PhysioMotion Artifact: A task-driven EEG dataset with point-wise motion artifact annotations

2026-04-09 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Health & Medical Research, Mathematics & Computational Sciences · Depth: Expert, medium

Summary

The PhysioMotion Artifact dataset, published on April 9, 2026, addresses persistent challenges in electroencephalogram (EEG) data analysis by providing a large-scale, task-driven collection with point-wise artifact annotations. This dataset was acquired from 30 healthy participants performing 16 systematically designed single-type and multi-type movement tasks, which induced 14 distinct types of physiological artifacts. To demonstrate its utility, a Convolutional Neural Networks-Transformer hybrid model was implemented for artifact detection and classification, achieving 95.4% accuracy in binary classification and 79.7% in 14-class classification tasks. The dataset is publicly available on OpenNeuro (https://doi.org/10.18112/openneuro.ds006386.v1.0.1), and all associated code, including modules for data preprocessing, annotation, and model training, is on GitHub (https://github.com/JiangweiYu221/PhysioMotion_Artifact).

Key takeaway

For machine learning engineers developing EEG signal processing solutions, this dataset offers a critical resource for improving artifact detection. Your models can achieve higher accuracy by training on PhysioMotion Artifact's 14 distinct, point-wise annotated physiological artifact types. Consider integrating the provided CNN-Transformer hybrid model as a baseline or starting point for your own artifact classification systems.

Key insights

PhysioMotion Artifact is a new EEG dataset with detailed point-wise annotations for 14 types of physiological motion artifacts.

Principles

• Systematic task design induces diverse artifact types.
• Point-wise annotations enhance artifact detection accuracy.

Method

EEG data from 30 participants performing 16 movement tasks were collected, annotated point-wise for 14 artifact types, and used to train a CNN-Transformer hybrid model.

In practice

• Utilize the dataset for training robust EEG artifact detectors.
• Adapt the CNN-Transformer model for specific artifact classification.
• Explore the annotation interface for custom labeling workflows.

Topics

• EEG Dataset
• Physiological Artifacts
• Motion Artifacts
• Point-wise Annotation
• CNN-Transformer Model

Code references

• JiangweiYu221/PhysioMotion_Artifact

Best for: AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• Reducing the Complexity of Deep Learning Models for EEG Analysis on Wearable Devices — Parameter quantization and electrode reduction can significantly reduce DNN complexity for wearable EEG analysis with minimal accuracy loss.
• Neural Response to Familiar Names Predicts Outcome of Comatose ICU Patients: A Prospective Observational Cohort Study — EEG responses to familiar names predict outcomes in comatose ICU patients.
• MOSAIC: Modality-Specific Adaptation for Incremental Continual Learning in Parkinson's Disease Gait Assessment — MOSAIC resolves modality-incremental learning challenges in PD gait assessment by stabilizing new sensor representations and decoupling statistics.
• EEGDancer: Dynamic Emotion Latent Space Masked Modeling with Reinforcement Learning for EEG Continuous Emotion Prediction — Continuous EEG emotion prediction benefits from a unified framework combining discrete latent space learning, masked temporal modeling, and reinforcement learning.
• SMART: A Flexible, Interpretable, and Scalable Spatio-temporal Brain Atlas from High-Resolution Imaging Data — SMART creates flexible, interpretable brain atlases by decoupling global disease dynamics from patient-specific anatomy using differential equations and Neural Cellular Automata.
• Phenotyping TPF via Self-Supervised Learning: A Label-Agnostic Framework with Expert Validation — Self-supervised learning can discover stable, clinically interpretable fracture phenotypes without relying on inconsistent human labels.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.