FROM PREDICTION TO POLICY: HEALTH TECHNOLOGY ASSESSMENT OF CLINICAL PREDICTION MODELS
Author(s)
Lavinbia Ferrante, PhD1, Tuba Saygin Avsar, PhD2, Alina Solomon, MD PhD3, Miia Kivipelto, MD, PhD4, Francesca Mangialasche, PhD4, Niranjan Bose, PhD5, Bethany Shinkins, PhD2.
1Scientific Advisor, National Institute for Health and Care Excellence (NICE), Manchester, United Kingdom, 2National Institute for Health and Care Excellence, Manchester, United Kingdom, 3UEF - Nordic-UK Brain Network (UEF-NBN), University of Eastern Finland, Kuopio, Finland, 4Karolinska Institutet, Stockholm, Sweden, 5Gates Ventures, LLC, Kirkland, WA, USA.
1Scientific Advisor, National Institute for Health and Care Excellence (NICE), Manchester, United Kingdom, 2National Institute for Health and Care Excellence, Manchester, United Kingdom, 3UEF - Nordic-UK Brain Network (UEF-NBN), University of Eastern Finland, Kuopio, Finland, 4Karolinska Institutet, Stockholm, Sweden, 5Gates Ventures, LLC, Kirkland, WA, USA.
OBJECTIVES: Clinical prediction models (CPMs) combine multiple sources of patient data to inform diagnosis, prognosis and treatment decisions. As their use expands, clear HTA processes are needed to identify models that improve outcomes, prevent ineffective or harmful implementation, and clarify the evidence requirements for reimbursement. This study provides an overview of the current HTA landscape for assessing CPMs across Europe.
METHODS: We reviewed HTA methods manuals from agencies across the European Economic Area, identified via EUnetHTA, ISPOR, and a previous EU mapping study. Documents were screened for content relevant to CPMs, and guidance was analysed using the EUnetHTA Core Model framework. We supplemented review findings with the perspectives of HTA experts (14 organisations across 10 European countries), hosting workshops to discuss their experience and challenges of evaluating CPMs, and to identify existing adoption pathways for novel CPMs.
RESULTS: We examined 29 methods guidance documents. EunetHTA, HAS, IQWIG, NICE, SBU, and ZiN offered the most comprehensive and relevant guidance on the assessment of diagnostics more generally, with only 3 (IQWIG, NICE and SBU) providing guidance specific to CPMs. Most agencies evaluated CPMs within broader diagnostic frameworks, lacking guidance on CPM-specific issues including calibration, time-to-event modelling, and model transportability. Workshop discussions supported these findings: HTA agency representatives shared limited experience of assessing CPMs and no clear adoption pathways. Decision-critical issues with insufficient guidance were: validation and applicability assessment, context of use, link to patient outcomes, and handling uncertainty.
CONCLUSIONS: The rapid increase in available CPMs creates a timely opportunity to strengthen methodological guidance for their assessment within existing European HTA frameworks. Incorporating CPM-specific issues and creating clearer adoption pathways could expedite their adoption in practice and facilitate equitable implementation. A follow-on study to develop methodological guidance for HTA of CPMs is being undertaken as part of the IHI-funded AD-RIDDLE project.
METHODS: We reviewed HTA methods manuals from agencies across the European Economic Area, identified via EUnetHTA, ISPOR, and a previous EU mapping study. Documents were screened for content relevant to CPMs, and guidance was analysed using the EUnetHTA Core Model framework. We supplemented review findings with the perspectives of HTA experts (14 organisations across 10 European countries), hosting workshops to discuss their experience and challenges of evaluating CPMs, and to identify existing adoption pathways for novel CPMs.
RESULTS: We examined 29 methods guidance documents. EunetHTA, HAS, IQWIG, NICE, SBU, and ZiN offered the most comprehensive and relevant guidance on the assessment of diagnostics more generally, with only 3 (IQWIG, NICE and SBU) providing guidance specific to CPMs. Most agencies evaluated CPMs within broader diagnostic frameworks, lacking guidance on CPM-specific issues including calibration, time-to-event modelling, and model transportability. Workshop discussions supported these findings: HTA agency representatives shared limited experience of assessing CPMs and no clear adoption pathways. Decision-critical issues with insufficient guidance were: validation and applicability assessment, context of use, link to patient outcomes, and handling uncertainty.
CONCLUSIONS: The rapid increase in available CPMs creates a timely opportunity to strengthen methodological guidance for their assessment within existing European HTA frameworks. Incorporating CPM-specific issues and creating clearer adoption pathways could expedite their adoption in practice and facilitate equitable implementation. A follow-on study to develop methodological guidance for HTA of CPMs is being undertaken as part of the IHI-funded AD-RIDDLE project.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR137
Topic
Health Technology Assessment, Medical Technologies, Methodological & Statistical Research
Disease
Neurological Disorders, Personalized & Precision Medicine