2026-10-13T11:04:31Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33709018 2026-09-13T20:06:44Z category_24892 item_type_3 month_year_09_2026

<b>End-to-End Reliability of Automated Systems for Diagnostic Evidence Extraction: A</b><b> </b><b>Protocol-Driven</b><b>Benchmark Study</b> Anton Kravchuk (24861190) Julio Ruben Rodas Juli (24861541) Gregor Donabauer (21071847) Samy Ateia (24862622) Udo Kruschwitz (21071850) Valentin Egert (24862635) Christoph Eckl (24862657) Emily Rinderknecht (24862521) Stefanie Herrmann (11089551) Stephan Siepmann (138085) Maximilian Burger (31656) Christian Gilfrich (24862540) Maximilian Haas (6193823) Matthias May (707496) Other biomedical and clinical sciences not elsewhere classified artificial intelligence (AI) diagnostics Benchmarking Care Diagnostic test accuracy studies data extraction formR 2. large language models, text detection tools, paper mills, research misconduct, publication ethics, verification mechanisms systematic review aggregates urinary biomarkers studied <p dir="ltr">This deposit contains the run-level data, the locked benchmark reference, the analysis code, the human-rater instructions, and the revision-stage verification documentation for a protocol-driven benchmark study of MedNuggetizer and three standalone large language models (ChatGPT-5.2, Claude Opus 4.5, Gemini 3 Pro) in uro-oncologic diagnostic data extraction.</p> 2026-09-13T20:06:44Z Dataset Dataset 10.6084/m9.figshare.33709018.v1 https://figshare.com/articles/dataset/_b_End-to-End_Reliability_of_Automated_Systems_for_Diagnostic_Evidence_Extraction_A_b_b_b_b_Protocol-Driven_b_b_Benchmark_Study_b_/33709018 CC BY 4.0