Skip to content

Signals Overview

PhysioKit bundles end-to-end utilities for common wearable signals—ECG, PPG, respiratory (RSP), inertial (IMU), and derived HRV. Each module ships cleaning, peak detection, metrics, and synthetic data helpers built on a consistent API.

Modules at a Glance

  • ECG: Cleaning, R-peak detection, RR intervals, HR/respiration, segmentation, and synthetic ECG.
  • PPG: Peak detection, HR/respiration from peaks or FFT, SpO₂ estimation, and PPG synthesis.
  • RSP: Respiratory peak detection, rate estimation (FFT/peaks), and dual-band ribcage/abdomen metrics.
  • IMU: ENMO, tilt angles, and activity counts for accelerometer data.
  • HRV: Time and frequency-domain HRV metrics and supporting dataclasses.
  • Signal: Shared filters, resampling, smoothing, noise/distortion helpers, and FFT utilities.

Quickstart Examples

Compute heart rate from ECG peaks 
import numpy as np
import physiokit as pk

fs = 250
t = np.arange(0, 10, 1 / fs)
ecg = np.sin(2 * np.pi * 1.2 * t)  # toy signal

bpm, qos = pk.ecg.compute_heart_rate(ecg, sample_rate=fs, method="peak")
print(bpm, qos)
Find PPG peaks and estimate SpO₂ 
import numpy as np
import physiokit as pk

fs = 100
t = np.arange(0, 8, 1 / fs)
ppg = np.sin(2 * np.pi * 1.1 * t)

peaks = pk.ppg.find_peaks(ppg, sample_rate=fs)
spo2 = pk.ppg.compute_spo2_in_time(ppg, ppg, sample_rate=fs)
print(len(peaks), spo2)
Respiratory rate from RSP using FFT 
import numpy as np
import physiokit as pk

fs = 25
t = np.arange(0, 40, 1 / fs)
rsp = np.sin(2 * np.pi * 0.2 * t)

bpm, qos = pk.rsp.compute_respiratory_rate(rsp, sample_rate=fs, method="fft")
print(bpm, qos)
IMU ENMO and tilt 
import numpy as np
import physiokit as pk

x = np.array([0, 0.1, 0.2])
y = np.array([0, 0.1, 0.2])
z = np.array([1, 1.0, 1.1])

enmo = pk.imu.compute_enmo(x, y, z)
angles = pk.imu.compute_tilt_angles(x, y, z, in_radians=False)
print(enmo, angles)
Signal helpers: filter and FFT 
import numpy as np
import physiokit as pk

fs = 100
t = np.arange(0, 5, 1 / fs)
sig = np.sin(2 * np.pi * 2 * t) + 0.2 * np.random.randn(t.size)

clean = pk.signal.filter_signal(sig, lowcut=0.5, highcut=20, sample_rate=fs)
freqs, sp = pk.signal.compute_fft(clean, sample_rate=fs)
print(freqs[np.argmax(np.abs(sp))])