การวิเคราะห์และแจ้งเตือนข้อร้องเรียนของผู้รับบริการด้านสุขภาพ

Abstract

The purpose of this independent study was to study the analysis and notification of complaints of health service recipients by using complaints from health care recipients and the severity of complaints assessed by experts. The study had foreseen the importance of prioritizing the problem solving of patients' complaints since complaints in hospitals could affect both the hospital and the patients. Therefore, the researcher conducted an independent study to analyze the level of severity of complaints requiring immediate action. The research started by studying the policy of complaint management and prioritizing the severity of the impact on service quality problems for complaint management. The level of severity was divided into 2 levels: the level that did not require warning was the level of complaints that were not severe and the level to be notified was complaints that were not severe but had a high risk of becoming severe, and severe level to be able to resolve complaints in a timely manner. The researcher obtained the information from the study to develop a system for analyzing complaints and notifying when there are complaints that are categorized as having to be resolved immediately through the LINE application of those involved in the handling of further complaints. The study found that from modeling complaints classification by TF-IDF feature extraction method, the efficiency of each model demonstrated that Multinomial NB had the highest efficiency in complaint classification compared to the Accuracy value of 71%. As a result, the Multinomial NB classification technique had the best accuracy compared to other classification techniques. The result of each complaint type of the Multinomial NB model where Support was the number of test datasets of complaints used to test the model. There were 161 complaints of data type 0 and 107 complaints of data type 1. There was still a difference in the number of complaints from both types. Therefore, the Precision, Recall, and F1 score must be taken into consideration. The F1 score indicates the performance of the Multinomial NB model calculated from the Precision and Recall values as the higher the F1 score, the better the model performs. When considering the precision value, it is a comparison of how many percent of each complaint classification is accurate. The model had the most accurate classification of complaints as type 0 complaints or complaints that did not need to be notified. The classification was correct 87% of the total number of 161 complaints, while type 1 complaints or complaints requiring notification were correctly classified 59% of the total number of 107 complaints and considering the Recall value as accuracy at what percentage of the classification was "true". The model classified 60% of Type 0 complaints as true and 87% of Type 1 complaints. When considering the overall picture from the F1 score, which is a value indicating the efficiency, it indicates that Multinomial NB can classify each type of complaint in good terms from all outcomes.