ECG Signal Classification for Detection of Hyperkalemia

Abstract

Potassium Imbalance is a serious problem that is the leading cause of sudden cardiac deaths in chronic kidney patients (CKD). Currently blood test is the gold standard for detection of potassium imbalances. Electrocardiogram (ECG) signals provide a non- invasive way to view the cardiac activity of the heart. It can also be used to detect potassium imbalance in Chronic Kidney patients. However the unique characteristics and complexity of ECG signals make it a very challenging process. This paper presents a machine learning classifier for detection of hyperkalaemia in patients using features extracted from ECG. Feature extraction plays a very crucial role in the machine learning process as it helps to capture the essential information for the learning task which can be used for applications such as classification and detection. Ten statistical features were extracted from ECG signals of patients having potassium within normal range and those with elevated levels of potassium. Classification performance of four different classifiers namely Naïve Bayes Classifier, Support Vector Machine(SVM), K Nearest Neighbors(kNN) and Artificial Neural Networks(ANN) was compared using statistical features. kNN and ANN performed best with classification accuracy of 97.9%.The results we found are in-line with other state-of-art hyperkalaemia classification approaches