Skip to main content

med.signal - Biosignal Processing

Peak detection, heart rate analysis, HRV metrics, rhythm classification, SpO2 calculation, respiratory rate estimation, and waveform analysis for medical device integration.

Heart Rate & Rhythm​

heart_rate_from_rr(rr_intervals_ms) → dict​

Calculate heart rate from R-R intervals (milliseconds). HR = 60000 / mean_RR.

Returns classification: BRADYCARDIA, NORMAL_SINUS, TACHYCARDIA, or SEVERE_TACHYCARDIA.

lib.signal.heart_rate_from_rr([800, 810, 790, 800, 820])
# {'mean_hr_bpm': 74.5, 'classification': 'NORMAL_SINUS'}

classify_rhythm(rr_intervals_ms) → dict​

Basic cardiac rhythm classification from R-R interval regularity:

RhythmCriteria
Normal SinusHR 60-100, regular
Sinus BradycardiaHR under 60, regular
Sinus TachycardiaHR 100-150, regular
AF SuspectedIrregular with CV over 25%
SVT PossibleHR over 150
lib.signal.classify_rhythm([800, 810, 790, 800, 820])
# {'rhythm': 'NORMAL_SINUS_RHYTHM', 'heart_rate': 74.5, 'regularity': 'REGULAR'}

hrv_metrics(rr_intervals_ms) → dict​

Heart Rate Variability time-domain analysis:

MetricDescriptionClinical Significance
SDNNSD of NN intervalsOverall HRV; under 50ms = high cardiac risk
RMSSDRoot mean square of successive diffsParasympathetic activity
pNN50% of successive diffs over 50msVagal tone marker
lib.signal.hrv_metrics([800, 810, 790, 800, 820, 830, 780])
# {'sdnn_ms': 17.1, 'rmssd_ms': 17.3, 'pnn50_pct': 0.0,
#  'autonomic_status': 'LOW_HRV'}

Pulse Oximetry​

spo2_from_ratio(red_ac, red_dc, ir_ac, ir_dc) → dict​

Calculate SpO2 from pulse oximeter raw data using Beer-Lambert approximation:

R = (Red_AC / Red_DC) / (IR_AC / IR_DC)
SpO2 = 110 - 25 * R

Returns status: NORMAL (≥95%), MILD_HYPOXEMIA, MODERATE_HYPOXEMIA, SEVERE_HYPOXEMIA.

perfusion_index(ac_component, dc_component) → dict​

Perfusion Index: PI = (AC / DC) * 100. Normal over 1.0%, low perfusion under 0.5%.

Waveform Analysis​

detect_peaks(waveform, threshold=0.5) → dict​

Threshold-based peak detection with refractory period. Works for ECG QRS complexes, pulse waveforms, etc.

lib.signal.detect_peaks([0, 0.2, 0.8, 1.5, 0.9, 0.1, 0, 0.3, 0.7, 1.4, 0.8, 0.2])
# {'peaks_detected': 2, 'peaks': [{'index': 3, 'amplitude': 1.5}, ...]}

moving_average(data, window=5) → dict​

Simple moving average filter for signal smoothing and noise removal.

detect_anomaly(data, baseline_mean=None, threshold_sd=2.0) → dict​

Statistical anomaly detection. Flags data points exceeding threshold_sd standard deviations from baseline.

Respiratory​

respiratory_rate(waveform, sampling_rate_hz=25.0) → dict​

Estimate respiratory rate from impedance or chest waveform using zero-crossing method.

Returns status: BRADYPNEA (under 12), NORMAL (12-20), TACHYPNEA (20-30), SEVERE_TACHYPNEA (over 30).

Integration with med.io​

med.signal pairs naturally with med.io for real-time device monitoring:

protocol WaveformMonitor {
    input: Patient p;

    // Capture ECG waveform from bedside monitor
    let ecg = med.io.read_waveform("Monitor_01", "ECG_II", 5.0, 250.0);

    // Analyze
    let peaks = med.signal.detect_peaks(ecg.data);
    let rhythm = med.signal.classify_rhythm(peaks.inter_peak_intervals);

    if rhythm.rhythm == "ATRIAL_FIBRILLATION_SUSPECTED" {
        alert "Possible AF detected" severity: critical;
    }
}

See Also​

  • med.io — device interfaces for real-time waveform capture
  • med.scores — clinical scoring from vital sign parameters
  • med.lab — correlate biosignal findings with lab results