Department of Anesthesiology, Washington University School of Medicine, St. Louis, MO, USA.
Department of Electrical and Computer Engineering, The University of Texas at Austin, Austin, TX, USA.
Br J Anaesth. 2024 Dec;133(6):1159-1172. doi: 10.1016/j.bja.2024.08.007. Epub 2024 Sep 24.
We lack evidence on the cumulative effectiveness of machine learning (ML)-driven interventions in perioperative settings. Therefore, we conducted a systematic review to appraise the evidence on the impact of ML-driven interventions on perioperative outcomes.
Ovid MEDLINE, CINAHL, Embase, Scopus, PubMed, and ClinicalTrials.gov were searched to identify randomised controlled trials (RCTs) evaluating the effectiveness of ML-driven interventions in surgical inpatient populations. The review was registered with PROSPERO (CRD42023433163) and conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Meta-analysis was conducted for outcomes with two or more studies using a random-effects model, and vote counting was conducted for other outcomes.
Among 13 included RCTs, three types of ML-driven interventions were evaluated: Hypotension Prediction Index (HPI) (n=5), Nociception Level Index (NoL) (n=7), and a scheduling system (n=1). Compared with the standard care, HPI led to a significant decrease in absolute hypotension (n=421, P=0.003, I=75%) and relative hypotension (n=208, P<0.0001, I=0%); NoL led to significantly lower mean pain scores in the post-anaesthesia care unit (PACU) (n=191, P=0.004, I=19%). NoL showed no significant impact on intraoperative opioid consumption (n=339, P=0.31, I=92%) or PACU opioid consumption (n=339, P=0.11, I=0%). No significant difference in hospital length of stay (n=361, P=0.81, I=0%) and PACU stay (n=267, P=0.44, I=0) was found between HPI and NoL.
HPI decreased the duration of intraoperative hypotension, and NoL decreased postoperative pain scores, but no significant impact on other clinical outcomes was found. We highlight the need to address both methodological and clinical practice gaps to ensure the successful future implementation of ML-driven interventions.
CRD42023433163 (PROSPERO).
我们缺乏关于机器学习 (ML) 驱动干预措施在围手术期效果的累积证据。因此，我们进行了一项系统评价，以评估 ML 驱动干预措施对围手术期结局的影响。
我们在 Ovid MEDLINE、CINAHL、Embase、Scopus、PubMed 和 ClinicalTrials.gov 中搜索了评估 ML 驱动干预措施在外科住院人群中有效性的随机对照试验 (RCT)。该综述在 PROSPERO（CRD42023433163）中进行了注册，并按照系统评价和荟萃分析的首选报告项目 (PRISMA) 指南进行了操作。对于有两个或更多研究的结局，使用随机效应模型进行了荟萃分析，对于其他结局，进行了投票计数。
在纳入的 13 项 RCT 中，评估了三种类型的 ML 驱动干预措施：低血压预测指数 (HPI)（n=5）、疼痛感知水平指数 (NoL)（n=7）和调度系统（n=1）。与标准护理相比，HPI 显著降低了绝对低血压的发生率（n=421，P=0.003，I=75%）和相对低血压的发生率（n=208，P<0.0001，I=0%）；NoL 显著降低了术后麻醉护理单元 (PACU) 的平均疼痛评分（n=191，P=0.004，I=19%）。NoL 对术中阿片类药物用量（n=339，P=0.31，I=92%）或 PACU 阿片类药物用量（n=339，P=0.11，I=0%）无显著影响。HPI 和 NoL 之间在住院时间（n=361，P=0.81，I=0%）和 PACU 停留时间（n=267，P=0.44，I=0%）方面无显著差异。
HPI 降低了术中低血压的持续时间，NoL 降低了术后疼痛评分，但对其他临床结局没有显著影响。我们强调需要解决方法学和临床实践方面的差距，以确保 ML 驱动干预措施的成功实施。
CRD42023433163（PROSPERO）。
