Research On The Online Diagnosis Text Based On Data Mining

In the recent two years,thanks to the online Ask&Diagnosis web,internet has built the link between doctors and patients.Now patients can solve their problems at home by communicating with doctors on the internet.This has produced lots of text data which affords to study patient’s need and doctor’s advice.This paper will choose patients of fatty liver as the research object.With the help of data mining technology,we use text data about fatty liver patients from online Ask&Diagnosis web《120ask》to study the following questions:what are patients like;what the patients care about;how do doctors advise the patients;what the doctors’ preference when choose medicine to treat patients;how do doctors think about Yishanfu,the medicine which can treat fatty liver.Firstly,we select the useful elements of text data from the web《120ask》and crawl the data through a tool web crawler written by Python.Secondly,we transform the unstructured data to structured data.We do this in three steps in sequence:coding every text data to several themes basis on patients’ buying process,building dictionaries for every theme by natural language processing method,tagging every text data with theme dictionaries.Then we preprocess the structured data in several steps,this will help us to prepare clean and available data for building model.At the last,we use three data mining models to study patients and doctors.At first,we use k-means clustering algorithm to divide patients into five groups.every group stands for a type of patients.Then,we use decision tree model to picture doctors’ decision rules and rank the factors that doctors care about when make decision.Lastly,we analyze the medicine advice with association mining method.We also especially study Yishanfu from its’ usage frequency and brand awareness.