AN INTEGRATED THREE-PLATFORM AI WORKFLOW FOR EU JOINT CLINICAL ASSESSMENT: AI-ASSISTED PICO DEFINITION, REAL-WORLD EVIDENCE VALIDATION, AND AI DOSSIER AUTHORING
Author(s)
Stefan Walzer, MA, PhD1, Shrutya Bhalla, MBA2, Anna Forsythe, MBA, MSc, PharmD3, Jonas Jost, BSc4, Rozee Liu, MSc5, Lutz Michael Vollmer, MSc4.
1MArS Market Access & Pricing Strategy GmbH, Germany, 2MArS GmbH – Moolya, MISSISSAUGA, ON, Canada, 3Oncoscope, Miami, FL, USA, 4MArS Market Access & Pricing Strategy GmbH, Weil am Rhein, Germany, 5Oncoscope AI, Miamia, FL, USA.
1MArS Market Access & Pricing Strategy GmbH, Germany, 2MArS GmbH – Moolya, MISSISSAUGA, ON, Canada, 3Oncoscope, Miami, FL, USA, 4MArS Market Access & Pricing Strategy GmbH, Weil am Rhein, Germany, 5Oncoscope AI, Miamia, FL, USA.
OBJECTIVES: The EU Joint Clinical Assessment (JCA) under Regulation (EU) 2021/2282 requires precise PICO operationalisation, real-world evidence (RWE) integration, and coordinated multi-country dossier preparation under compressed timelines. This proof-of-concept study evaluated the feasibility of an integrated three-platform AI workflow combining an AI-assisted PICO definition platform, an AI-RWE validation platform, and an AI dossier authoring platform operating in parallel.
METHODS: Three AI platforms operated simultaneously across predefined functional interfaces. The PICO definition platform generated and refined PICO frameworks based on EUnetHTA21 methodological guidance, clinical study data, and indication-specific evidence landscapes. The RWE platform interrogated real-world data sources, validating proposed comparators and outcome parameters against payer-relevant treatment patterns and standard-of-care. The AI dossier authoring platform received structured PICO outputs and RWE-validated parameters as upstream inputs and generated JCA dossier module drafts. Output completeness, cross-platform consistency, and workflow efficiency were assessed by market access and HTA experts (n=3; ≥5 years’ experience).
RESULTS: The integrated workflow produced a complete JCA submission package from PICO definition through to structured dossier drafts within a shorter timeframe versus conventional sequential expert-driven preparation. PICO frameworks demonstrated high structural completeness and traceable evidence sourcing. The RWE platform identified discrepancies between clinically proposed comparators and payer-observed real-world treatment sequences, enabling prospective PICO refinement before dossier authoring. The AI dossier authoring platform produced coherent, structured dossier module drafts directly informed by validated PICO inputs, with cross-platform consistency maintained throughout; expert review highlighted targeted areas requiring human refinement in clinical relevance argumentation and country-specific contextualisation, with no fundamental structural inconsistencies observed.
CONCLUSIONS: A parallel three-platform AI workflow demonstrated feasibility for end-to-end JCA submission preparation, from evidence-based PICO operationalisation through real-world validation to structured dossier drafting. The architecture may help address the PICO precision and real-world grounding required under EU JCA timelines. Quantitative benchmarking, quality assessment against EUnetHTA 21 standards, and external validation constitute priority next steps.
METHODS: Three AI platforms operated simultaneously across predefined functional interfaces. The PICO definition platform generated and refined PICO frameworks based on EUnetHTA21 methodological guidance, clinical study data, and indication-specific evidence landscapes. The RWE platform interrogated real-world data sources, validating proposed comparators and outcome parameters against payer-relevant treatment patterns and standard-of-care. The AI dossier authoring platform received structured PICO outputs and RWE-validated parameters as upstream inputs and generated JCA dossier module drafts. Output completeness, cross-platform consistency, and workflow efficiency were assessed by market access and HTA experts (n=3; ≥5 years’ experience).
RESULTS: The integrated workflow produced a complete JCA submission package from PICO definition through to structured dossier drafts within a shorter timeframe versus conventional sequential expert-driven preparation. PICO frameworks demonstrated high structural completeness and traceable evidence sourcing. The RWE platform identified discrepancies between clinically proposed comparators and payer-observed real-world treatment sequences, enabling prospective PICO refinement before dossier authoring. The AI dossier authoring platform produced coherent, structured dossier module drafts directly informed by validated PICO inputs, with cross-platform consistency maintained throughout; expert review highlighted targeted areas requiring human refinement in clinical relevance argumentation and country-specific contextualisation, with no fundamental structural inconsistencies observed.
CONCLUSIONS: A parallel three-platform AI workflow demonstrated feasibility for end-to-end JCA submission preparation, from evidence-based PICO operationalisation through real-world validation to structured dossier drafting. The architecture may help address the PICO precision and real-world grounding required under EU JCA timelines. Quantitative benchmarking, quality assessment against EUnetHTA 21 standards, and external validation constitute priority next steps.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
PT10
Topic
Health Technology Assessment
Topic Subcategory
Decision & Deliberative Processes, Value Frameworks & Dossier Format
Disease
No Additional Disease & Conditions/Specialized Treatment Areas