Deep significance clustering: a novel approach for identifying risk-stratified and predictive patient subgroups.
Deep significance clustering: a novel approach for identifying risk-stratified and predictive patient subgroups.
复制标题
深层群集：一种新的方法，用于识别风险分层和预测性的患者亚组。
DOI：
10.1093/jamia/ocab203
复制
发表时间：
2021-11-25
期刊：
影响因子：
--
通讯作者：
Zhang Y
中科院分区：
文献类型：
--
作者：
Huang Y;Liu Y;Steel PAD;Axsom KM;Lee JR;Tummalapalli SL;Wang F;Pathak J;Subramanian L;Zhang Y
深度显著性聚类（DICE）是一种自监督学习框架。DICE识别临床相似和风险分层的亚组，非监督聚类算法和监督风险预测算法都不能单独保证生成这些亚组。DICE通过优化过程实现结果和子组成员之间的统计显著性，在为深度表示提供可解释性的同时，DICE联合训练3个组件：表示学习、聚类和结果预测。DICE还允许将未见过的患者预测为经过训练的亚组，以进行人群水平的风险分层。我们使用来自2家城市医院的电子健康记录数据集评估DICE。结果和患者队列分别包括心力衰竭（HF）患者和COVID-19 （Cov-AKI）患者的急性肾损伤出院。与包括主成分分析在内的基线方法相比，DICE在聚类纯度指标：Silhouette评分（HF为0.48，Cov-AKI为0.51）、Calinski-Harabasz指数（HF为212，Cov-AKI为254）、davis - bouldin指数（HF为0.86，Cov-AKI为0.66）和预测指标：受试者工作特征（ROC）曲线下面积（HF为0.83，Cov-AKI为0.78）方面表现出更优的性能。对dice生成的亚组的临床评估显示，亚组成员特征的分布更有意义，亚组之间的风险比更高。此外，仅由dice生成的亚组成员就能适度预测预后。DICE解决了当前机器学习方法中的一个空白，即预测的风险可能无法直接导致可操作的临床步骤。DICE证明了在异质人群中应用的潜力，在异质人群中，具有相同的定量风险并不等同于具有相似的临床概况。
Deep significance clustering (DICE) is a self-supervised learning framework. DICE identifies clinically similar and risk-stratified subgroups that neither unsupervised clustering algorithms nor supervised risk prediction algorithms alone are guaranteed to generate. Enabled by an optimization process that enforces statistical significance between the outcome and subgroup membership, DICE jointly trains 3 components, representation learning, clustering, and outcome prediction while providing interpretability to the deep representations. DICE also allows unseen patients to be predicted into trained subgroups for population-level risk stratification. We evaluated DICE using electronic health record datasets derived from 2 urban hospitals. Outcomes and patient cohorts used include discharge disposition to home among heart failure (HF) patients and acute kidney injury among COVID-19 (Cov-AKI) patients, respectively. Compared to baseline approaches including principal component analysis, DICE demonstrated superior performance in the cluster purity metrics: Silhouette score (0.48 for HF, 0.51 for Cov-AKI), Calinski-Harabasz index (212 for HF, 254 for Cov-AKI), and Davies-Bouldin index (0.86 for HF, 0.66 for Cov-AKI), and prediction metric: area under the Receiver operating characteristic (ROC) curve (0.83 for HF, 0.78 for Cov-AKI). Clinical evaluation of DICE-generated subgroups revealed more meaningful distributions of member characteristics across subgroups, and higher risk ratios between subgroups. Furthermore, DICE-generated subgroup membership alone was moderately predictive of outcomes. DICE addresses a gap in current machine learning approaches where predicted risk may not lead directly to actionable clinical steps. DICE demonstrated the potential to apply in heterogeneous populations, where having the same quantitative risk does not equate with having a similar clinical profile.
登录
查看更多内容
影响因子：
6.2
作者：
Lauck, Sandra B.;Wood, David A.;Webb, John G.
通讯作者：
Webb, John G.
影响因子：
5.7
作者：
Cubbon, Richard M.;Witte, Klaus K.;Kearney, Mark T.
通讯作者：
Kearney, Mark T.
影响因子：
13.6
作者：
Fisher, Molly;Neugarten, Joel;Golestaneh, Ladan
通讯作者：
Golestaneh, Ladan
影响因子：
15.2
作者：
Beaulieu-Jones BK;Yuan W;Brat GA;Beam AL;Weber G;Ruffin M;Kohane IS
通讯作者：
Kohane IS
DOI：
10.2215/cjn.09330819
发表时间：
2020-11-06
影响因子：
9.8
作者：
Chaudhary, Kumardeep;Vaid, Akhil;Nadkarni, Girish N.
通讯作者：
Nadkarni, Girish N.