Sample data

Sample recordings support manual testing, demos, and Phase 1 acceptance checks.

Generate synthetic ECG CSV

conda activate biochem
python products/cardiology/experiments/example01/generate_ecg_csv.py

Default output: products/cardiology/experiments/example01/ecg_360hz_3min.csv (360 Hz, 3 minutes, NeuroKit2-simulated ECG).

Options:

python products/cardiology/experiments/example01/generate_ecg_csv.py --duration 600 --output ecg_10min.csv

MIT-BIH WFDB record 100

Recommended PhysioNet fixture for WFDB upload testing:

1. Download from MIT-BIH Arrhythmia Database
2. Get 100.hea and 100.dat
3. Upload both together in the app

Or fetch programmatically:

import wfdb

record = wfdb.rdrecord("100", pn_dir="mitdb")
wfdb.wrsamp(
    "100",
    fs=record.fs,
    units=record.units,
    sig_name=record.sig_name,
    p_signal=record.p_signal,
    fmt=record.fmt,
    write_dir="./products/cardiology/experiments/example01/wfdb",
)

Multimodal examples

The products/cardiology/experiments/ folder contains generators and usage notes for RR, BP, PPG, and multimodal cohort scenarios (example03–example06). See each folder's usage_notes.md.

Test fixtures

Automated tests use small fixtures under:

backend-analysis/tests/fixtures/

Prefer MIT-BIH record 100 or generated CSV for regression tests. Large recordings stay gitignored under data/.

Suggested demo workflow

1. Create study
2. Upload generated CSV (quick test)
3. Upload MIT-BIH 100 WFDB (format validation)
4. Preprocess + HRV on each
5. Import subjects CSV and link datasets for cohort demo

Full details: products/cardiology/experiments/README.md and products/cardiology/physionet.md.