Predicting patient revisits at the University of Virginia Health System Emergency Department

Brady Fowler, Monica Rajendiran, Timothy Schroeder, Nicholas Bergh, Abigail Flower, Hyojung Kang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This study focuses on the predictive identification of patients frequently revisiting the University of Virginia Health System Emergency Department. Identifying these patients can help the Emergency Department formulate strategies that improve patient care and decrease excess Emergency Department utilization. The Health System in particular faces a number of unique challenges in its ongoing mission to reduce extraneous patient revisits. In addition to its status as an academic hospital, it serves a broad geographic region as one of five level-I trauma centers in the Commonwealth of Virginia. In this study we utilized 5 years of data from the University of Virginia Health System data warehouse. These data contain information on 91,297 patients and 196,902 unique encounters, including details on patient demographics, diagnoses and hospital departments visited. From these raw data we engineered features, trained gradient boosted decision trees, and experimented with unsupervised clustering techniques to best approximate 30-day Emergency Department revisit risk at the conclusion of each patient encounter. Our best model for revisit risk resulted in a Receiver Operator Characteristic Area Under the Curve of 0.75. Furthermore, we exhibit the real-time performance of our model as a tool to rank which at-risk patients should receive priority for Emergency Department resources. This test demonstrated a significant improvement over the current allocation of Emergency Department social worker resources with a daily Mean Average Precision of 0.83. The methodologies proposed in this paper exhibit an end-to-end framework to transform raw administrative claims and limited clinical data into predictive models that help the Emergency Department better manage resources and target interventions.

Original languageEnglish (US)
Title of host publication2017 Systems and Information Engineering Design Symposium, SIEDS 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages253-258
Number of pages6
ISBN (Electronic)9781538618486
DOIs
StatePublished - May 31 2017
Event2017 Systems and Information Engineering Design Symposium, SIEDS 2017 - Charlottesville, United States
Duration: Apr 28 2017 → …

Publication series

Name2017 Systems and Information Engineering Design Symposium, SIEDS 2017

Conference

Conference2017 Systems and Information Engineering Design Symposium, SIEDS 2017
CountryUnited States
CityCharlottesville
Period4/28/17 → …

Fingerprint

Health
Data warehouses
Decision trees
Mathematical operators
Emergency department
Resources

Keywords

  • Clustering
  • Emergency Department Revisit Predictions
  • Gradient Boosted Trees
  • Healthcare

ASJC Scopus subject areas

  • Hardware and Architecture
  • Information Systems and Management
  • Computer Science Applications
  • Information Systems
  • Control and Systems Engineering
  • Decision Sciences (miscellaneous)

Cite this

Fowler, B., Rajendiran, M., Schroeder, T., Bergh, N., Flower, A., & Kang, H. (2017). Predicting patient revisits at the University of Virginia Health System Emergency Department. In 2017 Systems and Information Engineering Design Symposium, SIEDS 2017 (pp. 253-258). [7937726] (2017 Systems and Information Engineering Design Symposium, SIEDS 2017). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/SIEDS.2017.7937726

Predicting patient revisits at the University of Virginia Health System Emergency Department. / Fowler, Brady; Rajendiran, Monica; Schroeder, Timothy; Bergh, Nicholas; Flower, Abigail; Kang, Hyojung.

2017 Systems and Information Engineering Design Symposium, SIEDS 2017. Institute of Electrical and Electronics Engineers Inc., 2017. p. 253-258 7937726 (2017 Systems and Information Engineering Design Symposium, SIEDS 2017).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Fowler, B, Rajendiran, M, Schroeder, T, Bergh, N, Flower, A & Kang, H 2017, Predicting patient revisits at the University of Virginia Health System Emergency Department. in 2017 Systems and Information Engineering Design Symposium, SIEDS 2017., 7937726, 2017 Systems and Information Engineering Design Symposium, SIEDS 2017, Institute of Electrical and Electronics Engineers Inc., pp. 253-258, 2017 Systems and Information Engineering Design Symposium, SIEDS 2017, Charlottesville, United States, 4/28/17. https://doi.org/10.1109/SIEDS.2017.7937726
Fowler B, Rajendiran M, Schroeder T, Bergh N, Flower A, Kang H. Predicting patient revisits at the University of Virginia Health System Emergency Department. In 2017 Systems and Information Engineering Design Symposium, SIEDS 2017. Institute of Electrical and Electronics Engineers Inc. 2017. p. 253-258. 7937726. (2017 Systems and Information Engineering Design Symposium, SIEDS 2017). https://doi.org/10.1109/SIEDS.2017.7937726
Fowler, Brady ; Rajendiran, Monica ; Schroeder, Timothy ; Bergh, Nicholas ; Flower, Abigail ; Kang, Hyojung. / Predicting patient revisits at the University of Virginia Health System Emergency Department. 2017 Systems and Information Engineering Design Symposium, SIEDS 2017. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 253-258 (2017 Systems and Information Engineering Design Symposium, SIEDS 2017).
@inproceedings{d27bc69be71244eb8b96d7a63fae8503,
title = "Predicting patient revisits at the University of Virginia Health System Emergency Department",
abstract = "This study focuses on the predictive identification of patients frequently revisiting the University of Virginia Health System Emergency Department. Identifying these patients can help the Emergency Department formulate strategies that improve patient care and decrease excess Emergency Department utilization. The Health System in particular faces a number of unique challenges in its ongoing mission to reduce extraneous patient revisits. In addition to its status as an academic hospital, it serves a broad geographic region as one of five level-I trauma centers in the Commonwealth of Virginia. In this study we utilized 5 years of data from the University of Virginia Health System data warehouse. These data contain information on 91,297 patients and 196,902 unique encounters, including details on patient demographics, diagnoses and hospital departments visited. From these raw data we engineered features, trained gradient boosted decision trees, and experimented with unsupervised clustering techniques to best approximate 30-day Emergency Department revisit risk at the conclusion of each patient encounter. Our best model for revisit risk resulted in a Receiver Operator Characteristic Area Under the Curve of 0.75. Furthermore, we exhibit the real-time performance of our model as a tool to rank which at-risk patients should receive priority for Emergency Department resources. This test demonstrated a significant improvement over the current allocation of Emergency Department social worker resources with a daily Mean Average Precision of 0.83. The methodologies proposed in this paper exhibit an end-to-end framework to transform raw administrative claims and limited clinical data into predictive models that help the Emergency Department better manage resources and target interventions.",
keywords = "Clustering, Emergency Department Revisit Predictions, Gradient Boosted Trees, Healthcare",
author = "Brady Fowler and Monica Rajendiran and Timothy Schroeder and Nicholas Bergh and Abigail Flower and Hyojung Kang",
year = "2017",
month = "5",
day = "31",
doi = "10.1109/SIEDS.2017.7937726",
language = "English (US)",
series = "2017 Systems and Information Engineering Design Symposium, SIEDS 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "253--258",
booktitle = "2017 Systems and Information Engineering Design Symposium, SIEDS 2017",
address = "United States",

}

TY - GEN

T1 - Predicting patient revisits at the University of Virginia Health System Emergency Department

AU - Fowler, Brady

AU - Rajendiran, Monica

AU - Schroeder, Timothy

AU - Bergh, Nicholas

AU - Flower, Abigail

AU - Kang, Hyojung

PY - 2017/5/31

Y1 - 2017/5/31

N2 - This study focuses on the predictive identification of patients frequently revisiting the University of Virginia Health System Emergency Department. Identifying these patients can help the Emergency Department formulate strategies that improve patient care and decrease excess Emergency Department utilization. The Health System in particular faces a number of unique challenges in its ongoing mission to reduce extraneous patient revisits. In addition to its status as an academic hospital, it serves a broad geographic region as one of five level-I trauma centers in the Commonwealth of Virginia. In this study we utilized 5 years of data from the University of Virginia Health System data warehouse. These data contain information on 91,297 patients and 196,902 unique encounters, including details on patient demographics, diagnoses and hospital departments visited. From these raw data we engineered features, trained gradient boosted decision trees, and experimented with unsupervised clustering techniques to best approximate 30-day Emergency Department revisit risk at the conclusion of each patient encounter. Our best model for revisit risk resulted in a Receiver Operator Characteristic Area Under the Curve of 0.75. Furthermore, we exhibit the real-time performance of our model as a tool to rank which at-risk patients should receive priority for Emergency Department resources. This test demonstrated a significant improvement over the current allocation of Emergency Department social worker resources with a daily Mean Average Precision of 0.83. The methodologies proposed in this paper exhibit an end-to-end framework to transform raw administrative claims and limited clinical data into predictive models that help the Emergency Department better manage resources and target interventions.

AB - This study focuses on the predictive identification of patients frequently revisiting the University of Virginia Health System Emergency Department. Identifying these patients can help the Emergency Department formulate strategies that improve patient care and decrease excess Emergency Department utilization. The Health System in particular faces a number of unique challenges in its ongoing mission to reduce extraneous patient revisits. In addition to its status as an academic hospital, it serves a broad geographic region as one of five level-I trauma centers in the Commonwealth of Virginia. In this study we utilized 5 years of data from the University of Virginia Health System data warehouse. These data contain information on 91,297 patients and 196,902 unique encounters, including details on patient demographics, diagnoses and hospital departments visited. From these raw data we engineered features, trained gradient boosted decision trees, and experimented with unsupervised clustering techniques to best approximate 30-day Emergency Department revisit risk at the conclusion of each patient encounter. Our best model for revisit risk resulted in a Receiver Operator Characteristic Area Under the Curve of 0.75. Furthermore, we exhibit the real-time performance of our model as a tool to rank which at-risk patients should receive priority for Emergency Department resources. This test demonstrated a significant improvement over the current allocation of Emergency Department social worker resources with a daily Mean Average Precision of 0.83. The methodologies proposed in this paper exhibit an end-to-end framework to transform raw administrative claims and limited clinical data into predictive models that help the Emergency Department better manage resources and target interventions.

KW - Clustering

KW - Emergency Department Revisit Predictions

KW - Gradient Boosted Trees

KW - Healthcare

UR - http://www.scopus.com/inward/record.url?scp=85025611303&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85025611303&partnerID=8YFLogxK

U2 - 10.1109/SIEDS.2017.7937726

DO - 10.1109/SIEDS.2017.7937726

M3 - Conference contribution

AN - SCOPUS:85025611303

T3 - 2017 Systems and Information Engineering Design Symposium, SIEDS 2017

SP - 253

EP - 258

BT - 2017 Systems and Information Engineering Design Symposium, SIEDS 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -