Scholarlink

<p>Emergency department (ED) crowding necessitates rapid, accurate risk stratification to optimize care and resource allocation. Traditional clinical prediction tools like NEWS, APACHE II, and SOFA score have limited generalizability or rely on extensive inputs, including vitals. We developed RISK<sup>INDEX</sup>, a machine-learning tool predicting 31-day mortality using routine laboratory values, age, and sex. To evaluate the clinical impact of a machine learning–based risk score in routine care, we conducted an investigator-initiated, open-label, randomized, non-inferiority trial (MARS-ED) at the Maastricht University Medical Center+ ED. Adult patients ( ≥ 18 years) assessed by an internal medicine specialist and with ≥4 laboratory tests were eligible. Patients were randomized 1:1 using computer-generated permuted blocks to standard care (<i>n</i> = 659) or standard care plus access to the RISK<sup>INDEX</sup> (<i>n</i> = 644). No blinding was possible because physicians needed to view the RISK<sup>INDEX</sup>. The primary outcomes for this study were the prognostic accuracy for 31-day mortality and clinical impact of the RISK<sup>INDEX</sup>. In total, 1303 participants were analyzed. RISK<sup>INDEX</sup>’s prognostic accuracy matched or outperformed clinical intuition (AUROC 0.84 vs. 0.73–0.76) and was statistically higher than NEWS, APACHE II, and SOFA prediction tools (AUROC 0.65–0.75). RISK<sup>INDEX</sup> predictions aligned with clinicians’ expectations in only about half of cases, with highest discordance among less experienced physicians. Despite its prognostic accuracy, the RISK<sup>INDEX</sup> did not alter treatment plans (1/644 changes; 0.16%) or clinical outcomes, and clinicians perceived low added value. No adverse events related to the intervention occurred, and recruitment was completed as planned. These findings show that prognostic accuracy alone is insufficient to achieve clinical impact in the ED and that user-centered, actionable model design is needed to ensure relevance, trust, and responsiveness. ClinicalTrials.gov registration: NCT05497830.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning for risk stratification in the emergency department (MARS-ED): a randomized controlled trial

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月1日

查看全文
  • Paul M. E. L. van Dam,
  • William P. T. M. van Doorn,
  • Lotte Sevenich,
  • Lars Lambriks,
  • Jochen W. L. Cals,
  • Otto Bekers,
  • Patricia M. Stassen,
  • Steven J. R. Meex

摘要

Emergency department (ED) crowding necessitates rapid, accurate risk stratification to optimize care and resource allocation. Traditional clinical prediction tools like NEWS, APACHE II, and SOFA score have limited generalizability or rely on extensive inputs, including vitals. We developed RISKINDEX, a machine-learning tool predicting 31-day mortality using routine laboratory values, age, and sex. To evaluate the clinical impact of a machine learning–based risk score in routine care, we conducted an investigator-initiated, open-label, randomized, non-inferiority trial (MARS-ED) at the Maastricht University Medical Center+ ED. Adult patients ( ≥ 18 years) assessed by an internal medicine specialist and with ≥4 laboratory tests were eligible. Patients were randomized 1:1 using computer-generated permuted blocks to standard care (n = 659) or standard care plus access to the RISKINDEX (n = 644). No blinding was possible because physicians needed to view the RISKINDEX. The primary outcomes for this study were the prognostic accuracy for 31-day mortality and clinical impact of the RISKINDEX. In total, 1303 participants were analyzed. RISKINDEX’s prognostic accuracy matched or outperformed clinical intuition (AUROC 0.84 vs. 0.73–0.76) and was statistically higher than NEWS, APACHE II, and SOFA prediction tools (AUROC 0.65–0.75). RISKINDEX predictions aligned with clinicians’ expectations in only about half of cases, with highest discordance among less experienced physicians. Despite its prognostic accuracy, the RISKINDEX did not alter treatment plans (1/644 changes; 0.16%) or clinical outcomes, and clinicians perceived low added value. No adverse events related to the intervention occurred, and recruitment was completed as planned. These findings show that prognostic accuracy alone is insufficient to achieve clinical impact in the ED and that user-centered, actionable model design is needed to ensure relevance, trust, and responsiveness. ClinicalTrials.gov registration: NCT05497830.