Detection of Cardiac Arrhythmia using Spectral features

Due to rapid advancements in technology, early diagnosis of diseases have facilitated in tremendous growth in the medical sector. One of the most crucial conditions that demands diagnosis at the earliest stage is Cardiac Arrhythmia. Arrhythmia is a condition where the heart beats too quickly or too slowly thus causing irregular rhythm. It affects millions of people every year. ECG signals can be used for the detection procedure. In this paper, we have proposed a method to detect arrhythmia from the spectral features which can be extracted from the ECG strips. For the data, we have used the MIT-BIH dataset as well the Stanford Irhythmic public data for the testing purpose. Further with a 14 attribute data frame created from the features extracted using python libraries, it was fed to various classification models out of which Random Forest gives the highest accuracy.