NM000107: emg dataset, 100 subjects#

FRL Wrist Control: Wrist Movement Decoding from Surface Electromyography

Access recordings and metadata through EEGDash.

Citation: Patrick Kaifosh, Thomas R. Reardon, CTRL-labs at Reality Labs (2025). FRL Wrist Control: Wrist Movement Decoding from Surface Electromyography. 10.82901/nemar.nm000107

Modality: emg Subjects: 100 Recordings: 182 License: CC-BY-NC 4.0 Source: nemar

Metadata: Complete (100%)

Quickstart#

Get Started

Install

pip install eegdash

Access the data

from eegdash.dataset import NM000107

dataset = NM000107(cache_dir="./data")
# Get the raw object of the first recording
raw = dataset.datasets[0].raw
print(raw.info)

Query & Filter

Filter by subject

dataset = NM000107(cache_dir="./data", subject="01")

Advanced query

dataset = NM000107(
    cache_dir="./data",
    query={"subject": {"$in": ["01", "02"]}},
)

Iterate recordings

for rec in dataset:
    print(rec.subject, rec.raw.info['sfreq'])

Cite This Dataset

If you use this dataset in your research, please cite the original authors.

BibTeX

@dataset{nm000107,
  title = {FRL Wrist Control: Wrist Movement Decoding from Surface Electromyography},
  author = {Patrick Kaifosh and Thomas R. Reardon and CTRL-labs at Reality Labs},
  doi = {10.82901/nemar.nm000107},
  url = {https://doi.org/10.82901/nemar.nm000107},
}

About This Dataset#

wrist: Wrist Movement Control from EMG

Overview

Dataset: wrist - Wrist posture and movement from wrist-based surface electromyography Task: 1D continuous cursor control via wrist flexion/extension Participants: 100 subjects Sessions: 100 total (1 per subject)

View full README

wrist: Wrist Movement Control from EMG

Overview

Dataset: wrist - Wrist posture and movement from wrist-based surface electromyography Task: 1D continuous cursor control via wrist flexion/extension Participants: 100 subjects Sessions: 100 total (1 per subject) Publication: Kaifosh et al., 2025 - “A generic non-invasive neuromotor interface for human-computer interaction” (Nature)

Purpose

This dataset captures wrist-based sEMG signals during wrist movements for continuous cursor control. Motion capture provides ground-truth wrist angles. The goal is to enable gesture-free control through wrist posture alone, demonstrating sEMG’s ability to decode motor intent before visible movement occurs.

Dataset Details

Participants

Sample size: 100 participants
Demographics: Not available (marked as n/a)
Recording side: Dominant wrist
Sessions: 1 per participant

Hardware

Device: sEMG-RD (single wristband)
Channels: 16 (EMG0-EMG15)
Sampling rate: 2000 Hz
Reference: Bipolar differential
Ground truth: Motion capture wrist angles

Recording Protocol

Participant wears sEMG-RD on dominant wrist
Motion capture tracks wrist angles in real-time
Participant controls horizontal cursor position with wrist flexion/extension
Target acquisition task: Navigate to targets and hold for 500ms

Data Contents

Files per Session

sub-XXX/ses-XXX/emg/

├── sub-XXX_ses-XXX_task-wrist_emg.edf
├── sub-XXX_ses-XXX_task-wrist_emg.json
├── sub-XXX_ses-XXX_task-wrist_channels.tsv
├── sub-XXX_ses-XXX_task-wrist_events.tsv
└── sub-XXX_ses-XXX_electrodes.tsv

Events

Stage boundaries: Task phases and movement trials

Coordinate System

Single coordinate system at root (dominant wrist, percent units, no decimals)

Signal Processing

Note: This dataset has significant data quality issues: - Duplicate timestamps found in many sessions (up to 88% duplicates) - Irregular sampling requiring resampling (up to 916% deviation) - Post-processing: Duplicate removal followed by resampling to regular 2000 Hz

Baseline Performance

Published Results (Kaifosh et al., 2025)

Offline Evaluation: - Wrist angle velocity error: <13°/s - Error decreases with more training participants

Closed-loop Performance (n=17 naive test users): - Target acquisition time: Median 1.51s (sEMG decoder) - Dial-in time: Time to re-acquire after premature exit - Learning effects: Improvement from practice to evaluation blocks

Comparison: - Motion capture ground truth: 0.96s - MacBook trackpad: 0.68s - sEMG decoder: 1.51s (2.2× slower than trackpad)

Model architecture: MPF features + LSTM

Key Findings

Predictive signals: sEMG precedes movement by tens of milliseconds
Generic models work: Out-of-the-box cross-user generalization
Continuous control: Demonstrates feasibility of gesture-free interfaces
Room for improvement: Performance gap vs traditional inputs

Use Cases

Continuous control: Cursor/pointer movement
AR/VR navigation: Hands-free interface
Low-effort control: Minimal visible movement required
Predictive decoding: Intent detection before motion completion

Known Limitations

Single degree of freedom (1D control only)
Single wrist (dominant hand)
Duplicate timestamps (data quality issue)
Performance below traditional inputs
Extension to 2D control not demonstrated

Citation

Kaifosh, P., Reardon, T.R., & CTRL-labs at Reality Labs. (2025).
A generic non-invasive neuromotor interface for human-computer interaction.
Nature, 645(8081), 702-711. https://doi.org/10.1038/s41586-025-09255-w

Data Curator

Yahya Shirazi SCCN (Swartz Center for Computational Neuroscience) INC (Institute for Neural Computation) University of California San Diego

Version History

v1.0 (2025-10-01): Initial BIDS conversion

BIDS Version: 1.11 | EMG-BIDS: BEP-042 | Updated: Oct 1, 2025

Dataset Information#

Dataset ID	`NM000107`
Title	FRL Wrist Control: Wrist Movement Decoding from Surface Electromyography
Author (year)	`Kaifosh2025_107`
Canonical	—
Importable as	`NM000107`, `Kaifosh2025_107`
Year	2025
Authors	Patrick Kaifosh, Thomas R. Reardon, CTRL-labs at Reality Labs
License	CC-BY-NC 4.0
Citation / DOI	10.82901/nemar.nm000107
Source links	OpenNeuro \| NeMAR \| Source URL

Found an issue with this dataset?

If you encounter any problems with this dataset (missing files, incorrect metadata, loading errors, etc.), please let us know!

Report an Issue on GitHub

Technical Details#

Subjects & recordings

Subjects: 100
Recordings: 182
Tasks: 1

Channels & sampling rate

Channels: 16
Sampling rate (Hz): 2000.0
Duration (hours): 77.18355888888888

Electrode Layout#

Electrode layout — EMG · 16 sensors — 16 channels

Dataset Statistics#

Channel counts: 16 ch (n=182 recordings)

Sampling frequencies: 2000.0 Hz (n=182 recordings)

Total recording duration: 77 h

NEMAR Processing Statistics#

The plots below are generated by NEMAR’s automated EEG pipeline. The histogram shows pipeline success for data cleaning and ICA decomposition, the percentage of data frames and EEG channels retained after artefact removal, line noise per channel (RMS, dB), and the age/gender distribution of participants.

HED event descriptors word cloud

File Explorer#

Browse the BIDS file structure of this dataset. Records are fetched on demand from the EEGDash catalog the first time you open the explorer.

Files: —

Size: —

Subjects: —

Click to load file structure…

API Reference#

Use the NM000107 class to access this dataset programmatically.

class eegdash.dataset.NM000107(cache_dir: str, query: dict | None = None, s3_bucket: str | None = None, **kwargs)[source]#

Bases: EEGDashDataset

FRL Wrist Control: Wrist Movement Decoding from Surface Electromyography

Study:: nm000107 (NeMAR)
Author (year):: Kaifosh2025_107
Canonical:: —

Also importable as: NM000107, Kaifosh2025_107.

Modality: emg. Subjects: 100; recordings: 182; tasks: 1.

Parameters:

cache_dir (str | Path) – Directory where data are cached locally.
query (dict | None) – Additional MongoDB-style filters to AND with the dataset selection. Must not contain the key dataset.
s3_bucket (str | None) – Base S3 bucket used to locate the data.
**kwargs (dict) – Additional keyword arguments forwarded to EEGDashDataset.

data_dir#

Local dataset cache directory (cache_dir / dataset_id).

Type:: Path

query#

Merged query with the dataset filter applied.

Type:: dict

records#

Metadata records used to build the dataset, if pre-fetched.

Type:: list[dict] | None

Notes

Each item is a recording; recording-level metadata are available via dataset.description. query supports MongoDB-style filters on fields in ALLOWED_QUERY_FIELDS and is combined with the dataset filter. Dataset-specific caveats are not provided in the summary metadata.

References

OpenNeuro dataset: https://openneuro.org/datasets/nm000107 NeMAR dataset: https://nemar.org/dataexplorer/detail?dataset_id=nm000107 DOI: https://doi.org/10.82901/nemar.nm000107

Examples

>>> from eegdash.dataset import NM000107
>>> dataset = NM000107(cache_dir="./data")
>>> recording = dataset[0]
>>> raw = recording.load()

__init__(cache_dir: str, query: dict | None = None, s3_bucket: str | None = None, **kwargs)[source]#

save(path: str, overwrite: bool = False, offset: int = 0)[source]#

Save datasets to files by creating one subdirectory for each dataset:

path/
    0/
        0-raw.fif | 0-epo.fif
        description.json
        raw_preproc_kwargs.json (if raws were preprocessed)
        window_kwargs.json (if this is a windowed dataset)
        window_preproc_kwargs.json  (if windows were preprocessed)
        target_name.json (if target_name is not None and dataset is raw)
    1/
        1-raw.fif | 1-epo.fif
        description.json
        raw_preproc_kwargs.json (if raws were preprocessed)
        window_kwargs.json (if this is a windowed dataset)
        window_preproc_kwargs.json  (if windows were preprocessed)
        target_name.json (if target_name is not None and dataset is raw)