# PhysioNet/CinC Challenge 2017: AF Classification from a Short Single Lead ECG Recording

## Facts

- Provider: Gari D. Clifford, Chengyu Liu, Benjamin Moody, Li-wei Lehman, Ikaro Silva, Alistair Johnson, Roger Mark
- License: ODC-By-1.0 (https://physionet.org/content/challenge-2017/1.0.0/)
- Stored at: Amazon S3, s3://physionet-open/challenge-2017 (us-east-1)
- Access: public, readable without an account
- DOI: https://doi.org/10.13026/d3hm-sf11
- Category: Physiology
- Source: https://physionet.org/content/challenge-2017/1.0.0/
- Page: https://scigantic.com/archive/210af2a8-512b-5ffe-8e7b-ef310ae70005
- Keywords: Physiology, ECG, benchmark, AWS Open Data

## About

The 2017 PhysioNet/Computing in Cardiology Challenge asked for algorithms that classify short single-lead ECG recordings as normal sinus rhythm, atrial fibrillation (AF), other rhythm or too noisy to classify. The recordings were donated by AliveCor and collected with its handheld device as patient-initiated recordings of one minute or less. The training set contains 8,528 single-lead ECGs lasting from 9 s to just over 60 s and the hidden test set 3,658 recordings; signals were sampled at 300 Hz and band-pass filtered by the device. All data are in MATLAB V4 WFDB-compliant format, a .mat signal file and a .hea header per record. Measured 2026-11-13: training/ is split into subfolders A00, A01 and onward; labels are in REFERENCE-v0.csv to REFERENCE-v3.csv (successive relabelings), and training2017.zip, sample2017.zip, a Python dummy entry, results tables, the Clifford et al. CinC paper and sources/ sit at the top level. Read records with the wfdb Python package (wfdb.rdrecord) or scipy.io.loadmat and labels with pandas. A widely used benchmark for AF detection, arrhythmia classification and deep learning on wearable ECG. Released under the Open Data Commons Attribution License v1.0; cite the challenge paper and the standard PhysioNet reference (Goldberger et al., 2000) when using the data.

## Load the data

```python
import os
import pandas as pd
df = pd.read_csv(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], '1.0.0/REFERENCE-v2.csv'), nrows=1000)
df.head()
```

Run in a Scigantic notebook on 2026-11-14 against `1.0.0/REFERENCE-v2.csv`: CSV/TSV: columns ['A00001', 'N'], dtypes ['object', 'object'], first 1000 rows read.

From any machine:

```python
import boto3
from botocore import UNSIGNED
from botocore.client import Config
s3 = boto3.client('s3', region_name='us-east-1', config=Config(signature_version=UNSIGNED))
s3.list_objects_v2(Bucket='physionet-open', Prefix='challenge-2017/', Delimiter='/')
```

Lists the files with anonymous access. Runs on any machine, with no cloud account.

## Data structure

Example file: `challenge-2017/1.0.0/papers/065-469.pdf`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/210af2a8-512b-5ffe-8e7b-ef310ae70005
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/210af2a8-512b-5ffe-8e7b-ef310ae70005
- Over MCP: `get_schema_card` with id `210af2a8-512b-5ffe-8e7b-ef310ae70005` at https://api.scigantic.com/mcp

## Related topics

- [Benchmark datasets](https://scigantic.com/topics/benchmark)
- [Physiology datasets](https://scigantic.com/topics/physiology)
- [ECG datasets](https://scigantic.com/topics/ecg)
