BUILDING TRUST IN REAL-WORLD EVIDENCE: VALIDATION OF DATA, METHODS AND AI SOLUTIONS IN EUROPEAN CANCER CENTRES
Author(s)
Luisa Lopes Conceicao, PhD1, Andrea Roncadori, BSc, MSc2, Teresa Mota Garcia, MD1, Valentina Danesi, PhD3, Maria José Bento, MD1, Ilaria Massa, MSc3.
1Instituto Português de Oncologia do Porto FG, EPE (IPO-Porto), Porto, Portugal, 2Outcome Researcher, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori IRST IRCCS, Meldola, Italy, 3Istituto Romagnolo per lo Studio dei Tumori "Dino Amadori" - IRST IRCCS, Meldola, Italy.
1Instituto Português de Oncologia do Porto FG, EPE (IPO-Porto), Porto, Portugal, 2Outcome Researcher, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori IRST IRCCS, Meldola, Italy, 3Istituto Romagnolo per lo Studio dei Tumori "Dino Amadori" - IRST IRCCS, Meldola, Italy.
OBJECTIVES: The increasing use of real-world data (RWD) for health technology assessment (HTA) and regulatory decision-making requires validated approaches for data collection, analysis, and integration of artificial intelligence (AI) tools. The ONCOVALUE project aims to unlock the value of oncology RWD across Europe through standardized data collection, AI-enabled data extraction, and an innovative HTA framework. This abstract presents the validation framework developed to assess the implementation and usability of these solutions in participating European cancer centres.
METHODS: A multi-level validation framework was designed to evaluate three ONCOVALUE solutions: (1) guidelines and standard operating procedures for structured RWD collection, (2) a hybrid RWD-based HTA framework integrating structured and unstructured data, and (3) AI tools for automated extraction of outcomes from clinical notes and medical imaging. Validation levels were tailored to each solution and included assessments of data availability, data retrieval processes, analytical reproducibility, and the ability to address clinical or HTA research questions. For AI tools, additional levels evaluated legal, technical, and operational implementation requirements, usability, and predictive performance. Validation activities will be conducted through pilot studies in breast and lung cancer across multiple European centres.
RESULTS: The framework establishes a harmonized approach for evaluating technical, methodological, and operational aspects of RWD generation and use. Key validation outcomes include availability of variables required for HTA studies, degree of automation in data extraction, reproducibility of analytical workflows, and capacity to replicate clinical trial findings using RWD. AI validation metrics include sensitivity, positive predictive value and F1 score. Pilot studies will assess the framework's applicability in real-world oncology settings.
CONCLUSIONS: The ONCOVALUE validation framework provides a structured methodology for assessing the readiness, quality, and usability of RWD, HTA models, and AI tools across European cancer centres. By supporting standardized validation processes, it may facilitate wider adoption of RWD for HTA and regulatory decision-making in oncology.
METHODS: A multi-level validation framework was designed to evaluate three ONCOVALUE solutions: (1) guidelines and standard operating procedures for structured RWD collection, (2) a hybrid RWD-based HTA framework integrating structured and unstructured data, and (3) AI tools for automated extraction of outcomes from clinical notes and medical imaging. Validation levels were tailored to each solution and included assessments of data availability, data retrieval processes, analytical reproducibility, and the ability to address clinical or HTA research questions. For AI tools, additional levels evaluated legal, technical, and operational implementation requirements, usability, and predictive performance. Validation activities will be conducted through pilot studies in breast and lung cancer across multiple European centres.
RESULTS: The framework establishes a harmonized approach for evaluating technical, methodological, and operational aspects of RWD generation and use. Key validation outcomes include availability of variables required for HTA studies, degree of automation in data extraction, reproducibility of analytical workflows, and capacity to replicate clinical trial findings using RWD. AI validation metrics include sensitivity, positive predictive value and F1 score. Pilot studies will assess the framework's applicability in real-world oncology settings.
CONCLUSIONS: The ONCOVALUE validation framework provides a structured methodology for assessing the readiness, quality, and usability of RWD, HTA models, and AI tools across European cancer centres. By supporting standardized validation processes, it may facilitate wider adoption of RWD for HTA and regulatory decision-making in oncology.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
HTA15
Topic
Health Technology Assessment, Real World Data & Information Systems, Study Approaches
Disease
Oncology