Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025

Eugenia Miranti; Timothy Keyes; Alvaro Ayala; Nerissa Ambers; Gina Newman; Elmer De Leon; Erika Paola Viana-Cardenas; Wajeeha Tariq; Mindy Sampson; Jorge L. Salinas

doi:10.1017/ice.2026.10432

Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025

Eugenia Miranti
, Timothy Keyes
, Alvaro Ayala
, Nerissa Ambers
, Gina Newman
, Elmer De Leon
, Erika Paola Viana-Cardenas
, Wajeeha Tariq
, Mindy Sampson
, Jorge L. Salinas

Research output: Contribution to journal › Article › peer-review

Abstract

Our study evaluated a large language model (gpt-4o-mini) for surgical site infection (SSI) adjudication, achieving 100% sensitivity but 69.4% specificity. While reducing the manual screening workload by 66%, the agent generated many false positives, underscoring the need for refined models to improve specificity without compromising accuracy.

Original language	English
Pages (from-to)	633-635
Number of pages	3
Journal	Infection Control and Hospital Epidemiology
Volume	47
Issue number	6
DOIs	https://doi.org/10.1017/ice.2026.10432
State	Published - Jun 1 2026

Access to Document

10.1017/ice.2026.10432

Cite this

Miranti, E., Keyes, T., Ayala, A., Ambers, N., Newman, G., De Leon, E., Viana-Cardenas, E. P., Tariq, W., Sampson, M., & Salinas, J. L. (2026). Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025. Infection Control and Hospital Epidemiology, 47(6), 633-635. https://doi.org/10.1017/ice.2026.10432

@article{d9016b9926854525b1a9fa8fd99a2a6f,

title = "Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025",

abstract = "Our study evaluated a large language model (gpt-4o-mini) for surgical site infection (SSI) adjudication, achieving 100\% sensitivity but 69.4\% specificity. While reducing the manual screening workload by 66\%, the agent generated many false positives, underscoring the need for refined models to improve specificity without compromising accuracy.",

author = "Eugenia Miranti and Timothy Keyes and Alvaro Ayala and Nerissa Ambers and Gina Newman and \{De Leon\}, Elmer and Viana-Cardenas, \{Erika Paola\} and Wajeeha Tariq and Mindy Sampson and Salinas, \{Jorge L.\}",

note = "Publisher Copyright: {\textcopyright} The Author(s), 2026.",

year = "2026",

month = jun,

day = "1",

doi = "10.1017/ice.2026.10432",

language = "English",

volume = "47",

pages = "633--635",

journal = "Infection Control and Hospital Epidemiology",

issn = "0899-823X",

number = "6",

}

Miranti, E, Keyes, T, Ayala, A, Ambers, N, Newman, G, De Leon, E, Viana-Cardenas, EP, Tariq, W, Sampson, M & Salinas, JL 2026, 'Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025', Infection Control and Hospital Epidemiology, vol. 47, no. 6, pp. 633-635. https://doi.org/10.1017/ice.2026.10432

TY - JOUR

T1 - Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025

AU - Miranti, Eugenia

AU - Keyes, Timothy

AU - Ayala, Alvaro

AU - Ambers, Nerissa

AU - Newman, Gina

AU - De Leon, Elmer

AU - Viana-Cardenas, Erika Paola

AU - Tariq, Wajeeha

AU - Sampson, Mindy

AU - Salinas, Jorge L.

PY - 2026/6/1

Y1 - 2026/6/1

N2 - Our study evaluated a large language model (gpt-4o-mini) for surgical site infection (SSI) adjudication, achieving 100% sensitivity but 69.4% specificity. While reducing the manual screening workload by 66%, the agent generated many false positives, underscoring the need for refined models to improve specificity without compromising accuracy.

AB - Our study evaluated a large language model (gpt-4o-mini) for surgical site infection (SSI) adjudication, achieving 100% sensitivity but 69.4% specificity. While reducing the manual screening workload by 66%, the agent generated many false positives, underscoring the need for refined models to improve specificity without compromising accuracy.

UR - https://www.scopus.com/pages/publications/105036523125

U2 - 10.1017/ice.2026.10432

DO - 10.1017/ice.2026.10432

M3 - Article

C2 - 41972262

AN - SCOPUS:105036523125

SN - 0899-823X

VL - 47

SP - 633

EP - 635

JO - Infection Control and Hospital Epidemiology

JF - Infection Control and Hospital Epidemiology

IS - 6

ER -

Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections - Northern California, 2025

Abstract

Access to Document

Other files and links

Fingerprint

Cite this