Abstract: 【Purpose/significance】To study the influence of doctor text information and picture information on patients' medical choice behavior in on line medical community doctors’homepage, this paper hopes to provide more in-depth theoretical basis for multi-source information fusion for research in this field, as well as reference opinions to improve online consultation’quality of ser⁃vice.【Method/process】Based on the trust source theory, this paper constructs an online patient selection model from the three dimen⁃sions of reputation credibility, ability credibility and face credibility that affect trust, and collects the text, pictures and other relevant information on the Chunyu doctor’s website. Using deep learning technology to identify facial features in doctor photos, and verifying theoretical models through regression analysis based on multi-source information.【Result/conclusion】The credibility of a doctor's rep⁃utation has a significant positive impact on patients' behavior in choosing a doctor, while the credibility of a doctor's ability negatively
affects patients' behavior in choosing a doctor. In terms of photo information, a serious and unremarkable appearance of doctors can promote patients' trust in doctors, which in turn positively affects patients' behavior in choosing doctors. The facial attractiveness of doctor photos can weaken the negative impact of hospital level.【Innovation/limitation】Pictures information is added to the study of pa⁃tients' decision-making behavior, and the influence of patient generated information and system generated information on patients' be⁃
havior in choosing doctors is fully considered.