Purpose/Significance User portraits of doctors offering free medical advice in online live streaming are constructed and further used to analyze the behavioral characteristics of various types of doctors, so as to provide references for improving the effect of live streaming and their influence. Method/Process A sample of 1 816 doctors who offer free medical advice on the Haodf.com platform is used for clustering analysis to generate different doctor group portraits by using the two-step clustering algorithm. Then an evaluation system for live streaming influence is constructed to explore the differences in online influence among doctors with different styles of live streaming. Result/Conclusion Three types of doctor group portraits are identified, including knowledge-sharing, self-promotional and sociable. There are significant differences in their online influence and three dimensions of live streaming effects including activeness, reputation, and hotness among different doctor groups. Comparative analysis is further performed to find potential solutions for enhancing their live streaming influence.