Related Experiment Video
Updated: Aug 5, 2026
06:18
In Vivo Visualization of Spontaneous Activity in Neonatal Mouse Sensory Cortex at a Single-Neuron Resolution
Published on: November 21, 2023
A Lightweight Temporal-Spatial Fusion Network for Neonatal Sleep Staging
Ligang Zhou1, Laishuan Wang2, Yan Xu2
1Center for Medical Research and Innovation, Shanghai Pudong Hospital, Human Phenome Institute, Fudan University, 825 Zhangheng Road, Pudong, Shanghai 201203, China.
Bioengineering (Basel, Switzerland)
|July 28, 2026
Summary
We developed a lightweight AI model for automatic neonatal sleep staging using EEG data. This method efficiently analyzes brain activity, improving accuracy and aiding in the early detection of neurological disorders in infants.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Biomedical Engineering
Background:
- Accurate neonatal sleep assessment is vital for monitoring infant brain development and detecting neurological disorders.
- Manual scoring of electroencephalogram (EEG) recordings is time-consuming and inconsistent.
Purpose of the Study:
- To develop an efficient and accurate automatic sleep staging system for neonates.
- To address the limitations of manual EEG analysis in clinical settings.
Main Methods:
- Proposed a lightweight temporal-spatial feature fusion network for automatic neonatal sleep staging.
- Employed a dual-branch architecture to capture temporal and spatial EEG signal dependencies.
- Integrated features using concatenation and a compact classifier for comprehensive representation with low complexity.
Main Results:
- Achieved superior performance in sleep-wake classification, quiet sleep detection, and three-stage sleep staging on a neonatal dataset (CHFD).
- Demonstrated competitive accuracy and F1-score on an adult dataset (MASS-S3), indicating strong generalization.
- Outperformed several state-of-the-art methods in neonatal sleep staging tasks.
Conclusions:
- Jointly modeling temporal and spatial EEG features enables robust and efficient automatic sleep staging.
- The proposed approach provides a practical solution for clinical applications and edge deployment.
- Offers reliable, multi-dimensional assessment of neonatal brain activity, supporting future research.