Enabling Doctor-Centric Medical AI with LLMs through Workflow-Aligned Tasks and Benchmarks

The rise of large language models (LLMs) has profoundly influenced health-care by offering medical advice, diagnostic suggestions, and more. However, their deployment directly toward patients poses substantial risks, as limited domain knowledge may result in misleading or erroneous outputs. To address this challenge , we propose repositioning LLMs as clinical assistants that collaborate with experienced physicians rather than interacting with patients directly. We begin with a two-stage inspiration–feedback survey to identify real-world needs in clinical workflows. Guided by this, we construct DoctorFLAN, a large-scale Chi-nese medical dataset comprising 92,000 Q&A instances across 22 clinical tasks and 27 specialties. To evaluate model performance in doctor-facing applications, 1 we introduce DoctorFLAN-test (550 single-turn Q&A items) and DotaBench (74 multi-turn conversations mimicking realistic scenarios). Experimental results with over ten popular LLMs demonstrate that DoctorFLAN notably improves the performance of open-source LLMs in medical contexts, facilitating their alignment with physician workflows and complementing existing patient-oriented models. This work contributes a valuable resource and framework for advancing doctor-centered medical LLM development.