New onset delirium prediction using machine learning and long short-term memory (LSTM) in electronic health record.
New onset delirium prediction using machine learning and long short-term memory (LSTM) in electronic health record.
复制标题
DOI：
10.1093/jamia/ocac210
复制
发表时间：
2022-12-13
影响因子：
6.4
通讯作者：
Wright, Adam
中科院分区：
文献类型：
--
作者：
Liu, Siru;Schlesinger, Joseph J.;McCoy, Allison B.;Reese, Thomas J.;Steitz, Bryan;Russo, Elise;Koh, Brian;Wright, Adam
开发并测试一种用于预测住院成年患者新发谵妄的准确深度学习模型。
利用从一家大型学术医疗中心提取的电子健康记录（EHR）数据，我们开发了一种结合长短期记忆网络（LSTM）和机器学习的模型来预测新发谵妄，并将其性能与仅使用机器学习的模型（逻辑回归、随机森林、支持向量机、神经网络和LightGBM）进行比较。模型的标签是意识模糊评估法（CAM）评估结果。我们在一个预留数据集上对模型进行评估。我们计算了夏普利可加性解释（SHAP）指标来衡量特征对模型的影响。
共纳入了来自34035名患者的896个特征的331489次CAM评估。在机器学习模型中，LightGBM模型取得了最佳性能（曲线下面积AUC为0.927[0.924, 0.929]，F1值为0.626[0.618, 0.634]）。当与LSTM模型结合时，最终模型的性能显著提高（P = 0.001），AUC为0.952[0.950, 0.955]，F1值为0.759[0.755, 0.765]。在召回率固定为0.8的情况下，组合模型的准确率从0.497提高到0.751。利用平均绝对SHAP值，我们确定了前20个特征，包括年龄、心率、里士满激越 - 镇静量表评分、莫尔斯跌倒风险评分、脉搏、呼吸频率和护理级别。
利用LSTM捕捉时间趋势并将其与LightGBM模型相结合，可以显著提高对新发谵妄的预测能力，为后续开发用于积极谵妄干预的临床决策支持工具提供算法基础。
To develop and test an accurate deep learning model for predicting new onset delirium in hospitalized adult patients. Using electronic health record (EHR) data extracted from a large academic medical center, we developed a model combining long short-term memory (LSTM) and machine learning to predict new onset delirium and compared its performance with machine-learning-only models (logistic regression, random forest, support vector machine, neural network, and LightGBM). The labels of models were confusion assessment method (CAM) assessments. We evaluated models on a hold-out dataset. We calculated Shapley additive explanations (SHAP) measures to gauge the feature impact on the model. A total of 331 489 CAM assessments with 896 features from 34 035 patients were included. The LightGBM model achieved the best performance (AUC 0.927 [0.924, 0.929] and F1 0.626 [0.618, 0.634]) among the machine learning models. When combined with the LSTM model, the final model’s performance improved significantly (P = .001) with AUC 0.952 [0.950, 0.955] and F1 0.759 [0.755, 0.765]. The precision value of the combined model improved from 0.497 to 0.751 with a fixed recall of 0.8. Using the mean absolute SHAP values, we identified the top 20 features, including age, heart rate, Richmond Agitation-Sedation Scale score, Morse fall risk score, pulse, respiratory rate, and level of care. Leveraging LSTM to capture temporal trends and combining it with the LightGBM model can significantly improve the prediction of new onset delirium, providing an algorithmic basis for the subsequent development of clinical decision support tools for proactive delirium interventions.
登录
查看更多内容
影响因子：
168.9
作者：
Inouye, Sharon K.;Westendorp, Rudi G. J.;Saczynski, Jane S.
通讯作者：
Saczynski, Jane S.
影响因子：
5.7
作者：
Inouye, SK;Rushing, JT;Pompei, P
通讯作者：
Pompei, P
影响因子：
2.1
作者：
Bhattacharyya, Anirban;Sheikhalishahi, Seyedmostafa;Torbic, Heather;Yeung, Wesley;Wang, Tiffany;Birst, Jennifer;Duggal, Abhijit;Celi, Leo Anthony;Osmani, Venet
通讯作者：
Osmani, Venet
影响因子：
1.2
作者：
Adams, Carmen L.;Scruth, Elizabeth Ann;Cheng, Eugene
通讯作者：
Cheng, Eugene
DOI：
10.1111/j.1532-5415.1992.tb02111.x
发表时间：
1992-06-01
影响因子：
6.3
作者：
FRANCIS, J;KAPOOR, WN
通讯作者：
KAPOOR, WN