Personalized Predictions for Unplanned Urinary Tract Infection Hospitalizations with Hierarchical Clustering

EasyChair Preprint 4945

Abstract

Urinary Tract Infection (UTI) is the most frequent and preventable healthcare-associated infection in the US and an important cause of morbidity and excess healthcare costs. This study aims to predict the 30-day risk of a beneficiary for unplanned hospitalization for UTI. Using 2008-12 Medicare fee-for-service claims and several public sources, we extracted 784 potential variables, including patient demographics, clinical conditions, healthcare utilization, provider quality metrics, and community safety indicators. To address the challenge of high heterogeneity and imbalance in data, we propose a hierarchical clustering framework that separates patients of similar characteristics based on causal knowledge and supervised data-driven rules, thus increasing prediction accuracy via more finer-grained prediction models. Our prediction models are trained via 237,675 2011 Medicare beneficiaries and tested over 230,042 2012 Medicare beneficiaries. We evaluate the effectiveness of the framework using five performance metrics, including the area under the curve (AUC), the True Positive Rate (TPR), and the False Positive Rate (FPR). Results show that the hierarchical clustering approach achieves more accurate and precise predictions than a benchmark model without clustering.

Keyphrases: Hospitals, Urinary tract infection, cluster analysis, data analysis, health care, personalized prediction model, statistics

@booklet{EasyChair:4945,
  author    = {Lingchao Mao and Kimia Vahdat and Sara Shashaani and Julie Swann},
  title     = {Personalized Predictions for Unplanned Urinary Tract Infection Hospitalizations with Hierarchical Clustering},
  howpublished = {EasyChair Preprint 4945},
  year      = {EasyChair, 2021}}