College

Buchtel College of Arts and Sciences

Date of Last Revision

2026-05-07 06:09:15

Major

Statistics

Honors Course

STAT 498

Number of Credits

3

Degree Name

Bachelor of Science

Date of Expected Graduation

Spring 2026

Abstract

Unplanned 30-day hospital readmission remains a fundamental challenge in US healthcare, associated with increased risk to patient recovery and representing an estimated $52.4 billion in annual expenses (Beauvais et al., 2022). While the rigorously validated LACE index serves as the clinical standard for readmission modeling, its linear structure and four explanatory variables lack the complexity to capture the high-dimensional and interactive nature of patient risk. This study utilizes an admission granularity level cohort of the MIMIC-IV database to develop and compare machine learning architectures against the baseline LACE index. Due to the imbalanced prevalence of readmission, the penalized logistic regression, CART decision tree, random forest, XGBoost, and LACE index models were tuned and evaluated using the area under the receiver operator characteristic curve (AUC-ROC) metric, which measures discrimination across all decision thresholds. These findings suggest that the four machine learning architectures provide a notable lift above the LACE index (AUC 0.6471), with the XGBoost model (AUC 0.7310) achieving the highest discriminatory ability. Nonparametric ensemble methods, such as the random forest and XGBoost architectures, excel at modeling complex interactions between an array of high-dimensional predictors. To address the “black box” nature of these models, SHAP values were leveraged to provide insight into model logic and interpret prediction contributions. This research explores how modern machine learning methodologies, paired with explainable AI tools, can bridge the traditional divide between predictive power, actionable insight, and clinical trust.

Research Sponsor

Dr. Nao Mimoto

First Reader

Dr. Mark Fridline

Second Reader

Dr. Jun Ye

Honors Faculty Advisor

Dr. Nao Mimoto

Proprietary and/or Confidential Information

No

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.