Feature augmented stacked bagging for length of stay prediction on mimic III

Abstract

Managing hospital resources is one of the major requirements in providing qualitative and timely treatments do patients. From the patient’s standpoint, identifying the number of days of stay can aid in effective planning from the monetary perspective.This work presents the Feature Augmented Stacked Bagging (FASB) model for effective prediction of length of stay of patient based on their hospital records. The ensemble model created is composed of two levels; the first level operates on the hospital data and provides the intermediate predictions, while the second level uses predictions from the first level model to perform the final predictions. Experiments where performed using the MIMIC III data, and the results obtained were compared with existing state of the art models for analysis. Comparisons indicate that the proposed model exhibits 95% accuracy levels, which is 13% greater than the compared model. This indicates that the proposed model is highly effective in identifying the length of stay of patients.