Presentation (CTI)

OBJECTIVES: Swiss reimbursement submissions to the Federal Office of Public Health (BAG/OFSP) require distinct, structurally complex dossier types tailored to the applicable regulatory pathway. Preparation of these documents is resource-intensive and demands deep familiarity with Swiss-specific submission requirements. This study evaluated the feasibility and efficiency of an AI-assisted dossier authoring platform, DO-BO, for generating three BAG submission document types: new inclusion (,,Neuaufnahme“), change of an existing limitation (,,Gesuch auf Änderung der Limitation“), and the accompanying letter (,,Begleitbrief“).
METHODS: The dossier-creation platform DO-BO was configured, developed, tested, and implemented for three BAG-specific document types across a structured development-and-validation workflow comprising (i) template design and prompt engineering per document type, (ii) controlled testing using two real reimbursement dossiers from distinct therapeutic areas (oncology and immunology/nephrology) as reference cases, and (iii) comparative output assessment. AI-generated outputs were qualitatively compared against conventionally authored documents across four dimensions: turnaround time, personnel resource requirements, budget, and output quality. Assessment was conducted by an experienced market access expert reviewing both AI-generated and traditional outputs using the same structured evaluation framework.
RESULTS: For all three BAG submission document types, structured draft outputs were successfully generated. Across both reference dossiers, AI-assisted authoring was qualitatively assessed to reduce turnaround time, personnel resource requirements, and associated costs relative to conventional preparation. Output quality of AI-generated drafts was judged comparable to conventionally authored documents by the reviewing market access expert, with targeted human refinement required primarily for submission-context-specific argumentation and regulatory nuance. Efficiency gains were consistent across new applications, re-submission pathways, and the accompanying letter.
CONCLUSIONS: For AI-assisted preparation of Swiss BAG reimbursement dossiers across multiple submission document types the evaluated DO-BO demonstrated feasibility and efficiency gains. These findings complement prior validation data from AMNOG HTA dossiers and other dossier types, establishing DO-BO as a platform applicable across heterogeneous HTA submission contexts.