Abstract:This study proposes a medical dialogue intention fusion generation network (MedIFG-Net), which generates outpatient medical records through multi-turn doctor-patient dialogues on online consultation platforms. This system assists in quickly summarizing patient medical history and generating medical documentation. A dialogue intention recognition method based on the ERNIE model is introduced to identify the dialogue intentions of doctors and patients. To integrate dialogue and intentions, an attention-based fusion method is also proposed. By analyzing and recognizing intention information from multi-turn dialogues, and performing the fusion process, the system generates outpatient medical records more efficiently and accurately. Experimental results demonstrate that the proposed method significantly outperforms traditional approaches, exhibiting enhanced generation capabilities and practical application value.