{"slug": "hybrid-ensemble-learning-for-eeg-based-epileptic-seizure-forecasting", "title": "Hybrid Ensemble Learning for EEG-Based Epileptic Seizure Forecasting", "summary": "A calibrated hybrid ensemble combining five deep learning models and three classical machine learning models via a logistic regression stacking meta-learner achieved 74.2% seizure-level sensitivity at 1.24 false alarms per hour, with an average warning time of 16.9 minutes, on the CHB-MIT dataset under strict Leave-One-Patient-Out cross-validation. The framework, evaluated on a filtered cohort excluding patients with anomalous preictal rates below 1% or above 15%, also reached 60.9% sensitivity at 0.951 false alarms per hour when a test-tuned oracle was constrained to the target false-alarm budget. The authors state the results highlight the importance of reporting sensitivity together with realized false-alarm rates, and released code at a public GitHub repository.", "body_md": "arXiv:2609.35876v1 Announce Type: new \nAbstract: Epileptic seizure forecasting aims to provide actionable warnings before seizure onset, yet patient-independent generalization and false-alarm control remain major challenges. We propose a calibrated hybrid ensemble for EEG-based seizure forecasting that combines five deep learning models and three classical machine learning models through a logistic regression stacking meta-learner. The proposed pipeline integrates signal preprocessing, handcrafted feature extraction, class-imbalance handling, probability calibration, and clinically motivated post-processing. We evaluate the framework on CHB-MIT using strict Leave-One-Patient-Out (LOPO) cross-validation, with threshold and post-processing parameters selected only on held-out meta data. On the filtered cohort, excluding patients with anomalous preictal rates below 1\\% or above 15\\%, the model achieves 74.2\\% seizure-level sensitivity at 1.24 false alarms per hour, with an average warning time of 16.9 minutes. A test-tuned oracle constrained to the target false-alarm budget achieves 60.9\\% sensitivity at 0.951 false alarms per hour, highlighting the importance of reporting sensitivity together with realized false-alarm rates. Our code is available at: https://github.com/DanaMason/IEEE-CARS-Hybrid-Ensemble-Learning-for-EEG-Based-Epileptic-Seizure-Forecasting", "url": "https://wpnews.pro/news/hybrid-ensemble-learning-for-eeg-based-epileptic-seizure-forecasting", "canonical_source": "https://arxiv.org/abs/2609.35876", "published_at": "2026-09-30 04:00:00+00:00", "updated_at": "2026-09-30 04:19:33.754748+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "neural-networks"], "entities": ["CHB-MIT", "GitHub", "DanaMason"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/hybrid-ensemble-learning-for-eeg-based-epileptic-seizure-forecasting", "markdown": "https://wpnews.pro/news/hybrid-ensemble-learning-for-eeg-based-epileptic-seizure-forecasting.md", "text": "https://wpnews.pro/news/hybrid-ensemble-learning-for-eeg-based-epileptic-seizure-forecasting.txt", "jsonld": "https://wpnews.pro/news/hybrid-ensemble-learning-for-eeg-based-epileptic-seizure-forecasting.jsonld"}}