Al-Jefri Majed, Lee Joon, James Matthew
Annu Int Conf IEEE Eng Med Biol Soc. 2020 Jul;2020:5606-5609. doi: 10.1109/EMBC44109.2020.9175448.
Acute Kidney Injury (AKI) is a common complication after surgery. Recognition of patients at risk of AKI at an earlier stage is a priority for researchers and health care providers. The objective of this study is to develop machine learning prediction models of acute kidney injury (AKI) in patients who undergo surgery. The dataset used in this study consists of in-hospital patients' data of five different cohorts coming from different major procedure types. This data was collected from the SunRiseClinical Manager (SCM) electronic medical records system that is used in the Calgary Zone, Alberta, Canada from 2008 to 2015 where the patients are >=18 years of age. Five classifiers were experimented with: support vector machine, random forest, logistic regression, k-nearest neighbors, and adaptive boosting. The area under the receiver operating characteristics curve (AUROC) ranged between 0.62-0.84 and sensitivity and specificity ranged between 0.81-0.83 and 0.43-0.85, respectively. Predictions from these models can facilitate early intervention in AKI treatment.
急性肾损伤（AKI）是手术后常见的并发症。早期识别有AKI风险的患者是研究人员和医疗保健提供者的首要任务。本研究的目的是开发接受手术患者急性肾损伤（AKI）的机器学习预测模型。本研究中使用的数据集包括来自不同主要手术类型的五个不同队列的住院患者数据。这些数据是从2008年至2015年在加拿大艾伯塔省卡尔加里地区使用的SunRiseClinical Manager（SCM）电子病历系统中收集的，患者年龄≥18岁。对五个分类器进行了试验：支持向量机、随机森林、逻辑回归、k近邻和自适应增强。受试者工作特征曲线下面积（AUROC）在0.62 - 0.84之间，敏感性和特异性分别在0.81 - 0.83和0.43 - 0.85之间。这些模型的预测可以促进AKI治疗的早期干预。
