Literature Review on the Use of Synthetic Data and AI Advances for Patient-Centered, Sustainable HEOR

OBJECTIVES: Recent advances in AI algorithms have increased the availability of real-world data as well as options for synthetic data, an artificial version of real data created algorithmically without compromising privacy. The aim of this study was to summarize how these advances could provide avenues for patient-centered, sustainable HEOR.

METHODS: Literature review of Embase/Medline (January 1, 2022 through June 15, 2024) was conducted with a focus on: i) the anticipated effect of the approaching patent cliff of many blockbuster therapeutics and the introduction of the Inflation Reduction Act on market competition; ii) AI advancements in generating data; iii) AI methodological advancements in manipulating and analyzing data.

RESULTS: The review highlighted the following surfacing themes. With respect to anticipated increased market competition, demands for accelerated clinical development with integrated digital health technology are expected to continue to increase. The use of digital twins technology and dynamic Bayesian borrowing partial information from external data sources will be more prominent in order to conduct well-powered, yet more efficient, clinical trials requiring reduced sample size. AI will continue to enable the conversion of semi-structured or unstructured data into structured data as well as impute missing data fields. This in turn will allow for the pooling and interoperability of multiple disparate sources of data, from clinical trial data to real-world data and vice versa. This will further enable the creation of fit-for-purpose synthetic structured data that could be used, alongside real data, to augment the efficiency and speed of clinical trials as well as economic evaluations when available comparative effectiveness evidence is scarce.

CONCLUSIONS: The expanding digitalization of healthcare data and the evolution of AI algorithms along with the blending of real and synthetic patient data will increase the applicability and impact of HEOR in conducting equitable patient-centered research.