Forecasting and real-time detection of neonatal seizures: A machine learning perspective.

Real-Time Epileptic Seizure Detection from Raw EEG Using Classical Machine Learning and Time-Domain Feature

Automated detection of epileptic seizures from electroencephalogram (EEG) recordings is essential for timely clinical intervention and long-term patient monitoring. Deep learning achieves high accuracy, but its limited interpretability and computational demands restrict deployment in resource-constrained, real-time clinical environments; furthermore, classical machine learning studies have concentrated on a narrow group of well-known temporal features. This study systematically introduces and evaluates 25 less-explored time-domain features, 13 of which have no documented precedent as classification features in scalp EEG seizure detection, against 25 classical features and their 50-feature combination. Raw, unfiltered recordings from the CHB-MIT database were segmented into 10-second windows, and seven classical classifiers were optimized with GridSearchCV under subject-wise StratifiedGroupKFold cross-validation, with the data partitioned at the patient level such that no patient appeared in both the training and test sets. The multilayer perceptron trained on the combined 50-feature set performed best (accuracy 86.59%, F1-score 86.35%), exceeding the classical features alone by 5.1 and 4.6 percentage points, respectively; the less-explored features alone remained competitive (84.63%, 84.67%). SHAP analysis identified the exponent of the detrended fluctuation analysis (DFA)—a long-range, nonlinear measure of temporal correlation—as the most influential predictor. The curated feature set, its rigorous subject-level validation, and its interpretability provide a reproducible and computationally efficient foundation for future clinical deployment.

Bridging Feature-Based and Temporal Learning for EEG-Based Epileptic Seizure Detection

A novel lightweight comparative ML-DL framework for detecting seizures with EEG signals recorded using the UCI Epileptic Seizure Recognition Dataset is presented in this article. The data collected from the experiments was adjusted to form a binary classification problem, where the elements of the dataset were divided into seizure versus non-seizure. The number of features was reduced from 178 to 148 with a variance feature reduction method while maintaining the discriminative quality of the features. Four traditional machine learning models were compared with a Long Short-Term Memory (LSTM) network model to capture the temporal aspects of the EEG signal. The Random Forest algorithm produced an overall accuracy of 97.08%, while the LSTM produced an accuracy of 98%. This finding is indicative that when a framework is developed with an LSTM and temporal modeling capabilities, the performance increases, therefore making it more suitable for real-time seizure detection.

An Efficient Dual-Specialist Framework for Real-Time Epileptic Seizure Detection and Prediction

Real-time epileptic seizure monitoring requires simultaneous detection and prediction capabilities, yet existing systems address only one task, leading to either delayed response or excessive false alarms. This paper presents a dual-specialist deep learning framework that performs concurrent seizure detection and prediction through three key innovations: (1) seizure-sensitive preprocessing with cross-frequency coupling detection and adaptive channel weighting, (2) complementary dual-specialist architecture where a detection specialist optimizes for rapid response while a prediction specialist optimizes for conservative forecasting, and (3) probabilistic temporal state machine providing interpretable risk assessment through four clinical states with sustained evidence validation. Evaluation on the CHB-MIT database using Leave-One-Seizure-Out cross-validation across 20 patients (98 seizures, 851.1 hours) demonstrates 99.2% sensitivity, 99.7% specificity, and 0.31/h false alarm rate with 7.9-minute average warning time, achieved with 1.2M parameters and 1.58M multiply-accumulate operations. Real-time Raspberry Pi 4 deployment achieves 108.0 ± 13.5 ms inference latency, validating practical edge deployment. The framework demonstrates potential for wearable seizure monitoring applications with improved false alarm control.

Neonatal Seizure Detection Using Combined aEEG and Compressed Spectral Array Features: A Machine-Learning Proof-of-Concept Study

Epileptic Seizure Detection from EEG Signal Using Progressive Channel Selection and Deep Learning

: Epileptic Seizures (ES), marked by abrupt abnormal electrical discharges in the brain, represent the primary symptom of epilepsy, making timely and accurate detection critical because of their unpredictable recurrence and potentially severe consequences. The Electroencephalogram (EEG), a widely used, non-invasive, and economical method for monitoring brain activity, plays a key role in automated ES detection through Machine Learning (ML) and Deep Learning (DL) techniques. Most existing ML/DL approaches typically utilize all available EEG channels or apply patient-specific channel selection to maximize detection performance. This study investigates a novel Progressive Channel Selection (PCS) framework designed to identify and retain only the most informative EEG channels across patients. The proposed method progressively selects channels in multiple stages according to their contribution to classification accuracy, ensuring that only the most influential channels are preserved. Experimental evaluation was performed using both Neural Network (NN) and Convolutional Neural Network (CNN) models on the CHB-MIT dataset. The CNN model achieved an accuracy of 98.99% while utilizing fewer than half of the available EEG channels. Although some existing approaches report slightly higher accuracy by employing all channels, the proposed method provides a more effective trade-off between detection accuracy and channel efficiency. The identification of informative EEG channels could support hardware simplification and may facilitate low-cost EEG-based ES detection systems.

An Optimized Deep Learning Framework for Robust Epileptic Seizure Detection Using EEG Signals

Epilepsy affects a person's behavior and is characterized by seizures. Epileptic seizures are infamously difficult to treat because of how unpredictable they are. Automatic electroencephalogram (EEG) seizure detection allows medical professionals to better predict when patients may have seizures and adjust treatment plans appropriately. The signals only show one side of the brain, making it impossible to accurately identify epileptic seizures. When done manually, feature extraction also consumes a significant amount of time. To build seizure detection systems reliant on multi-view feature learning via automatic means is not feasible in their absence. A proposed method for detecting epileptic seizures from electroencephalogram data involves creating an optimal deep learning architecture. Using sophisticated hyperparameter tuning techniques, the system incorporates deep learning architectures, feature optimisation, and wavelet-based preprocessing. Results show that compared to previous approaches, ours are much more accurate, robust, and computationally efficient, according to extensive trials conducted on several benchmark datasets. Clinical and real-time deployment of the suggested method is confirmed to be effective by statistical validation.