Publication Date: 2023/12/08
Abstract: This project aims to enhance hospital management by predicting patients’ length of stay using the MIMIC dataset, ultimately resulting in substantial cost savings and improved resource allocation. In our initial approach, we categorized the target variable, “length of stay” into three classes: short, medium, and long. Employing classification models including Logistic Regression, Random Forests, and Gradient Boosting, we attempted to predict patient outcomes. However, the initial results were unsatisfactory, prompting us to refine our methodology. We expanded the target variable classes to five: very short, short, medium, long, and very long, leading to improved accuracy in predicting short hospital stays. In the second approach, we treated the length of stay as a continuous variable and employed Multiple Linear Regression for modeling. Unfortunately, this ap- proach yielded sub-optimal results compared to the classification techniques. We analyzed the encountered limitations and further propose future steps to enhance the efficiency and accuracy of prediction models, ultimately contributing to more effective hospital resource management.
Keywords: Length of Stay, MIMIC III, Classification, Random Forest, Healthcare.
DOI: https://doi.org/10.5281/zenodo.10297083
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT23NOV1911.pdf
REFERENCES