riteshhf's picture
Upload folder using huggingface_hub
74286a5 verified
|
Raw
History Blame Contribute Delete
2 kB
metadata
license: cc0-1.0
tags:
  - eeg
  - medical-time-series
  - icml2026
  - open-reproductions
task_categories:
  - time-series-forecasting

ADFTD (OpenNeuro ds004504) — processed for the TS-Fingerprint reproduction

Processed arrays for the ICML 2026 reproduction of "Learning Fingerprints for Medical Time Series with Redundancy-Constrained Information Maximization" (OpenReview lrRBHFIgaK, arXiv 2605.00130).

Provenance

Source: OpenNeuro ds004504A dataset of EEG recordings from Alzheimer's disease, Frontotemporal dementia and Healthy subjects (Miltiadous et al., 2023), CC0. We use the derivatives/ (ASR + ICA cleaned) .set recordings, 19 channels at 500 Hz, 88 subjects.

Processing

Reproduces data_preprocessing/ADFTD_preprocessing.ipynb of DL4mHealth/Medformer (the protocol the paper cites), see prep_adftd.py:

  1. linear-interpolation resample 500 Hz → 256 Hz
  2. non-overlapping 256-timestamp (1 s) windows, centre-anchored on each recording
  3. labels {C: 0 healthy, F: 1 FTD, A: 2 AD}
  4. subject-wise 60/20/20 split within each class (Medformer ADFTDLoader, a=0.6, b=0.8)
  5. per-sample, per-channel standardisation (Medformer uea.normalize_batch_ts)

Contents

File Shape Notes
X_train.npy (40446, 256, 19) 51 subjects
X_val.npy (14658, 256, 19) 18 subjects
X_test.npy (14648, 256, 19) 19 subjects
y_*.npy (N,) int64 0 = CN, 1 = FTD, 2 = AD
subj_*.npy (N,) int64 subject id per window, for leakage checks

Total 69,752 samples — this exactly matches the paper's Table 5 (#Sample = 69,752), which is the main evidence that the preprocessing pipeline is faithful.

Please cite the original data descriptor: doi:10.3390/data8060095