Torshizi Hamid Mokhtari, Omidi Negar, Khorgami Mohammad Rafie, Jamali Razieh, Ahmadi Mohsen
Department of Biomedical Engineering and Physics, School of Medicine, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Department of Cardiology, Tehran Heart Center, School of Medicine, Tehran University of Medical Sciences, Tehran, Iran.
Ann Pediatr Cardiol. 2024 Mar-Apr;17(2):116-123. doi: 10.4103/apc.apc_54_24. Epub 2024 Jul 20.
An abnormal variation in blood electrolytes, such as potassium, contributes to mortality in children admitted to intensive care units. Continuous and real-time monitoring of potassium serum levels can prevent fatal arrhythmias, but this is not currently practical. The study aims to use machine learning to estimate blood potassium levels with accuracy in real time noninvasively.
Hospitalized patients in the Pediatric Department of the Rajaie Cardiology and Medical Research Center and Tehran Heart Center were recruited from December 2021 to June 2022. The electrocardiographic (ECG) features of patients were evaluated. We defined 16 features for each signal and extracted them automatically. The dimension reduction operation was performed with the assistance of the correlation matrix. Linear regression, polynomials, decision trees, random forests, and support vector machine algorithms have been used to find the relationship between characteristics and serum potassium levels. Finally, we used a scatter plot and mean square error (MSE) to display the results.
Of 463 patients (mean age: 8 ± 1 year; 56% boys) hospitalized, 428 patients met the inclusion criteria, with 35 patients having a high noise of ECG were excluded. After the dimension reduction step, 11 features were selected from each cardiac signal. The random forest regression algorithm showed the best performance with an MSE of 0.3.
The accurate estimation of serum potassium levels based on ECG signals is possible using machine learning algorithms. This can be potentially useful in predicting serum potassium levels in specific clinical scenarios.
血液电解质（如钾）的异常变化会导致入住重症监护病房的儿童死亡。持续实时监测血钾水平可预防致命性心律失常，但目前这并不实际可行。本研究旨在利用机器学习实时、无创且准确地估算血钾水平。
2021年12月至2022年6月期间，招募了拉贾伊心脏病学与医学研究中心儿科以及德黑兰心脏中心的住院患者。对患者的心电图（ECG）特征进行评估。我们为每个信号定义了16个特征并自动提取。在相关矩阵的辅助下进行降维操作。使用线性回归、多项式、决策树、随机森林和支持向量机算法来寻找特征与血钾水平之间的关系。最后，我们使用散点图和均方误差（MSE）来展示结果。
在463名住院患者（平均年龄：8±1岁；56%为男孩）中，428名患者符合纳入标准，35名心电图噪声高的患者被排除。经过降维步骤后，从每个心脏信号中选择了11个特征。随机森林回归算法表现最佳，MSE为0.3。
使用机器学习算法基于心电图信号准确估算血钾水平是可行的。这在特定临床场景中预测血钾水平可能具有潜在用途。
