Model Keputusan untuk Transfer ICU yang Tidak Direncanakan di Sebuah Rumah Sakit Menggunakan Pendekatan Optimasi Deep Learning

Abstract

Unplanned ICU transfer is one of the most important initial decisions after pa- tient management in the ER because apart from being an indicator of the quality of care for ER practitioners, it is also needed to achieve health goals, namely im- proving the quality of critical care and preventing death. Research that has been done in predicting the initial decision of unplanned ICU transfer using univari- ate analysis and logistic regression analysis as well as deep learning optimization (association rule). The association rule algorithm generates rules that are used to form a decision model for unplanned ICU transfers. In this study, we compare two association rule algorithms to get a more efficient algorithm in generating rules. The results of the study obtained that the Apriori Algorithm requires a completion time of 3 ms and the completion time required by the FP-Growth Al- gorithm is 31 ms so that the FP-Growth Algorithm is 28 ms more efficient than the Apriori Algorithm, while for rule generation, the resulting number is the same as 67 rules. Only 11 rules meet the minsupp and minconf thresholds and include the set of Class Association Rules (CAR) which are used to form a decision model for unplanned ICU transfers with binary integer programming.