SubmissionNumber#=%=#28
FinalPaperTitle#=%=#LAMAR at MedExACT 2026: Agreement-Driven Large Language Model Ensembles for Clinical Decision Extraction from Discharge Summaries
ShortPaperTitle#=%=#
NumberOfPages#=%=#
CopyrightSigned#=%=#
JobTitle#==#
Organization#==#
Abstract#==#This paper presents an ensemble of Qwen3.5-4B language models for extracting medical decisions from discharge summaries in the MedDec dataset. The models were trained to annotate discharge summaries with inline XML-like tags. Three different training strategies were used including dynamic fine-tuning, reinforcement learning, and pseudo-label augmentation. By combining predictions based on inter-model agreement, the system improved performance across evaluation metrics, achieving an overall F1 of 0.5942 and ranking second on the test leaderboard. The results also showed stable performance across demographic groups, suggesting fairness for underrepresented populations.
Author{1}{Firstname}#=%=#Monrada
Author{1}{Lastname}#=%=#Chiewhawan
Author{1}{Username}#=%=#monradach
Author{1}{Orcid}#=%=#
Author{1}{Email}#=%=#monradach@gmail.com
Author{1}{Affiliation}#=%=#Mahidol University
Author{2}{Firstname}#=%=#Keetawan
Author{2}{Lastname}#=%=#Limaroon
Author{2}{Orcid}#=%=#
Author{2}{Email}#=%=#keetawan.lima@kmutt.ac.th
Author{2}{Affiliation}#=%=#King Mongkut's University of Technology Thonburi
Author{3}{Firstname}#=%=#Titipat
Author{3}{Lastname}#=%=#Achakulvisut
Author{3}{Orcid}#=%=#
Author{3}{Email}#=%=#titipat.ach@mahidol.ac.th
Author{3}{Affiliation}#=%=#Mahidol University

==========
èéáğö