| Summary: | 碩士 === 明志科技大學 === 工業工程與管理研究所 === 99 === Medical treatment plays a substantial part in health services with an objective of curing and improving the life quality of patients. Due to increase in drug categories in recent years, there are more treatment options available to doctors. In addition, given that doctors in different hospital departments share various medication experiences and patients are likely to see doctors of different departments in short gap, the possibility of patients receiving repeated prescriptions and drug interactions is increasing. Consequently patients’ health is at risk and the medical resources are wasted. Accordingly the use of data mining techniques is expected to show the seriousness and percentage of drug interactions and to provide information for health service institutions to help elevate the healthcare quality.
This research uses the database from a teaching hospital in Taipei to investigate the association between drug interactions and prescriptions in different departments. First of all, the differences in seriousness and percentage of drug interactions among various departments are analyzed, and then the association between prescriptions and drug interactions. The results show: (1) four out of every 100 patients are under the threat of drug interactions; (2) there is a higher likelihood of patients in the divisions of Cardiology and Cardiovascular Surgery to be affected by drug interactions; (3) nearly 70% of first-order drug-drug interactions cases are from Cardiology and Cardiovascular Surgery, which is distinctively higher than the drug interactions in the other departments; (4) as people age, there is an evident increase in drug interactions; (6) among the first five combinations of first-order drug-drug interactions, four of them which contain Digoxin would cause cardiac arrhythmia; (7) among the first five combinations of second-order drug-drug interactions, four of them which contain Aspirin would cause stomach upset and even lead to hemorrhagic ulcer.
On the other hand, in a further data mining, cluster analyses identify six representative patient groups, which may provide a reference for the health service institutions to simplify data. In respect of the association rule, this research investigates the relationships between patients’ sex, age, hospital department, diagnosis code, drug category, and the level of interactions and obtains 71 association rules. After organizing and compiling, there are 7 association rules. They may become a reference for health service institutions in developing the prescription system and the regulation of making medications decisions.
|
|---|
