PP21 Are We Ready For It? Developing Criteria To Include Artificial Intelligence Medical Devices (AI-MDs) For Health Technology Assessment

Abstract

Introduction

The increasing pace at which artificial intelligence medical devices (AI-MDs) or digital health technologies (DHTs) have been introduced and integrated in healthcare has not been matched with appropriate selection criteria for health technology assessment (HTA) to inform funding decision-making. To align with international best practice and local regulatory guidance, the Agency for Care Effectiveness (ACE) developed criteria to include AI-MDs as part of its 2022 topic prioritization process for medical technologies. This abstract describes ACE’s approach to develop the inclusion criteria.

Methods

To develop key principles for including AI-MDs in ACE’s topic prioritization process, relevant information from overseas HTA agencies, local regulatory guidelines, and ACE’s existing topic selection criteria were reviewed. A search of international HTA agency websites was conducted in September 2022 to identify relevant information on inclusion of AI-MDs in healthcare for reimbursement recommendations.

Additionally, local regulatory guidelines for AI-MDs in healthcare were also identified. The inclusion criteria were then piloted with AI-MDs identified from ACE’s horizon scanning workstream to examine their feasibility for HTA topic selection.

Results

One overseas framework on DHTs from the National Institute for Health and Care Excellence (NICE) and two local regulatory guidelines were identified. Based on the key finding that the purpose of AI-MD use in guiding clinical management and its associated risks were important considerations, the following criteria were developed: (i) full registration with the regulatory body;(ii) device characteristics should be interventional, have direct impact on patient safety, or support accurate diagnosis or treatment which is critical to avoid death and serious health deterioration; and (iii) the AI algorithm should be fixed as opposed to adaptable as per regulatory requirements. Using this inclusion criteria, eight AI-MDs surfaced from horizon scanning were screened with the above criteria and deemed suitable for HTA topic selection.

Conclusions

As AI technologies are increasingly used to replace or supplement current clinical practice, continuous adaptation of HTA method is needed to ensure appropriate topic selection.