Abstract:Aiming at the problems existing in traditional consultation processes for acute and critical illnesses, such as department allocation relying on manual experience, slow response speed, and imprecise matching of expert resources, an intelligent expert-case matching method based on Graph Neural Network (GNN) is proposed, and a one-click emergency consultation platform for acute and critical illnesses is constructed. The system utilizes Natural Language Processing (NLP) technology to conduct deep semantic parsing of patient basic information, chief complaints, past medical history, examination and test results, and physical examination data. Combined with deep learning models, disease features are extracted, and a heterogeneous medical knowledge graph composed of case nodes, expert nodes, and disease knowledge nodes is constructed. Through graph neural network learning of the association relationships between cases and experts, intelligent recommendation of consultation departments and precise matching of experts are achieved. The research results show that this method can significantly improve the accuracy of consultation triage and expert matching efficiency, shorten the waiting time of acute and critical patients, and provide technical support for smart healthcare and multidisciplinary collaborative diagnosis and treatment in hospitals.