The application discloses a Mongolian
medicine prescription recommendation method based on multi-graph
convolution fusion, and relates to the technical field of
medicine, in particular to a Mongolian
medicine prescription recommendation method based on multi-graph
convolution fusion. The method comprises the following steps: constructing a Mongolian medicine prescription
recommendation model based on multi-graph
convolution fusion
Transformer, designing
feature coding to introduce
drug properties and medication time, combining syndrome information, constructing multiple relationship graphs, using graph convolution to capture features, fusing comprehensive representations of drugs and symptoms, and then performing
drug recommendation. The method is characterized in that the unique
drug 'taste, nature and effect' characteristics, 'nature and taste formula' prescription principles and'medication at different times' medication methods of Mongolian medicine are analyzed, coding and
processing are performed accordingly, a special
loss function is designed for the prescription process to simulate the Mongolian prescription process, the drug recommendation problem is converted into a drug recommendation and combination problem, and special regularization is designed to simulate the prescription process based on in-depth research on the Mongolian prescription principles, so that the whole scheme is changed from simple drug recommendation to drug recommendation and drug combination, and the method is more in line with the Mongolian theoretical
system and clinical practice.