OBJECTIVES: The rapid pace of evidence generation and clinical development along with the advent of machine-learning technologies (MLT) has led NICE and other reimbursement bodies toward a “living” approach to health technology assessment (HTA). Traditional development methods for global value dossiers (GVDs), which serve as the basis for HTA submissions, cannot support the Living HTA approach. We aimed to explore the utility of an MLT-assisted curated reference library to support Living GVD through a case study in TRMM.

METHODS: LiveRef is a continuously updated, web-based, library of indication-specific publications reporting epidemiology, disease burden, treatment practices, and comparative effectiveness. MLT-assisted data reviews and extractions are performed in compliance with PRISMA guidelines; EMBASE, MEDLINE, and Cochrane databases, scientific congresses, trial registries, regulatory and HTA websites are searched and data on relevant populations, interventions, geographies, study designs, and results are extracted, loaded to an interactive platform, linked to original publications, and organized by GVD chapter.

RESULTS: An initial GVD for a product in TRMM was created in April 2021, with 188 original references extracted and loaded into LiveRef. With 13 relevant scientific congresses presenting 605 interventional, humanistic, economic, and real-world-evidence studies for MM in 2021, a GVD update in March 2022 required inclusion of 59 new references, 35 of which were conference presentations, primarily contributing to the disease burden, current treatment, and unmet need sections of the GVD. Efficiency realized from the use of the LiveRef library translated into an estimated 63% reduction in time required to complete the initial update (3 weeks vs 8 weeks conventional).

CONCLUSIONS: HEOR professionals face significant challenges regarding how to efficiently assess a higher volume of evidence while employing rigorous review methods, addressing an emerging requirement to support Living HTAs. This case study demonstrates the value of an MLT-aided Living Reference tool to support rapid GVD updates.