OBJECTIVES: Data from healthcare and life sciences are typically high dimensional; in other words, there is a high ratio of potential variables to data points. Artificial intelligence (AI) and machine learning (ML) methods are particularly adept at handling such datasets without overfitting, and recent advances in causal ML and explainable AI enable these methodologies to address some of the challenges with Health Technology Assessment (HTA) and the preparation, generation and planning of payer evidence. The objective of this research is to provide an overview of the use cases for AI/ML to support HTA.

METHODS: The IQVIA HTA Accelerator database was used to determine whether AI/ML (or related terms) had been mentioned in global English language HTA documents for single or multiple drug assessments between January 2016 and June 2020. Additionally, relevant publications from the last five years were reviewed and experts were consulted to determine the potential use cases for AI/ML to support HTA submissions.

RESULTS: Two HTAs were found to have made use of AI/ML: one to identify predictors of glycaemic control in diabetes, and the other an Evidence Review Group request to use LASSO for variable selection in a Cox Proportional Hazards model for a Multiple Myeloma treatment.

Potential and future applications of AI/ML to support HTA fell broadly into three categories: generating evidence and understanding disease (including efficacy/effectiveness, safety, disease burden and natural history), informing HTA strategy and planning (including identifying the right target population and understanding predictors of HTA success), and supporting operational aspects of HTA submission preparation (increasing automation and efficiency).

CONCLUSIONS: AI/ML is not currently commonly used to support HTA; however, there are a number of potential applications that may help manufacturers to strengthen their evidence and support their submissions. Some of these applications of AI/ML may become essential for successful submissions in the future.