机器学习聚类算法在急性呼吸窘迫综合征治疗效果异质性检测中的比较：三项随机对照试验的二次分析。

Comparison of machine learning clustering algorithms for detecting heterogeneity of treatment effect in acute respiratory distress syndrome: A secondary analysis of three randomised controlled trials.

机构信息

Division of Clinical and Translational Research, Division of Critical Care, Department of Anesthesia, Washington University School of Medicine, Saint Louis, MO.

Department of Medicine, University of Wisconsin- Madison, Madison, Wisconsin.

出版信息

EBioMedicine. 2021 Dec;74:103697. doi: 10.1016/j.ebiom.2021.103697. Epub 2021 Dec 1.

DOI:10.1016/j.ebiom.2021.103697

PMID:34861492

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC8645454/

Abstract

BACKGROUND

Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected.

METHODS

Five unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable.

FINDINGS

No single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms.

INTERPRETATION

Machine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation.

FUNDING

NIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC).

摘要

背景

由于急性呼吸窘迫综合征（ARDS）的非特异性定义，其异质性导致了大量负面的随机对照试验（RCT）。研究人员试图使用聚类算法来确定治疗效果的异质性（HTE）。我们评估了几种常用的机器学习算法识别可能检测到 HTE 的聚类的能力。

方法

五种无监督算法：潜在类别分析（LCA）、K-均值、中位数分区、层次和谱聚类；以及四种监督算法：基于模型的递归分区、因果森林（CF）和带有随机森林的 X-learner（XL-RF）和贝叶斯加法回归树分别应用于三项先前的 ARDS RCT。临床数据和研究蛋白生物标志物被用作分区变量，后者在二次分析中被排除。对于聚类方案，根据治疗组和聚类与 90 天死亡率作为因变量的交互项评估 HTE。

结果

没有一种算法能够在所有三项试验中识别出具有显著 HTE 的聚类。LCA、XL-RF 和 CF 最常识别出 HTE（2/3 RCT）。无监督方法中的重要分区变量在算法和 RCT 之间是一致的。在监督模型中，重要的分区变量在算法之间和 RCT 之间有所不同。在同一试验中算法中聚类显示出 HTE 的算法中，患者在算法之间经常从治疗受益聚类切换到治疗危害聚类。除了 LCA 之外，所有其他算法的结果都受到聚类组成和 HTE 的显著改变，并且随着随机种子的改变而改变。去除研究生物标志物作为分区变量大大降低了所有算法检测 HTE 的机会。