Pre-Processing Data Admisi dan ICU Rumah Sakit untuk Prediksi Transfer ICU

Abstract

In the field of healthcare analytics, the importance of data quality and completeness cannot be overstated. The recurring nature of pandemics also serves as a motivation for this study, which focuses on preprocessing admission and ICU data linked to patient parameters obtained from the Emergency Department (ED). Vital signs data from the ED are compared using the NEWS2 scoring model, while triage data are compared using the TREWS scoring model. The data preprocessing process adheres to the basic rules of EDPAI, and its outcomes are subsequently analyzed to identify risk factors and significant patterns that influence conclusions based on data distribution. Validation of the conducted tests is performed using accuracy, precision, recall, F1-score, and confusion matrix performance scales. The research findings demonstrate the success of the performed data preprocessing, where the Random Forest model outperforms traditional scoring systems in predicting ICU occurrences.