By: Catherine Joachin
Artificial Intelligence and Epilepsy
What is Artificial intelligence?
Artificial intelligence (AI) refers to a digital computer or artificial system’s ability to carry out complex tasks typically performed by humans, such as those involving reasoning and decision-making (NASA,2024). AI tools deliver a wide range of tasks and outputs; therefore, the term ‘artificial intelligence’ lacks a single clear definition (NASA, 2024).
AI tools often fall into two categories: machine learning and deep learning approaches.
- Machine learning uses algorithms that learn from data, allowing computers to make predictions and decisions autonomously. Although human intervention is still required to ensure accuracy.
- Deep learning is a subfield of machine learning that relies on multi-layered artificial neural networks that learn from large, complex datasets. This type of machine learning is well-suited for image and speech recognition tasks and does not require human intervention.
(National Institute of Biomedical Imaging and Bioengineering [NIBIB], 2025; Singapore Computer Society, 2020).
Today, the integration of AI into health and biomedical research is reshaping medical care by supporting clinical decision-making, patient monitoring and enhanced imaging data interpretation (NIBIB, 2025).
AI in epilepsy
AI is largely used to optimize data collection and analysis for the study of epilepsy. Thanks to recent advances in data processing power, AI-driven methods have emerged as powerful tools to analyze medical imaging data across different brain imaging modalities, including structural MRI, PET and fMRI (Sollee et al. 2022).
Current applications of AI in neuroimaging suggest that it can improve epilepsy diagnosis and management by aiding in the detection of structural brain abnormalities, helping localize epileptic activity, offering better neuroimaging quality by leveraging deep learning algorithms, and having the potential to predict surgical outcomes using FDG PET data (Lucas, Revell & Davis, 2024).
Challenges in integrating of AI into clinical practice
Despite a rise in the development AI tools for epilepsy management, few have been successfully implemented into clinical practice (Lucas, Revell & Davis, 2024). An important challenge in the application of AI models is obtaining sufficiently large datasets, as medical imaging data is limited (Sollee et al. 2022). Moreover, the data found in many epilepsy studies is imbalanced, meaning that it contains an significant underrepresentation of a group and the overrepresentation of another. Coupled with low image quality and small sample size, these factors increases the risk of producing biased AI models (Sollee et al. 2022).
Contamination due to data augmentation in deep learning models also poses an issue in terms of data portioning, but the biggest issue remains generalizability, underscoring the need for more external validation studies and publicly available datasets to support the role of AI models in clinical practice (Sollee et al. 2022).
Conclusion
Research suggests that artificial intelligence is a valuable tool for optimizing epilepsy diagnosis, prognosis and treatment; however, more studies and available medical imaging data are needed to integrate AI models into clinical settings with success.
References
Lucas, A., Revell, A. & Davis, K.A. Artificial intelligence in epilepsy — applications and pathways to the clinic. Nat Rev Neurol 20, 319–336 (2024). https://doi.org/10.1038/s41582-024-00965-9
National Aeronautics and Administration (2024). What is Artificial Intelligence? NASA. Retrieved from: https://www.nasa.gov/what-is-artificial-intelligence/
National Institute of Biomedical Imaging and Bioengineering (2025). Artificial Intelligence (AI). National Institute of Biomedical Imaging and Bioengineering. Retrieved from: https://www.nibib.nih.gov/science-education/science-topics/artificial-intelligence-ai
Singapore Computer Society (2020). Machine learning vs. deep learning. Singapore Computer Society. Retrieved from: https://www.scs.org.sg/articles/machine-learning-vs-deep-learning
Sollee, J., Tang, L., Igiraneza, A. B., Xiao, B., Bai, H. X., & Yang, L. (2022). Artificial intelligence for medical image analysis in epilepsy. Epilepsy Research, 182, Article 106861. https://doi.org/10.1016/j.eplepsyres.2022.106861
