The use of classification algorithms and partial least squares for patient satisfaction

Hung-Pin Hou, Ping-Feng Pai, Hsin-Mei Lin
2018 Journal of Scientific and Engineering Research   unpublished
After customer satisfaction investigation has been widely applied in many countries, the medical insurance payment system has change from pay by service to pay by quality, even pay by satisfaction. Therefore, how to collaborate with medical professionals and patient experiences have revealed the important issue for both medical providers and patients. Using partial least squares (PLS) and classification algorithms to examine 237 patients interviewed on Maternal in clinics and hospitals of
more » ... . Firstly, we proposed a measurement structure of patient satisfaction index (PSI) model based on direct and indirect experiences of medical service form patient. Our study shows that the continued medical treatment of patients and the accountability of medical staffs perceived by patients play fully mediating effect between medical institute"s reputation, comprehensiveness, and accessibility and patient satisfaction. Among the routes, the medical institute"s reputation has the highest influence upon patient satisfaction through continued medical treatment of patients and the accountability of medical staffs that patients perceived. Secondly, using classification algorithms exploring the majority influence factors which correlative with the patient satisfaction we proved that decision tree result shows consistence with PSI model. Thirdly, we proposed the Confusing-Satisfaction Matrix for pruning skewed distribution and poor data to strengthen the PSI model from a weak one to a strong one. For some of the weak structure of exploratory research, we provide an effective way for data cleaning and sampling.