Research And Implementation Of Generation Method Of Medical Strategy Based On Topic Model And ILP

The rapid development of internet plus era has played a great role in promoting the inheritance and development of traditional medical industry.internet plus medical model is rapidly changing the way of seeking medical advice..The emergence of a number of mature online doctor-patient communication platform,not only gathered many online users to discuss their experience of treatment and healing,but also produce and collect a large number of medical information.In-depth analysis and mining of these online data,extraction of the hidden valuable medical information or knowledge,is of great significance,both for website service promotion,or the research of data mining methods and application.Therefore,based on the online community text generated by the user's real medical experience,this paper attempts to explore the generation of medical strategy for common chronic diseases.The main content includes the following five aspects:First,based on the research of mature online doctor-patient communication platform and patient community,the paper analyzes the needs of online medical text mining,and designs the overall scheme of medical strategy generation,which includes two modules:entity-based disease theme joint modeling and topic-oriented automatic summarization.Second,we collect and analyze online medical experience data,domain knowledge dictionary and medical knowledge of common chronic diseases.And we build a medical experience database and strategy template.Third,based on the research of topic modeling and named entity recognition method,we proposed and implemented Joint Entity-based Disease Topic Model.The model makes full use of the structured information of strategy template and entity information in medical treatment experience,and effectively deals with the linguistic style diversity and data heterogeneity of network text.Through the generated theme-word distribution,we select medical experience texts for each theme.Experimental results show that the classification perfomance of the model is better than LDA and LF-LDA.Fourth,the automatic summarization technology is studied,and an automatic summarization method of medical experience based on ILP(Integer Linear Programming)optimization is proposed and implemented.This programming method combines the topic distribution relevance and entity saliency combination analysis of sentences.Through the objective function optimization solution,we generate topic-oriented medical strategy summarization.Fifth,based on the above research,on the basis of building the Wordmate Home system,the research results of medical strategy are realized and applied.