Severin Schramm, B Le Guellec, Marlene Topka, Mortimer Svec, Paul Backhaus, Viktor Maria Eisenkolb, Evamaria O. Riedel, Mirjam Beyrle, Paul Platzek, Constanze Ramschütz, Karolin J. Paprottka, Martin Renz, Jannis Bodden, Jan S. Kirschke, Sebastian Ziegelmayer, Felix Busch, Marcus R. Makowski, Lisa C. Adams, K Bressem, Dennis M. Hedderich, B Wiestler, S Kim
With increasing reader experience, large language model accuracy improved, whereas diagnostic gains from model assistance paradoxically diminished, highlighting the divergence between model performance and clinically relevant benefits for readers.