SYSTEMATIC EVALUATION OF CLINICAL DATA MODEL COVERAGE AND ALIGNMENT WITH ESMO PRECISION ONCOLOGY MOLECULAR TUMOUR BOARD RECOMMENDATIONS
Author(s)
Abhishek Nayak, Msc1, Marina Ainciburu, PhD1, Noor Khehrah, PhD1, Aedín Culhane, PhD2.
1School of Medicine, University of Limerick, Limerick, Ireland, 2Limerick Digital Cancer Research Centre, School of Medicine, University of Limerick, Limerick, Ireland.
1School of Medicine, University of Limerick, Limerick, Ireland, 2Limerick Digital Cancer Research Centre, School of Medicine, University of Limerick, Limerick, Ireland.
OBJECTIVES: Molecular Tumour Boards (MTBs) are central to precision oncology, but the readiness of existing clinical data models (CDMs) to capture MTB-relevant data has not been systematically evaluated. We compared seven precision oncology CDMs against the reference dataset of 52 elements derived from ESMO MTB recommendations, to assess their fitness for European data exchange and real-world evidence generation.
METHODS: Seven CDMs were evaluated: Precision-DM, WayFindR, DIGICORE MEDOC, HL7 European Cancer Common Data Model (ECCDM), OSIRIS, HL7 minimal Common Oncology Data Elements (mCODE), and Genomic Data Infrastructure (GDI) Harmonised Minimal Data Model (HMDM). Each ESMO data element was mapped against each CDM's published specification using a three-level taxonomy (Complete, Partial, Missing) across nine thematic sections and three ESMO priority tiers (Minimum n=29; Recommended n=15; Optimal n=8), producing a field-level coverage matrix.
RESULTS: Substantial heterogeneity was observed: complete-match rates ranged from 5.8% (DIGICORE) to 42.3% (Precision-DM); any-match from 34.6% (HL7-ECCDM) to 82.7% (Precision-DM). Core clinical and demographic data were broadly represented. However, variability was substantial in genomics, deliberation outputs, and longitudinal follow-up - the domains most specific to MTB function. The MTB Discussion section (n=11) was the most discriminating: Precision-DM was the only model with meaningful coverage (27.3% complete; 90.9% any-match), while HL7-ECCDM achieved 0% any-match. Coverage declined from Minimum to Optimal ESMO tiers, indicating reduced interoperability for advanced use cases.
CONCLUSIONS: No single CDM supports both deliberation outputs and longitudinal post-MTB monitoring; elements (concomitant medications, quality of life) were absent from every CDM. We propose a federated three-layer architecture (Precision-DM, WayFindR, OSIRIS) on the GDI HMDM backbone, extended with four European-specific additions: ESCAT actionability scoring, structured adverse event capture, quality of life assessment, and European Health Data Space FHIR alignment. This provides an immediately actionable foundation for European MTB data infrastructure supporting Real World Evidence generation and Health Technology Assessment.
METHODS: Seven CDMs were evaluated: Precision-DM, WayFindR, DIGICORE MEDOC, HL7 European Cancer Common Data Model (ECCDM), OSIRIS, HL7 minimal Common Oncology Data Elements (mCODE), and Genomic Data Infrastructure (GDI) Harmonised Minimal Data Model (HMDM). Each ESMO data element was mapped against each CDM's published specification using a three-level taxonomy (Complete, Partial, Missing) across nine thematic sections and three ESMO priority tiers (Minimum n=29; Recommended n=15; Optimal n=8), producing a field-level coverage matrix.
RESULTS: Substantial heterogeneity was observed: complete-match rates ranged from 5.8% (DIGICORE) to 42.3% (Precision-DM); any-match from 34.6% (HL7-ECCDM) to 82.7% (Precision-DM). Core clinical and demographic data were broadly represented. However, variability was substantial in genomics, deliberation outputs, and longitudinal follow-up - the domains most specific to MTB function. The MTB Discussion section (n=11) was the most discriminating: Precision-DM was the only model with meaningful coverage (27.3% complete; 90.9% any-match), while HL7-ECCDM achieved 0% any-match. Coverage declined from Minimum to Optimal ESMO tiers, indicating reduced interoperability for advanced use cases.
CONCLUSIONS: No single CDM supports both deliberation outputs and longitudinal post-MTB monitoring; elements (concomitant medications, quality of life) were absent from every CDM. We propose a federated three-layer architecture (Precision-DM, WayFindR, OSIRIS) on the GDI HMDM backbone, extended with four European-specific additions: ESCAT actionability scoring, structured adverse event capture, quality of life assessment, and European Health Data Space FHIR alignment. This provides an immediately actionable foundation for European MTB data infrastructure supporting Real World Evidence generation and Health Technology Assessment.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
RWD15
Topic
Health Technology Assessment, Medical Technologies, Real World Data & Information Systems
Topic Subcategory
Distributed Data & Research Networks
Disease
Genetic, Regenerative & Curative Therapies, Oncology