The invention provides a patient consultation intelligent reply method and
system based on an RAG
knowledge base, and is applied to the technical field of
data processing. The method comprises the following steps: preprocessing multi-
modal data of a patient to generate text semantic features, voice features, intestinal
barrier function features, intestinal immune function features, intestinal
nerve function features and image features; performing
feature extraction processing on the historical dialogue data of the patient to generate past consultation features and medication feedback features;
processing the previous consultation features and the medication feedback features to generate patient portrait information; preprocessing the
medical knowledge base data, extracting preset semantic features and storing the preset semantic features to a mixed index structure; the intestinal
barrier function features, the intestinal immune function features and the intestinal
nerve function features are processed, and intestinal influence factors are generated; and
processing the text semantic features, the image features, the patient portrait information and the intestinal tract influence factors based on the dynamically updated
medical knowledge base data and an RAG technology to generate target reply information.