EUVIMEDEUVIMED entdecken
BeobachtungsstudieOpen Access · CC BY 4.0

Reinforcement Learning for Intraoperative Hypotension Management with Consideration to Postoperative Acute Kidney Injury

Esra Adiyeke iD, Tianqi Liu iD, Venkata Sai Dheeraj Naganaboina iD, Han Li iD, Tyler J Loftus, Yuanfang Ren iD, Benjamin Shickel iD, Matthew M Ruppert iD, Karandeep Singh iD, Ruogu Fang iD, Parisa Rashidi iD, Azra Bihorac iD und 1 weitere

Kidney360 · 2026-07 · 7 (7) · 1535-1544

DOI: 10.34067/kid.0000001212 ↗

Kurz gefasst

Beobachtungsstudie in der Zeitschrift Kidney360, online erschienen am 05.05.2026. Verfasst von Esra Adiyeke und 12 weiteren Autor:innen. Beteiligte Einrichtung: Intelligent Clinical Care Center, University of Florida, Gainesville, Florida. Hauptthemen laut MEDLINE: Postoperative Complications, Acute Kidney Injury, Intraoperative Complications, Hypotension. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NCATS NIH HHS, University of Florida Research, Clinical and Translational Science Institute, University of Florida.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Key Points Intraoperative hypotension is associated with postoperative AKI, which is a common and morbid postoperative complication. We developed a reinforcement learning model to guide intraoperative fluids and vasopressors, avoiding hypotension and postoperative AKI. The model's policy showcases the potential to lower postoperative AKI and improve outcomes driven by intraoperative hypotension.

Background

Traditional methods of surgical decision making heavily rely on human experience and prompt actions, which are variable. A data-driven system that generates treatment recommendations based on patient states can be a substantial asset in perioperative decision making for cases of intraoperative hypotension in which suboptimal management is associated with AKI, a common and morbid postoperative complication.

Methods

In this retrospective cohort study, we analyzed 50,021 surgeries from 42,547 adult patients who underwent major surgery at a quaternary care hospital between 2014 and 2020. We developed a deep reinforcement learning model to recommend the optimum doses of intravenous fluids and vasopressors during surgery to avoid intraoperative hypotension and AKI defined by Kidney Disease Improving Global Outcomes serum creatinine criteria within 3 days after surgery.

Results

The developed model replicated 69% of physicians' decisions for the dosage of vasopressors and proposed higher or lower dosage of vasopressors than received in 10% and 21% of the treatments, respectively. In terms of intravenous fluids, the model's recommendations were within 0.05 ml/kg per 15 minutes of the actual dose in 41% of the cases, with higher or lower doses recommended for 27% and 32% of the treatments, respectively. The reinforcement learning policy resulted in a higher estimated policy value compared with the physicians' actual treatments, as well as random policies and zero-drug policies. AKI incidence was the lowest in patients who received medication dosages that aligned with our agent model's decisions.

Conclusions

Our findings suggest that implementation of the model's policy has the potential to lower postoperative AKI and improve other outcomes driven by intraoperative hypotension.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Postoperative ComplicationsAcute Kidney InjuryIntraoperative ComplicationsHypotensionReinforcement Machine LearningRetrospective StudiesLongitudinal StudiesHumansMaleFemaleAdultMiddle AgedAgedTreatment OutcomeVasoconstrictor Agents

Stichwörter der Autor:innen

AKIBPacute renal failureartificial intelligencecohort studieshypotensionvasopressin

Substanzen

Vasoconstrictor Agents

Autor:innen und Institutionen

  • Intelligent Clinical Care Center, University of Florida, Gainesville, Florida.
  • Division of Nephrology, Hypertension, and Renal Transplantation, Department of Medicine, University of Florida, Gainesville, Florida.
  • Department of Electrical and Computer Engineering, University of Florida, Gainesville, Florida.
  • Department of Computer Science, University of Florida, Gainesville, Florida.
  • Department of Surgery, University of Florida, Gainesville, Florida.
  • Department of Medicine, University of California San Diego, San Diego, California.

ORCID-Profile: Esra Adiyeke, Tianqi Liu, Venkata Sai Dheeraj Naganaboina, Han Li, Yuanfang Ren, Benjamin Shickel, Matthew M Ruppert, Karandeep Singh, Ruogu Fang, Parisa Rashidi, Azra Bihorac, Tezcan Ozrazgat-Baslanti

Förderung

  • NCATS NIH HHS
  • University of Florida Research
  • Clinical and Translational Science Institute, University of Florida
  • NIDDK NIH HHS
  • National Institute of Diabetes and Digestive and Kidney Diseases
  • National Center for Advancing Translational Sciences

Zitieren

APA

Adiyeke, E., Liu, T., Naganaboina, V. S. D., Li, H., Loftus, T. J., Ren, Y., Shickel, B., Ruppert, M. M., Singh, K., Fang, R., Rashidi, P., Bihorac, A., & Ozrazgat-Baslanti, T. (2026). Reinforcement Learning for Intraoperative Hypotension Management with Consideration to Postoperative Acute Kidney Injury. Kidney360, 7(7), 1535-1544. https://doi.org/10.34067/kid.0000001212

Vancouver

Adiyeke E, Liu T, Naganaboina VSD, Li H, Loftus TJ, Ren Y, et al. Reinforcement Learning for Intraoperative Hypotension Management with Consideration to Postoperative Acute Kidney Injury. Kidney360. 2026;7(7):1535-1544. doi:10.34067/kid.0000001212

BibTeX
@article{Adiyeke2026,
  author = {Adiyeke, Esra and Liu, Tianqi and Naganaboina, Venkata Sai Dheeraj and Li, Han and Loftus, Tyler J and Ren, Yuanfang and Shickel, Benjamin and Ruppert, Matthew M and Singh, Karandeep and Fang, Ruogu and Rashidi, Parisa and Bihorac, Azra and Ozrazgat-Baslanti, Tezcan},
  title = {Reinforcement Learning for Intraoperative Hypotension Management with Consideration to Postoperative Acute Kidney Injury},
  journal = {Kidney360},
  year = {2026},
  volume = {7},
  number = {7},
  pages = {1535--1544},
  doi = {10.34067/kid.0000001212}
}
Kostenlos starten →