Empowering AI assisted clinical drug development: tactics to address data bias, the digital divide and missing patient populations through AI and digital solutions
Abstract
The use of AI methodologies in drug development holds great potential to expedite clinical trials by helping to identify patients most likely to respond favorably to a given medication. To fully harness the benefits of such algorithmic approaches to precision medicine however requires that the datasets from which such predictive analyses are performed are fully representative of the populations intended to benefit. Due to a multitude of factors, there remains extant need to increase the heterogeneity of these data including the contribution of under-represented and other absent populations. There is also a parallel need to broaden patient representation in the clinical trials themselves which are used to demonstrate efficacy of predictive models deployed. Herein we outline tactics, approaches and measures that could be deployed to drive these elements, the potential impact of such on precision medicine, and the ethical, legal, and privacy issues which need to be considered. The benefits of such would be myriad, including helping realize more fully the potential of precision medicine, which aims to pair the most appropriate care and medications with patients based on their individual phenotypic and pharmacogenomic profiles.