Patient-Oriented Herb Recommendation System Based on Multi-Graph Convolutional Network

The presented herb recommendation system aims to analyze the patients’ symptoms and recommends a set of herbs as prescription treat diseases. In addition symptoms, personal properties induced diagnoses are also essential for treatment making. Specifically, different age groups, treatments different. However, existing studies only use represent patients ignore multidimensional features modeling. Thus, these models insufficiently personalized. Meanwhile, most based on graphs have not distinguished effects node types. To address above limitations, we propose model named Patient-Oriented Multi-Graph Convolutional Network-based Herb Recommendation (PMGCN). prediction contains two effective modules, patient portraits modeling interactions modeling, learn representations enhance interactions. First, depict portrait enrich individualized features. distinguish properties, diagnoses, adopt type-aware attention mechanism, thereby improving accuracy personalized recommendation. Next, build herb-interaction design multigraph convolution networks capture this way, our emphasizes impact diagnosis induction selection. Experimental demonstrate that method outperforms compared methods confirms significance portraits. conclusion, research proposes adds simulate TCM prescriptions