Text Mining Of Patient Satisfaction In Online Health Community And The Research Of Its Impact On Patient Selection Of Doctors

The development of Online Health Community(OHC)boosts the booming of various online healthcare service and the online-offline interaction.Patients can make appointment online and then go to doctor offline.At the same time,offline patients can also surf the internet to comment on doctors for their service to make reference for other patients who want to choose an appropriate doctor,which accelerates the transparency of doctor's information and alleviates the information asymmetry in healthcare industry.However,tensions between doctors and patients still a tricky problem in China.Patient Satisfaction Questionnaire distributed in hospital cannot reflect patients' true attitude towards doctors,which exists positivity bias.Patients tend to write down positive comments because they rely on hospital and doctors' service to treat the disease.Hospital have no access to patients' personal information if patients comment on doctor via OHC's doctor rating system.Therefore,it is reasonable to believe that doctor rating in OHC is more objective.Having an understanding of patient satisfaction and their doctor selection behavior in OHC is helpful for promoting the quality of healthcare service and providing advice on hospital operating.Thus,we attach importance to mining patient satisfaction in OHC,and further investigate its influence on patients'doctor selection behavior.We crawl data of haodf.com,which includes panel data of 3181 doctors and their 79.8.thousand of online reviews.Firstly,we mining patient satisfaction from text reviews using LDA model and summarize its contents into six topics,we sort them according to patients' attention,and the inverted order are:satisfaction of service attitude,satisfaction of technical skill,satisfaction of ethics,satisfaction of clarity of explanation,satisfaction of medical procedures and satisfaction of infrastructure.After that,we recognize topic sentences in text reviews,and then calculate satisfaction score of each topic sentence and each doctor through sentiment analysis.We divide satisfaction into several dimensions:overall satisfaction,satisfaction evaluation quantity(volume of reviews)and six topics of satisfaction mentioned before.This paper take them as independent variable.We regard doctor's title and tier level of hospital as doctor's properties and add them into the independent variable group.The risk of disease is a moderate variable.The dependent variable is the number of appointment every week.We also add some control variables into the model.The dependent variable is a count variable,and its standard deviation is much larger than its mean value.Thuswe take negative binomial regression model to analyze the panel data to verify our hypotheses.At last,we make robust checks to ensure the robustness and effectiveness of the regression results.The regression results indicate that patients prefer doctors who have higher title and working in hospitals with higher tier level.Overall satisfaction,volume of reviews,satisfaction of service attitude,satisfaction of technical skill,satisfaction of ethics,satisfaction of clarity of explanation are proved to have positive effects on patients'doctor selection.We also find that the risk of disease can positively moderate the influence of volume of reviews on appointments.Besides,there are negatively interaction effect between online recommendation score and doctor's title,namely,doctor can remedy their low title through getting a higher recommendation score.