Analyzing medical transaction data by using association rule mining with multiple minimum supports

Abstract

The quick development of IS has a huge impact on the healthcare industry. almost all the existing hospitals, clinics and other healthcare-related institutes have adopted a functionally powerful and highly integrated Hospital Information System (HIS) for management of clinic or medical-related affairs. The medical data stored in the HIS are collected from many different medical subsystems, However, problems of failed data sharing and inconsistent data content often occur among these subsystems, resulting in many hospitals collect a large amount of medical data, but not the ability to process and analyse these data properly, letting the valuable data in the HIS all go to waste. In this study, we made a practical visit to a certain hospital in Taiwan and collected radioimmunoassay (RIA) data from the Laboratory Information System (LIS) and the Departmental Registration System (DRS) of this hospital. Further, we proposed a method of the association rule mining in combination with the concept of multiple minimum supports to analyse and find valuable association rules from the RIA data. The analytical results found the method we proposed can indeed find association rules that would not be able to be found with the traditional association mining methods. It is very helpful in improving doctorpatient relationship and upgrading health care quality.