bundles / scipy 2.0.0.dev0+git20260907.41eeb59 / scipy / datasets / _fetchers / electrocardiogram

function

`scipy.datasets._fetchers:electrocardiogram`

source: build-install/usr/lib/python3.14/site-packages/scipy/datasets/_fetchers.py:83

Signature

def electrocardiogram()

Summary

Load an electrocardiogram as an example for a 1-D signal.

Extended Summary

The returned signal is a 5 minute long electrocardiogram (ECG), a medical recording of the heart's electrical activity, sampled at 360 Hz.

Returns

ecg : ndarray: The electrocardiogram in millivolt (mV) sampled at 360 Hz.

Notes

The provided signal is an excerpt (19:35 to 24:35) from the record 208 (lead MLII) provided by the MIT-BIH Arrhythmia Database ^[1] on PhysioNet ^[2]. The excerpt includes noise induced artifacts, typical heartbeats as well as pathological changes.

Array API Standard Support

electrocardiogram is not in-scope for support of Python Array API Standard compatible backends other than NumPy.

See dev-arrayapi for more information.

Examples

from scipy.datasets import electrocardiogram
ecg = electrocardiogram()

✓

ecg
ecg.shape, ecg.mean(), ecg.std()

✗

((108000,), np.float64(-0.16510875), np.float64(0.5992473991177295))

As stated the signal features several areas with a different morphology. E.g., the first few seconds show the electrical activity of a heart in normal sinus rhythm as seen below.

import numpy as np
import matplotlib.pyplot as plt
fs = 360
time = np.arange(ecg.size) / fs

✓

plt.plot(time, ecg)
plt.xlabel("time in s")
plt.ylabel("ECG in mV")
plt.xlim(9, 10.2)
plt.ylim(-1, 1.5)

✗

(-1.0, 1.5)

plt.show()

✓

After second 16, however, the first premature ventricular contractions, also called extrasystoles, appear. These have a different morphology compared to typical heartbeats. The difference can easily be observed in the following plot.

plt.plot(time, ecg)
plt.xlabel("time in s")
plt.ylabel("ECG in mV")
plt.xlim(46.5, 50)
plt.ylim(-2, 1.5)

✗

(-2.0, 1.5)

plt.show()

✓

At several points large artifacts disturb the recording, e.g.:

plt.plot(time, ecg)
plt.xlabel("time in s")
plt.ylabel("ECG in mV")
plt.xlim(207, 215)
plt.ylim(-2, 3.5)

✗

(-2.0, 3.5)

plt.show()

✓

Finally, examining the power spectrum reveals that most of the biosignal is made up of lower frequencies. At 60 Hz the noise induced by the mains electricity can be clearly observed.

from scipy.signal import welch
f, Pxx = welch(ecg, fs=fs, nperseg=2048, scaling="spectrum")

✓

plt.semilogy(f, Pxx)
plt.xlabel("Frequency in Hz")
plt.ylabel("Power spectrum of the ECG in mV**2")
plt.xlim(f[[0, -1]])

✗

(0.0, 180.0)

plt.show()

✓