Real-time prediction of inpatient length of stay for discharge prioritization

Sean Barnes, Eric Hamrock, Matthew Toerper, Sauleh Siddiqui, Scott Levin

Research output: Contribution to journalArticle

Abstract

Objective: Hospitals are challenged to provide timely patient care while maintaining high resource utilization. This has prompted hospital initiatives to increase patient flow and minimize nonvalue added care time. Real-time demand capacity management (RTDC) is one such initiative whereby clinicians convene each morning to predict patients able to leave the same day and prioritize their remaining tasks for early discharge. Our objective is to automate and improve these discharge predictions by applying supervised machine learning methods to readily available health information. Materials and Methods: The authors use supervised machine learning methods to predict patients' likelihood of discharge by 2 p.m. and by midnight each day for an inpatient medical unit. Using data collected over 8000 patient stays and 20 000 patient days, the predictive performance of the model is compared to clinicians using sensitivity, specificity, Youden's Index (i.e., sensitivity + specificity - 1), and aggregate accuracy measures. Results: The model compared to clinician predictions demonstrated significantly higher sensitivity (P.10). Early discharges were less predictable than midnight discharges. The model was more accurate than clinicians in predicting the total number of daily discharges and capable of ranking patients closest to future discharge. Conclusions: There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions. This approach may be used to automate and support daily RTDC predictions aimed at improving patient flow.

Original languageEnglish (US)
Pages (from-to)e2-e10
JournalJournal of the American Medical Informatics Association
Volume23
Issue numbere1
DOIs
StatePublished - Apr 1 2016

Fingerprint

Inpatients
Length of Stay
Patient Discharge
Sensitivity and Specificity
Health
Patient Care
Supervised Machine Learning

Keywords

  • Length of stay
  • Machine learning
  • Operational forecasting
  • Patient flow

ASJC Scopus subject areas

  • Health Informatics

Cite this

Real-time prediction of inpatient length of stay for discharge prioritization. / Barnes, Sean; Hamrock, Eric; Toerper, Matthew; Siddiqui, Sauleh; Levin, Scott.

In: Journal of the American Medical Informatics Association, Vol. 23, No. e1, 01.04.2016, p. e2-e10.

Research output: Contribution to journalArticle

Barnes, Sean ; Hamrock, Eric ; Toerper, Matthew ; Siddiqui, Sauleh ; Levin, Scott. / Real-time prediction of inpatient length of stay for discharge prioritization. In: Journal of the American Medical Informatics Association. 2016 ; Vol. 23, No. e1. pp. e2-e10.
@article{d58f34702e964b098e410cca43937285,
title = "Real-time prediction of inpatient length of stay for discharge prioritization",
abstract = "Objective: Hospitals are challenged to provide timely patient care while maintaining high resource utilization. This has prompted hospital initiatives to increase patient flow and minimize nonvalue added care time. Real-time demand capacity management (RTDC) is one such initiative whereby clinicians convene each morning to predict patients able to leave the same day and prioritize their remaining tasks for early discharge. Our objective is to automate and improve these discharge predictions by applying supervised machine learning methods to readily available health information. Materials and Methods: The authors use supervised machine learning methods to predict patients' likelihood of discharge by 2 p.m. and by midnight each day for an inpatient medical unit. Using data collected over 8000 patient stays and 20 000 patient days, the predictive performance of the model is compared to clinicians using sensitivity, specificity, Youden's Index (i.e., sensitivity + specificity - 1), and aggregate accuracy measures. Results: The model compared to clinician predictions demonstrated significantly higher sensitivity (P.10). Early discharges were less predictable than midnight discharges. The model was more accurate than clinicians in predicting the total number of daily discharges and capable of ranking patients closest to future discharge. Conclusions: There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions. This approach may be used to automate and support daily RTDC predictions aimed at improving patient flow.",
keywords = "Length of stay, Machine learning, Operational forecasting, Patient flow",
author = "Sean Barnes and Eric Hamrock and Matthew Toerper and Sauleh Siddiqui and Scott Levin",
year = "2016",
month = "4",
day = "1",
doi = "10.1093/jamia/ocv106",
language = "English (US)",
volume = "23",
pages = "e2--e10",
journal = "Journal of the American Medical Informatics Association : JAMIA",
issn = "1067-5027",
publisher = "Oxford University Press",
number = "e1",

}

TY - JOUR

T1 - Real-time prediction of inpatient length of stay for discharge prioritization

AU - Barnes, Sean

AU - Hamrock, Eric

AU - Toerper, Matthew

AU - Siddiqui, Sauleh

AU - Levin, Scott

PY - 2016/4/1

Y1 - 2016/4/1

N2 - Objective: Hospitals are challenged to provide timely patient care while maintaining high resource utilization. This has prompted hospital initiatives to increase patient flow and minimize nonvalue added care time. Real-time demand capacity management (RTDC) is one such initiative whereby clinicians convene each morning to predict patients able to leave the same day and prioritize their remaining tasks for early discharge. Our objective is to automate and improve these discharge predictions by applying supervised machine learning methods to readily available health information. Materials and Methods: The authors use supervised machine learning methods to predict patients' likelihood of discharge by 2 p.m. and by midnight each day for an inpatient medical unit. Using data collected over 8000 patient stays and 20 000 patient days, the predictive performance of the model is compared to clinicians using sensitivity, specificity, Youden's Index (i.e., sensitivity + specificity - 1), and aggregate accuracy measures. Results: The model compared to clinician predictions demonstrated significantly higher sensitivity (P.10). Early discharges were less predictable than midnight discharges. The model was more accurate than clinicians in predicting the total number of daily discharges and capable of ranking patients closest to future discharge. Conclusions: There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions. This approach may be used to automate and support daily RTDC predictions aimed at improving patient flow.

AB - Objective: Hospitals are challenged to provide timely patient care while maintaining high resource utilization. This has prompted hospital initiatives to increase patient flow and minimize nonvalue added care time. Real-time demand capacity management (RTDC) is one such initiative whereby clinicians convene each morning to predict patients able to leave the same day and prioritize their remaining tasks for early discharge. Our objective is to automate and improve these discharge predictions by applying supervised machine learning methods to readily available health information. Materials and Methods: The authors use supervised machine learning methods to predict patients' likelihood of discharge by 2 p.m. and by midnight each day for an inpatient medical unit. Using data collected over 8000 patient stays and 20 000 patient days, the predictive performance of the model is compared to clinicians using sensitivity, specificity, Youden's Index (i.e., sensitivity + specificity - 1), and aggregate accuracy measures. Results: The model compared to clinician predictions demonstrated significantly higher sensitivity (P.10). Early discharges were less predictable than midnight discharges. The model was more accurate than clinicians in predicting the total number of daily discharges and capable of ranking patients closest to future discharge. Conclusions: There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions. This approach may be used to automate and support daily RTDC predictions aimed at improving patient flow.

KW - Length of stay

KW - Machine learning

KW - Operational forecasting

KW - Patient flow

UR - http://www.scopus.com/inward/record.url?scp=84964921508&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84964921508&partnerID=8YFLogxK

U2 - 10.1093/jamia/ocv106

DO - 10.1093/jamia/ocv106

M3 - Article

C2 - 26253131

AN - SCOPUS:84964921508

VL - 23

SP - e2-e10

JO - Journal of the American Medical Informatics Association : JAMIA

JF - Journal of the American Medical Informatics Association : JAMIA

SN - 1067-5027

IS - e1

ER -