Arrhythmia Detection: Dataset

In the next few concepts, we’ll be building an arrhythmia classifier. The data we will use comes from the Computing in Cardiology (CinC) Challenge 2017 dataset hosted on Physionet.

The dataset contains thousands of short ECG snippets (30s - 60s) from the AliveCor mobile ECG monitor. The original challenge was to build a 4-class classifier for sinus rhythm, atrial fibrillation, alternative rhythm, and noisy record. We will throw out the noisy records and build a two-class classifier distinguishing between sinus rhythm and another rhythm (atrial fibrillation included).

From exploring our data, we see that we have 1.5x more sinus rhythm records than other rhythm records. Most of the records are 30 seconds long; some are 60 seconds long, and a few are somewhere in between.

We plot the data and visualize the QRS complex detections provided in the dataset. The QRS complex detector still detects QRS complexes during periods of high noise, but these detections are suspect.

Download the notebook and play with the data for yourself!

Notebook Review

If you wanted to interact with the notebook in the video, you can access it here in the repo /ecg-processing/walkthroughs/af-data-exploration/ or in the workspace below.