Page 28 - AN-4-3

Advanced Neurology Artificial intelligence in epilepsy education

Meanwhile, Harvard Medical School is working on LSTM- sensitivity. This model reduced false positives, thus
60
based models for real-time seizure prediction, which are aiding accurate early diagnosis. Similarly, Altaheri et al.
61
60
vital tools for mitigating risks associated with SUDEP and applied a deep learning-based EEG analysis method that
enhancing patient safety. Collectively, these institutions enhanced detection rates and simplified the identification
61
are shaping the future of epilepsy care by creating accessible, of complex seizure patterns, improving real-time clinical
accurate diagnostic tools that significantly improve patient interventions. Another study that incorporated predictive
61
outcomes globally. 58-61 Table 3 highlights the contributions models using EEG and electrocardiogram (ECG) data
of these institutions in advancing AI’s role in epilepsy achieved high performance in seizure prediction,
management. facilitating timely interventions and comprehensive patient
monitoring. These examples illustrate the advancement
63
9. Case studies and practical of engineering science in utilizing AI to enhance epilepsy
implementations diagnosis, streamline clinical workflows, and support
Various research efforts have underscored the utility of AI informed decision-making in patient care. 60-63
in epilepsy outreach and the central goals within this field. 10. Barriers to AI adoption in healthcare
57
AI models for detecting and forecasting epileptiform EEG
patterns and seizures have yielded significant outcomes in and education
clinical applications. For instance, the AI-based model Despite the need to integrate AI into teaching, learning,
58
mjn-SERAS exhibits notable sensitivity and specificity in and healthcare, several challenges must be addressed
early seizure detection using customized mathematical to facilitate better integration. 54,55 Concerns include a
models for individualized patient care through EEG lack of technical proficiency in medical schools, often
analysis. Research highlights specific AI applications stemming from inaccurate descriptions and volunteered
59
in epilepsy, focusing on seizure detection, prediction, information given by generative intelligence models.
61
and localization. Jeon et al. demonstrated a deep- In addition, issues related to data privacy, moral and
60
59
learning model for identifying epileptiform discharges legal aspects, compatibility issues, and complexities in
in self-limited epilepsy, achieving high specificity and human-AI communication hinder the integration of AI in
Table 2. Key findings of machine learning models for detecting epileptic seizures based on electroencephalogram

Authors Year Type of study Results (Statistical values)
Hannun et al. 47 2019 Deep neural network for arrhythmia detection ROC AUC: 0.97, F1 Score: 0.837
Zhou et al. 48 2024 Heartbeat classification using CNNs and transformer Accuracy: 99.4%
Siddiqui et al. 49 2020 Review of machine learning classifiers for seizure detection Overview of classifiers and features
Yang et al. 50 2022 An AI system for clinical seizure recognition High sensitivity and specificity
Mirowski et al. 51 2019 Comparison of SVM and CNN for seizure prediction High accuracy, CNN slightly better
Abbreviations: AUC: Area under the curve; CNN: Convolutional neural network; LSTM: Long short-term memory; RNN: Recurrent neural network;
ROC: Receiver operating characteristic; SVM: Support vector machine.
Table 3. Contributions of different institutions in advancing artificial intelligence in epilepsy management

Institution Location Key focus areas in epilepsy AI Selected contributions
Mayo Clinic USA EEG signal analysis, seizure prediction, patient monitoring Mayo Clinic AI Lab on Epilepsy
Boston Children’s Hospital USA Pediatric epilepsy, EEG monitoring, and AI-driven educational AI-EEG study for early seizure
resources diagnosis
University College London UK EEG and MRI imaging analysis, epilepsy surgery decision support UCL AI for epilepsy diagnostics
Seoul National University South Korea Machine learning for seizure classification, patient adherence Epilepsy AI Lab, Seoul National
strategies University
Indian Institute of Science India Deep learning for EEG analysis and epilepsy care in low-resource Epilepsy AI in developing regions
settings
WHO Collaborating Centre International Ethical and social aspects of AI in epilepsy global access to AI WHO Report on AI in Epilepsy 62
on Epilepsy diagnostics
Abbreviations: AI: Artificial intelligence; EEG: Electroencephalography; MRI: Magnetic resonance imaging; UCL: University College London;
UK: United Kingdom, USA: United States of America; WHO: World Health Organization.

Volume 4 Issue 3 (2025) 22 doi: 10.36922/an.4777