JMIR Publications

ABSTRACT

Background:

Monitoring hyperkalemia is very important in CKD patients, but so far blood testing is the only way to test serum potassium levels, so more closely and reliably monitoring requires development and verification of non-invasive and continuous monitoring methods.

Objective:

This study aimed to propose a novel method for noninvasive screening tool of hyperkalemia using a single-lead electrocardiogram (ECG) based on a deep learning model.

Methods:

For this study, 2,958 patients with hyperkalemia events from July 2009 to June 2019 were enrolled at one regional emergency center, and of which 1,790 patients were diagnosed with chronic renal failure (CRF) before hyperkalemic events. The patients that we cannot extract with the available form to analyze from the original 12-lead ECG signal were excluded, then we used data of the 855 (555 patients with CRF, 300 patients without CRF) patients. 12-lead ECG signal was collected at the time of hyperkalemic event, previous normal time, and post-event normal time in each of the patients. All 12-lead ECG signal have matched with an electrolyte test within 2 hours on each ECG to form a dataset. Then, we analyzed ECG signal with a duration of 2 s and a segment composed of 1,400 samples. The dataset randomly was divided into the training set, validation set, and test set according to the ratio of 6:2:2 percent. The proposed noninvasive screening tool is conducted with the deep learning model that can express the complex and cyclic rhythm of cardiac activity. The deep learning model consists of the convolutional and pooling layers for noninvasive screening to the serum potassium (K+) level from ECG signal. To extract an optimal single-lead ECG, we evaluated the performances of the proposed deep learning model for each leads including lead I, II, and V1 to V6 leads.

Results:

The proposed noninvasive screening tool using a single-lead ECG shows high performances with F1-score of 100%, 96%, and 95% for the training set, validation set, and test set, respectively. The lead II signal was shown the highest performance among other ECG leads.

Conclusions:

We developed a novel method for noninvasive screening of hyperkalemia using a single-lead ECG signal and it can be used as a helpful tool in emergency medicine.