Related Experiment Video
Updated: Aug 6, 2026
Continuous Video Electroencephalogram during Hypoxia-Ischemia in Neonatal Mice
Published on: June 11, 2020
Forecasting and real-time detection of neonatal seizures: A machine learning perspective
Tamara Skoric1, Marija Djermanovic2, John M O'Toole3
1University of Novi Sad, Faculty of Technical Sciences, Trg Dositeja Obradovica 6, Novi Sad, 21000, Serbia.
Insights
This study introduces a machine learning model for forecasting and detecting neonatal seizures using electroencephalogram (EEG) data. The model shows promise in improving timely seizure management in intensive care settings.
Area of Science:
- Neuroscience
- Medical Technology
- Machine Learning
Background:
- Neonatal seizures require prompt recognition and treatment to prevent brain damage.
- Current treatment delays, with only 11% of seizures treated within an hour, highlight the need for advanced detection systems.
- Limited neurologist availability in intensive care units necessitates automated seizure forecasting and detection solutions.
Purpose of the Study:
- To develop and evaluate a dual-function machine learning (ML) architecture for both forecasting and real-time detection of neonatal seizures.
- To utilize explainable, feature-based methods for improved accuracy and interpretability in seizure prediction.
- To assess the model's performance on both single-channel and multi-channel electroencephalogram (EEG) datasets.
Main Methods:
- Proposed an ML architecture employing entropy, singular value decomposition, power spectrum, and statistical moment-based features.
- Utilized an AdaBoost classifier with a unified feature set for both seizure forecasting (5, 10, 15 min intervals) and real-time detection.
- Trained and validated the model on a single-channel (n=82) and a public multi-channel (n=79) EEG dataset.
Main Results:
- The ML model achieved high sensitivity and specificity for seizure forecasting, particularly at a 5-min interval (95.6%/88.9%) with a Matthews correlation coefficient (MCC) of 0.55.
- Short-range forecasting performance surpassed previous studies on a multi-channel dataset (MCC=0.48).
- The proposed model demonstrated over 30% improvement in MCC and Pearson's coefficient compared to ConvNeXt on the single-channel dataset.
Conclusions:
- EEG segmentation and feature selection significantly enhance ML model performance for seizure detection, even with limited data.
- Feature-based ML models require validation across diverse EEG channel configurations (single and multi-channel).
- While outperforming existing methods, further validation and performance enhancement are crucial for clinical integration of this seizure forecasting system.
Background And Objectives:
Recognition and treatment of neonatal seizures, primarily diagnosed using EEG, are essential to protect the developing brain, yet only 11% of seizures are treated within 1 hour of onset. Developing seizure forecasting and detection systems could help address this issue in intensive care settings with limited neurologist availability.
Methods:
We propose an ML architecture with dual functionality: (1) forecasting seizures within short prediction intervals (5, 10, and 15 min), and (2) detecting seizures in real time, using explainable entropy-, singular value decomposition-, power spectrum-, and statistical moment-based features with an AdaBoost classifier. The same feature set supports both functionalities, with forecast reliability improving when features are derived from longer EEG intervals. The model was trained and tested on a single-channel (P3-P4, n = 82) dataset and a multi-channel public dataset (n = 79).
Results:
The proposed ML model was able to forecast seizures at a 15-min prediction interval (86.6%/92.4% sensitivity/specificity; Matthews correlation coefficient, MCC = 0.39), with better performance at the shorter 5-min prediction interval (95.6%/88.9% sensitivity/specificity; MCC = 0.55) for the single-channel EEG seizure group. Short-range forecasting outperformed the pioneering study on the public multi-channel EEG dataset (MCC = 0.48). The proposed ML model achieved over a 30% improvement in MCC and Pearson's coefficient on the single-channel dataset compared to the state-of-the-art deep-learning architecture (ConvNeXt).
Conclusion:
EEG segmentation and appropriate feature selection enhance ML model performance for seizure detection in small datasets. Validation of feature-based ML models is required on both single and multi-channel EEG, given that quantitative features can be sensitive to different bipolar channel pairs. Although the method has outperformed the pioneering studies in seizure forecasting, further improvement in performance and more extensive validation are needed for clinical adoption.
