Myths and Promises of Artificial Intelligence in RWE, HEOR, and Clinical Development

Speaker(s)

ABSTRACT WITHDRAWN

OBJECTIVES: The development and use of artificial intelligence (AI) is gaining traction, including in real world evidence generation (RWE), health economics and outcomes research (HEOR), and broader clinical development. Acceptance is mixed, ranging from early and enthusiastic adopters to traditionalists and even doomsayers, perhaps with most user perspectives falling somewhere between these two extremes. Given the rise and utility of AI, researchers and industry stakeholders can benefit from accurate information about AI and its use in our field.

We present common myths and promises, on how AI can contribute to the clinical landscape. This study examines AI transparency, reliability, regulatory acceptance, and usefulness in RWE, HEOR, and clinical development.

METHODS: The synthesis of myths and promises, regulatory acceptance, and the history of development and use of AI are based on targeted literature reviews, and complemented by unstructured interviews with 10 AI users. Additionally, a simple Natural Language Processing algorithm was used on clinicaltrials.gov to identify increasing applications of (specific) application in clinical trials/development.

RESULTS: In general, transparency in AI (explainable AI/XAI) is considered best practice in algorithm/model development, and the mathematics used in AI are robust and have been established over many years. AI methods have been tested and verified in highly-regulated spaces. For example, the US Food and Drug Administration and European Medicines Agency accept AI in regulatory submissions to varying degrees. Adoption is mainly in automation rather than wholistically used to improve research methods contributing to RWE and HEOR (e.g. causal and predictive modelling, and determining variational treatment effect), and clinical development (e.g. target trial emulation).

CONCLUSIONS: AI can have a pivotal role in our field, including advancing understanding of disease, treatment effect, and precision medicine. Accurately understanding what AI is and how it can be used in RWE, HEOR, and clinical development can contribute to reducing hesitancy and unease about AI.

Code

MSR72

Topic

Methodological & Statistical Research

Topic Subcategory

Artificial Intelligence, Machine Learning, Predictive Analytics

Disease

No Additional Disease & Conditions/Specialized Treatment Areas