeling.
(1) Literature reviews: generative AI has the potential to assist in automating aspects of systematic literature reviews by proposing search terms, screening abstracts, extracting data, and generating code for meta-analyses; (2) real-world evidence: generative AI can facilitate automating processes and analyze large collections of real-world data, including unstructured clinical notes and imaging; (3) health economic modeling: generative AI can aid in the development of health economic models, from conceptualization to validation. Limitations in the use of foundation models and large language models include challenges surrounding their scientific rigor and reliability, the potential for bias, implications for equity, as well as nontrivial concerns regarding adherence to regulatory and ethical standards, particularly in terms of data privacy and security. Additionally, we survey the current policy landscape and provide suggestions for HTA agencies on responsibly integrating generative AI into their workflows, emphasizing the importance of human oversight and the fast-evolving nature of these tools.
Although generative AI technology holds promise with respect to HTA applications, it is still undergoing rapid developments and improvements. Continued careful evaluation of their applications to HTA is required. Both developers and users of research incorporating these tools, should familiarize themselves with their current capabilities and limitations.
This report discusses the use of generative artificial intelligence (AI) in health technology assessment (HTA) and highlights its potential benefits and challenges. Generative AI, especially large language models, could significantly improve how evidence is generated for healthcare decisions. This overview is important because it addresses the growing need for efficient methodologies in assessing health technologies and provides insights into the application of advanced AI tools in this field.
The report identifies 3 main areas where generative AI can enhance HTA: conducting systematic literature reviews, analyzing real-world evidence, and developing health economic models. In literature reviews, generative AI can automate tasks like proposing search terms, screening abstracts, and extracting data. For real-world evidence, it can analyze large datasets, including unstructured clinical notes, improving data processing and accuracy. In health economic modeling, generative AI can assist in creating and validating models.
Despite these opportunities, the report also points out significant challenges. The use of generative AI is still in its early stages and there are concerns about scientific accuracy, potential biases, and ethical implications like data privacy. AI-generated outputs may be unreliable, which necessitates human oversight in its application. The report emphasizes that while generative AI can support human efforts, it should not replace them entirely.
For healthcare decision makers, the findings highlight the need for clear guidelines on how to responsibly integrate generative AI into HTA processes. Policy makers are encouraged to collaborate with HTA agencies and develop standardized practices that ensure transparency in using AI technologies.
Researchers are urged to stay informed about the capabilities and limitations of generative AI tools as they evolve. Ongoing evaluation of their application in HTA is essential to ensure their responsible use and to enhance the quality of their assessments.
In summary, generative AI holds promise for transforming HTA by improving efficiency and accuracy in evidence generation. However, careful consideration of its limitations and the importance of human oversight are crucial as this technology continues to develop.