Do the Demands of European Health Technology Assessment (HTA) Regulation Necessitate the Adoption of Artificial Intelligence (AI) in the Joint Clinical Assessment (JCA) Development?

Speaker(s)

Povsic M, Fisher F
AMICULUM, Bollington, LAN, UK

OBJECTIVES: With the first stages of JCA rollout beginning in January 2025, the scale of work for countries and manufacturers is expected to be significant. We examined whether organic expansion could meet the challenges presented or if only a technical leap such as AI could meet expected needs.

METHODS: Via directed multidisciplinary workshopping involving therapy area, market access, regulatory, and AI specialists, we developed the most likely JCA process. We analyzed typical timeframes and estimated resourcing needs. We mapped JCA objectives (efficiency, speed, harmonization, evidence-based decision-making, patient-centric approach) and measured AI performance against them. We identified candidate areas where AI efficiencies could be expected quickly, and estimated the range of efficiencies that could be delivered.

RESULTS: Our findings suggest a significant resource gap between what is currently available and what is expected for JCA. The short period required for JCA creation in particular, proved near impossible with limited resources and without automation/AI.

The areas where AI could provide the largest (~20%) time savings were input/output templates and literature screening. Due to the need for explainability and consistent outputs, AI tools that had defined datasets and were guided by ‘desirable’ templates, ie Retrieval Augmented Generation models and similar, were found to be most beneficial.

CONCLUSIONS: We report on a first comprehensive review of AI-supported JCA feasibility. Some level of AI support will be required, primarily to address resource gaps and timeline constraints. The largest efficiencies were found across literature screening and document templates. However, hurdles for successful implementation of AI assistance for JCA creation remain and will be best mitigated by a close partnership between all stakeholders. Exponential growth of AI in this area is likely to prompt a rapid response from policymakers and the eventual development of a consistent framework for assessing AI processes.

Code

MSR211

Topic

Methodological & Statistical Research

Topic Subcategory

Artificial Intelligence, Machine Learning, Predictive Analytics

Disease

Drugs, No Additional Disease & Conditions/Specialized Treatment Areas, Oncology