Levin S, Toerper M, Hamrock E, Hinson J, Barnes S, Gardner H, Dugas A, Linton B, Kirsch T, Kelen G. Machine Learning-Based Triage More Accurately Differentiates Patients with Respect to Clinical Outcomes Compared to the Emergency Severity Index. Ann Emerg Med. 71(5):565-574, 2018.

E-Triage highlights opportunities for machine learning and predictive analytics to improve emergency triage decision-making.

Levin S, Toerper M, Dugas A, Kelen G. E-Triage: An electronic emergency triage system. US Patent App No. 62296753. Filed 2/15, 2017.

E-Triage patent.


Levin S, Toerper M, Hinson J, Gardner H, Henry S, McKenzie C, Whalen M, Hamrock E, Barnes S, Martinez D, Kelen G. Machine-Learning Based Electronic Triage: A Prospective Evaluation. Ann Emerg Med. 72(4), S116.

Harmonizing nurses’ clinical judgment with data-driven e-triage decision support was capable of improving differentiation of patients measured by clinical outcomes at triage.


Chan W, Mason J, Grock A. The Long and Winding Triage Road. Ann Emerg Med. 71:575-577, 2018.

E-triage is done autonomously by a machine, which may sound a little 2001: A Space Odyssey, but just remember that about 90% of your commercial flight time is spent in autopilot these days. E-Triage is not meant to replace the triage provider; it spits out a triage-level score that triagers can override according to patient appearance, clinical history, and gestalt, just like they do with the current ESI system.


Schenkel S, Wears R. Triage, Machine Learning, Algorithms, and Becoming the Borg. Ann Emerg Med. 71:578-580, 2018. 


StoCastic. Data-Driven Decision Support for Efficient Patient Progression. National Science Foundation Award 1738440, 2018. 

Grant award to develop and commercialize E-Triage as part of StoCastic’s patient progression platform.


StoCastic. Risk Adaptive Triage in Emergency Medicine. National Science Foundation Award 1621899, 2017. 

Grant award to develop and commercialize E-Triage as part of StoCastic’s patient progression platform.


Barnes S, Saria S, Levin S. An Evolutionary Computation Approach for Optimizing Multi-Level Data to Predict Patient Outcomes. Journal of Healthcare Engineering. ID 7174803, 2018.

Evolutionary computation approaches may be used to optimize multi-level healthcare data for predictive models.


Dugas A, Kirsch T, Toerper M, Korley F, Yenokyan G, France D, Hager D, Levin S.  An Electronic Emergency Triage System Improves Patient Distribution. J Emerg Med. 50(6):910-918, 2016.

E-Triage is a triage system based on the frequency of critical outcomes that demonstrate improved differentiation of patients compared to the current standard ESI


Levin S, France DJ, Aronsky D. Implementation of a Computerized Triage System in the Emergency Department. Transforming Health Care through Information: 3rd Edition. Eds. Einbeinder L. Lorenzi N. Ash J. Gadd C. Springer Science. 236-269, 2009.


Hinson J, Martinez D, Cabral S, George K, Whalen M, Hansoti B, Levin S. Triage Performance in Emergency Medicine: A Systematic Review. Ann Emerg Med. 74(1):140-152, 2019.

Using standardized triage systems, including ESI, a substantial proportion of ED patients who die post-encounter or are critically ill are not designated as high acuity at triage.


Mistry B, Stewart De Ramirez S, Kelen G, Schmitz P, Balhara K, Levin S, Martinez D, Psoter K, Anton X, Hinson J. Accuracy and Reliability of Emergency Department Triage using the Emergency Severity Index: An International Multicenter Assessment. Ann Emerg Med. 71(5):581-587, 2017.

In this multinational study, concordance of nurse-assigned ESI score with reference standard was universally poor and variability was high.


Hinson J, Martinez D, Schmitz P, Toerper M, Radu D, Scheulen J, Stewart De Ramirez S, Levin S. Accuracy of the Emergency Severity Index and Independent Predictors of Under-Triage and Over-Triage in Brazil: A Retrospective Cohort Analysis. Int J Emerg Med. 11(1), 2018. 

Despite rigorous and ongoing training of ESI users, a large number of patients in this cohort were under- or over-triaged. Advanced age, vital sign derangements, and specific chief complaints—all subject to limited guidance by the ESI algorithm—were particularly under-appreciated.


Balhara K, Mistry B, Hinson J, Levin S, De Ramirez S. Nursing perceptions of the Emergency Severity Index as a Triage Tool in the United Arab Emirates. J Emerg Nurs. 44(4):360-367, 2017.

Although emergency nurses perceive the ESI as easy to use, there are concerns about the subjectivity and variability inherent in the ESI that can lead to a functional lack of triage and a burden of undifferentiated ESI level 3 patients.


Lentz BA, Jenson A, Hinson JS, Levin S, Cabral S, George K, Hsu EB, Kelen G, Hansoti B. Validity of emergency department triage tools: Addressing heterogeneous definitions of over-triage and under-triage. Am J Emerg. 35(7):1023-1025, 2017.

Based on the results of our systematic review and our recent survey of experts in the field of emergency department triage, we suggest that future investigations of triage tool validity utilize outcome-based metrics for defining over- under-triage when possible and that reporting of over− under-triage be based on sensitivity and specificity.




Barnes S, Hamrock E, Toerper M, Siddiqui S, Levin S. Real-Time Prediction of Inpatient Length of Stay for Discharge Prioritization. J Am Med Inform Assoc. 23:2-10, 2016.

There is potential to use readily available health information to predict daily patient discharges with accuracies comparable to clinician predictions.


Levin S, Zeger S, McCarthy M, Fackler J. Hospital unit demand forecasting tool. 2016; US Patent No. 20140136458. Granted 4/12, 2016.

E-LOS Patent.


Levin S, Zeger S, McCarthy M, Fackler J, France D. Forecasting Demand for Pediatric Critical Care. National Science Foundation Award 1738440, 2018. 

Grant award to develop E-LOS for the pediatric intensive care unit.


Levin S, Harley E, Fackler J, Lehmann C, Custer J, France D, Zeger S. Real Time Forecasting of Pediatric Intensive Care Unit Length of Stay Using Computerized Provider Orders. Crit Care Med. 40(11):3058-64, 2012.

Provider orders reflect dynamic changes in patients' conditions, making them useful for real-time length of stay prediction and patient flow management.