Electrocardiography (ECG) is used to measure the electrical activity of the heart. Electrodes are placed on various parts of the human skin to measure the differences in electrical changes. By deriving these changes the condition of the heart can be assessed in order to identify if the patients heart is in a healthy our unhealthy state. Often this technique is used to identify if a patient has suffered from a heart attack.

The normal rhythm of the heart follows a very strict wave form, with only small variance. If an alternative wave form is identified, one can immediately tell that something is wrong. The identification of the rhythm type is indispensable for determining the next steps of the patients treatment. Although routinized cardiologists can identify the rhythm type precisely, it often takes a lot of time to identify if and what kind of abnormal rhythm the patient’s heart has. Having a classification algorithm which supports the physician in his work would save him time and will let him focus on the next steps more closely.

The data set includes only the electrical signals with a frequency of 300 Hz and the following labels:

1. Normal Rhythm 2. Artrial Fibrillation 3. Other Frequency 4. Noisy

Link to dataset: https://www.physionet.org/challenge/2017/