Construction and validation of an algorithm to separate focal and generalised epilepsy using clinical variables - a comparison of machine learning approaches

Aug 5, 2026

Upload by Rowan

Purpose: Distinguishing focal versus generalised epilepsy is critical for selecting anti-seizure medication, especially in resource-limited settings where EEG and neuroimaging are often unavailable. Methods: An algorithm was built using 28 clinical variables from 503 patients, using a naïve Bayes approach and mutual-information feature selection, then compared with five other machine learning methods. Results: The best naïve Bayes model achieved 92.2% accuracy, and an eight-variable variant reached 91.0%. Conclusion: A clinician-friendly eight-variable clinical algorithm can accurately separate focal from generalised epilepsy without EEG or imaging.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages22

Reading Time55 min

AccessFree to download

Abstract

Purpose

Methods

Results

Conclusion

Introduction

Construction and validation of an algorithm to separate focal and generalised epilepsy using clinical variables: A comparison of machine learning approaches Patterson, V. , Glass, D. H. , Kumar, S. , El-Sadig, S. M. , Mohamed, I. , El-Amin, R. , & Singh, M. (2024) . Construction and validation of an algorithm to separate focal and generalised epilepsy using clinical variables: A comparison of machine learning approaches. Epilepsy & behavior : E&B, 155, 1-8. Article 109793. Advance online publication. [https://doi.org/10.1016/j.yebeh.2024.109793](https://doi.org/10.1016/j.yebeh.2024.109793) Link to publication record in Ulster University Research Portal Published in: Epilepsy & behavior : E&B Publication Status: Published online: 25/04/2024 DOI: 10.1016/j.yebeh.2024.109793 Document Version Author Accepted version Document Licence: CC BY-NC-ND General rights The copyright and moral rights to the output are retained by the output author(s), unless otherwise stated by the document licence. Unless otherwise stated, users are permitted to download a copy of the output for personal study or non-commercial research and are permitted to freely distribute the URL of the output. They are not permitted to alter, reproduce, distribute or make any commercial use of the output without obtaining the permission of the author(s) . If the document is licenced under Creative Commons, the rights of users of the documents can be found at [https://creativecommons.org/share-your-work/cclicenses/](https://creativecommons.org/share-your-work/cclicenses/) . Take down policy The Research Portal is Ulster University's institutional repository that provides access to Ulster's research outputs. Every effort has been made to ensure that content in the Research Portal does not infringe any person's rights, or applicable UK laws. If you discover content in the Research Portal that you believe breaches copyright or violates any law, please contact [pure-support@ulster.ac.uk](pure-support@ulster.ac.uk) Download date: 10/05/2025 Construction and Validation of an Algorithm to Separate Focal and Generalised Epilepsy using Clinical Variables: a comparison of machine learning approaches Victor Patterson 1 David H. Glass2 Shambhu Kumar3 Sarah Misbah El-Sadig4 Inaam Mohamed 5 Rahba El-Amin4 Mamta Singh3 1 Independent neurologist, Belfast, UK 2 School of Computing, Ulster University, Belfast, UK 3 Department of Neurology, All India Institute of Medical Sciences, New Delhi, India 4 Department of Medicine, University of Khartoum, Khartoum, Sudan 5 Department of Paediatrics, University of Khartoum, Khartoum, Sudan Corresponding author: Dr Victor Patterson, [vhp498@gmail.com](vhp498@gmail.com) Abstract Purpose Epilepsy type, whether focal or generalised, is important in deciding anti-seizure medication (ASM) . In resource-limited settings, investigations are usually not available, so a clinical separation is required. We used a naïve Bayes approach to devise an algorithm to do this , and compared its accuracy with algorithms devised by five other machine learning methods. Methods We used data on 28 clinical variables from 503 patients attending an epilepsy clinic in India with defined epilepsy type, as determined by an epileptologist with access to clinical, imaging, and EEG data. We adopted a machine learning approach to select the most relevant variables based on mutual information, to train the model on part of the data, and then to evaluate it on the remaining data (testing set) . We used anaïve Bayes approach and compared the results in the testing set with those obtained by several other machine learning algorithms by measuring sensitivity, specificity, accuracy , area under the curve, and Cohen’s kappa. Results The six machine learning methods produced broadly similar results. The best naïve Bayes algorithm contained eleven variables, and its accuracy was 92.2% in determining epilepsy type (sensitivi

FAQs

Why is separating focal and generalised epilepsy clinically important?

Epilepsy type guides the choice of anti-seizure medication and affects treatment effectiveness and prognosis, including seizure-freedom likelihood after withdrawal of ASMs.

How was the algorithm constructed and evaluated?

The study used 28 clinical variables from 503 patients, selected via mutual information, trained on a portion of the data, and evaluated on a held-out testing set using metrics such as sensitivity, specificity, accuracy, AUC, and Cohen’s kappa.

What were the key performance results compared with other machine learning methods?

Six machine learning approaches produced broadly similar results. The best naïve Bayes model reached 92.2% accuracy, while an eight-variable naïve Bayes model achieved 91.0% accuracy and was easier for clinicians to use.

Aug 5, 2026

Upload by Rowan