Knowledge abstraction from MIMIC II using Apriori algorithm for clinical decision support system

Abstract

Clinical Decision Support Systems provide physicians with a sustainable solution to treat their patients more effectively and improve the quality of care provided to them. These Systems can interpret large volumes of real-time patient data and provide doctors with a snapshot view of actionable information, ultimately allowing them to make better decisions, intervene in a timelier manner. icuARM is one of the Clinical Decision Support Systems, developed so far. This System uses MIMIC II, a publicly accessible Database created to archive records involving multimodal measurement data about ICU patients. icuARM is a tool which uses association rule mining technique that analyzes data for frequent patterns in the database. Overall, icuARM is a CDSS that assists physicians in choosing proper medication based on clinical status of patient's in real time, which will substantially improve the efficiency, accuracy and timeliness for clinical decision making in intensive care. The proposed Apriori association rule mining algorithm generates pattern to associate type of medication to the disease suffered by ICU Patients. The proposed approach decreases the probability of prolonging the patients stay in the ICU and improve their condition. The confidence score achieved in the algorithm is very effective in providing proper medication to the patients.