Reinforcement Learning to Optimize Ventilator Settings for Patients on Invasive Mechanical Ventilation: Retrospective Study
Reinforcement Learning zur Optimierung von Beatmungseinstellungen bei Patient:innen unter invasiver mechanischer Beatmung: retrospektive Studie
Journal of Medical Internet Research · 2024-10-16 · 26 · e44494
DOI: 10.2196/44494 ↗Auf einen Blick
Eine KI auf Basis von Reinforcement Learning wurde mit ICU-Daten aus den USA trainiert. Die geschätzte Krankenhausmortalität unter der lernbasierten Policy lag niedriger als die beobachtete klinische Praxis.
Fragestellung
Kann eine auf Reinforcement Learning basierende KI optimale Beatmungseinstellungen für kritisch kranke Patient:innen unter invasiver mechanischer Beatmung ableiten?
Methode
Retrospektive Studie mit Beobachtungsdaten aus mehreren ICUs in den USA; Off-policy-Evaluation der KI-Policy mit eICU und MIMIC-IV.
Ergebnisse
21.595 ICU-Aufenthalte aus eICU und 5105 aus MIMIC-IV. Geschätzte Krankenhausmortalität: eICU 12,1 % (SD 3,1 %), MIMIC-IV 29,1 % (SD 0,9 %).
Schlussfolgerung der Autor:innen
Die Anpassung der Beatmungseinstellungen an einzelne Patient:innen war mit niedrigeren geschätzten Krankenhausmortalitätsraten als die tatsächlichen Raten verbunden. Die Autor:innen sehen Potenzial für weitere prospektive Validierung.
Kernaussagen
- Geschätzte Krankenhausmortalität unter der KI-Policy niedriger als beobachtet
- Trainiert mit Daten aus eICU und MIMIC-IV
- Potenzial für Einsatz in einem klinischen Entscheidungssystem
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
At a glance
Journal article in Journal of Medical Internet Research, published online on 2024-10-16. Written by Siqi Liu and 7 co-authors. Institution: National University of Singapore Graduate School for Integrative Science and Engineering, National University of Singapore, Singapore, Singapore. Major MEDLINE topics: Respiration, Artificial. Cited 27 times so far (Crossref); 35 references. The full text is freely available in PubMed Central.
Topics: Beatmung und Weaning
Abstract (original language)
Background
One of the significant changes in intensive care medicine over the past 2 decades is the acknowledgment that improper mechanical ventilation settings substantially contribute to pulmonary injury in critically ill patients. Artificial intelligence (AI) solutions can optimize mechanical ventilation settings in intensive care units (ICUs) and improve patient outcomes. Specifically, machine learning algorithms can be trained on large datasets of patient information and mechanical ventilation settings. These algorithms can then predict patient responses to different ventilation strategies and suggest personalized ventilation settings for individual patients.
Objective
In this study, we aimed to design and evaluate an AI solution that could tailor an optimal ventilator strategy for each critically ill patient who requires mechanical ventilation.
Methods
We proposed a reinforcement learning–based AI solution using observational data from multiple ICUs in the United States. The primary outcome was hospital mortality. Secondary outcomes were the proportion of optimal oxygen saturation and the proportion of optimal mean arterial blood pressure. We trained our AI agent to recommend low, medium, and high levels of 3 ventilator settings—positive end-expiratory pressure, fraction of inspired oxygen, and ideal body weight–adjusted tidal volume—according to patients’ health conditions. We defined a policy as rules guiding ventilator setting changes given specific clinical scenarios. Off-policy evaluation metrics were applied to evaluate the AI policy.
Results
We studied 21,595 and 5105 patients’ ICU stays from the e-Intensive Care Unit Collaborative Research (eICU) and Medical Information Mart for Intensive Care IV (MIMIC-IV) databases, respectively. Using the learned AI policy, we estimated the hospital mortality rate (eICU 12.1%, SD 3.1%; MIMIC-IV 29.1%, SD 0.9%), the proportion of optimal oxygen saturation (eICU 58.7%, SD 4.7%; MIMIC-IV 49%, SD 1%), and the proportion of optimal mean arterial blood pressure (eICU 31.1%, SD 4.5%; MIMIC-IV 41.2%, SD 1%). Based on multiple quantitative and qualitative evaluation metrics, our proposed AI solution outperformed observed clinical practice.
Conclusions
Our study found that customizing ventilation settings for individual patients led to lower estimated hospital mortality rates compared to actual rates. This highlights the potential effectiveness of using reinforcement learning methodology to develop AI models that analyze complex clinical data for optimizing treatment parameters. Additionally, our findings suggest the integration of this model into a clinical decision support system for refining ventilation settings, supporting the need for prospective validation trials.
Subjects
MeSH terms (MEDLINE)
Author keywords
Authors and institutions
- National University of Singapore Graduate School for Integrative Science and Engineering, National University of Singapore, Singapore, Singapore.
- Saw Swee Hock School of Public Health, National University of Singapore, Singapore, Singapore.
- Ping An Healthcare Technology, Beijing, China.
- Institute of Data Science, National University of Singapore, Singapore, Singapore.
- Division of Respiratory and Critical Care Medicine, Department of Medicine, National University Hospital, Singapore, Singapore.
ORCID profiles: Siqi Liu, Qianyi Xu, Zhuoyang Xu, Zhuo Liu, Xingzhi Sun, Guotong Xie, Mengling Feng, Kay Choong See
Cite
APA
Liu, S., Xu, Q., Xu, Z., Liu, Z., Sun, X., Xie, G., Feng, M., & See, K. C. (2024). Reinforcement Learning to Optimize Ventilator Settings for Patients on Invasive Mechanical Ventilation: Retrospective Study. Journal of Medical Internet Research, 26, e44494. https://doi.org/10.2196/44494
Vancouver
Liu S, Xu Q, Xu Z, Liu Z, Sun X, Xie G, et al. Reinforcement Learning to Optimize Ventilator Settings for Patients on Invasive Mechanical Ventilation: Retrospective Study. Journal of Medical Internet Research. 2024;26:e44494. doi:10.2196/44494
BibTeX
@article{Liu2024,
author = {Liu, Siqi and Xu, Qianyi and Xu, Zhuoyang and Liu, Zhuo and Sun, Xingzhi and Xie, Guotong and Feng, Mengling and See, Kay Choong},
title = {Reinforcement Learning to Optimize Ventilator Settings for Patients on Invasive Mechanical Ventilation: Retrospective Study},
journal = {Journal of Medical Internet Research},
year = {2024},
volume = {26},
pages = {e44494},
doi = {10.2196/44494}
}
EUVIMED