Using machine learning to model older adult inpatient trajectories from electronic health records data.
Using machine learning to model older adult inpatient trajectories from electronic health records data.
复制标题
DOI：
10.1016/j.isci.2022.105876
复制
发表时间：
2023-01-20
期刊：
影响因子：
5.8
通讯作者：
Keevil, Victoria L.
中科院分区：
文献类型：
--
作者：
Herrero-Zazo, Maria;Fitzgerald, Tomas;Taylor, Vince;Street, Helen;Chaudhry, Afzal N.;Bradley, John R.;Birney, Ewan;Keevil, Victoria L.
电子健康记录（EHR）数据可以为住院患者的轨迹提供新的见解。将来自去识别患者的住院事件（AE）的血液测试和生命体征表示为多变量时间序列（MVTS）以训练无监督隐马尔可夫模型（HMM），并将每个AE日表示为17个状态之一。所有HMM状态均基于其MVTS变量模式及其与临床信息的关系进行临床解释。可视化区分患者进展到稳定的“出院样”状态与那些仍然存在住院死亡率（IM）的风险。卡方检验证实了这些关系（2个状态与IM相关; 12个状态有≥1个诊断）。用MVTS数据而不是状态训练的逻辑回归和随机森林（RF）模型具有更高的IM预测性能，但结果相当（最佳RF模型AUC-ROC：MVTS数据= 0.85; HMM状态= 0.79）。ML模型从医院数据中提取临床可解释的信号。ML为EHR系统开发决策支持工具的潜力值得研究。将老年住院患者的时间序列血液测试和生命体征数据提交给HMM提取隐藏的临床可解释状态，与诊断和死亡联系起来状态建模住院患者轨迹，区分入院-出院风险HMM状态的临床解释有助于解释ML模型如何组织数据健康技术;健康技术中的诊断技术;医学科学中的应用计算;机器学习
Electronic Health Records (EHR) data can provide novel insights into inpatient trajectories. Blood tests and vital signs from de-identified patients’ hospital admission episodes (AE) were represented as multivariate time-series (MVTS) to train unsupervised Hidden Markov Models (HMM) and represent each AE day as one of 17 states. All HMM states were clinically interpreted based on their patterns of MVTS variables and relationships with clinical information. Visualization differentiated patients progressing toward stable ‘discharge-like’ states versus those remaining at risk of inpatient mortality (IM). Chi-square tests confirmed these relationships (two states associated with IM; 12 states with ≥1 diagnosis). Logistic Regression and Random Forest (RF) models trained with MVTS data rather than states had higher prediction performances of IM, but results were comparable (best RF model AUC-ROC: MVTS data = 0.85; HMM states = 0.79). ML models extracted clinically interpretable signals from hospital data. The potential of ML to develop decision-support tools for EHR systems warrants investigation. Time-series blood test & vital sign data from older inpatients were presented to HMM Hidden clinically interpretable states were extracted, linked with diagnoses and death States modeled inpatient trajectories, differentiating risk from admission-discharge The clinical interpretation of HMM states helped explain how ML models organize data Health technology; Diagnostic technique in health technology; Applied computing in medical science; Machine learning
登录
查看更多内容
影响因子：
4.1
作者：
Fox MT;Persaud M;Maimets I;Brooks D;O'Brien K;Tregunno D
通讯作者：
Tregunno D
影响因子：
6.7
作者：
Parsons, Matthew;Parsons, John;Kenealy, Tim
通讯作者：
Kenealy, Tim
影响因子：
5.8
作者：
Kowarik, Alexander;Templ, Matthias
通讯作者：
Templ, Matthias
影响因子：
15.2
作者：
Rajkomar, Alvin;Oren, Eyal;Dean, Jeffrey
通讯作者：
Dean, Jeffrey
影响因子：
5.8
作者：
Arandjelovic, Ognjen
通讯作者：
Arandjelovic, Ognjen