Machine-learning algorithms for the identification of visual field loss associated with the anti-seizure medication vigabatrin - a proof of concept

Aug 5, 2026

Upload by Levi

Vigabatrin, an antiseizure medication, is linked to visual field loss (VAVFL), yet clinical interpretation is difficult because the defect pattern may be unfamiliar and perimetry can be unreliable, especially in epilepsy patients with cognitive limitations. Two machine-learning pattern-recognition algorithms were developed to identify VAVFL objectively using an EMA-approved screening and threshold protocol. Algorithms compared each eye’s measured field with severity-matched reference patterns and optionally applied symmetrisation as a signal-to-noise enhancement. Results showed excellent agreement with clinical “gold standard” interpretation, with symmetrisation improving performance in poorly recorded and concomitant homonymous loss cases.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages23

Reading Time58 min

AccessFree to download

Abstract

Background/ Aims

Methods

Results

Conclusion

What Is Already Known

What This Study Adds

How This Study Might Affected Research, Practice Or Policy

Introduction

ORCA – Online Research @ Cardiff This is an Open Access document downloaded from ORCA, Cardiff University's institutional repository:[https://orca.cardiff.ac.uk/id/eprint/177314/](https://orca.cardiff.ac.uk/id/eprint/177314/) This is the author’s version of a work that was submitted to / accepted for publication. Citation for final published version: Wild, John M. , Smith, Philip E.M. and Knupp, Carlo 2025. Machine-learning algorithms for the identification of visual field loss associated with the antiseizure medication vigabatrin-a proof of concept. British Journal of Ophthalmology 109 (8) , pp. 932-940. 10.1136/bjo-2024-325804 Publishers page: [https://doi.org/10.1136/bjo-2024-325804](https://doi.org/10.1136/bjo-2024-325804) Please note: Changes made as a result of publishing processes such as copy-editing, formatting and page numbers may not be reflected in this version. For the definitive version of this publication, please refer to the published source. You are advised to consult the publisher’s version if you wish to cite this paper. This version is being made available in accordance with publisher policies. See [http://orca.cf.ac.uk/policies.html](http://orca.cf.ac.uk/policies.html) for usage policies. Copyright and moral rights for publications made available in ORCA are retained by the copyright holders. Machine-learning algorithms for the identification of visual field loss associated with the anti-seizure medication vigabatrin-a proof of concept. John M Wild*1 , Philip EM Smith2 and Carlo Knupp1 1. Cardiff Centre for Vision Sciences Research College of Biomedical and Life Sciences Cardiff University Cardiff CF5 6DE United Kingdom 2. The Alan Richens Unit Welsh Epilepsy Centre University Hospital of Wales Heath Park Cardiff CF14 4XW United Kingdom * Corresponding author John M Wild [wildjm@cardiff.ac.uk](wildjm@cardiff.ac.uk) Philip E M Smith [smithpe@cardiff.ac.uk](smithpe@cardiff.ac.uk) Carlo Knupp [knuppc@cardiff.ac.uk](knuppc@cardiff.ac.uk) ABSTRACT Background/ Aims The antiseizure medication, vigabatrin, is associated with visual field loss (VAVFL) . However, the fields can be challenging to interpret due to unfamiliarity with the characteristics of the defect and/ or to difficulty in obtaining a reliable examination, particularly in patients with cognitive limitations associated with the epilepsy. Two machine-learning pattern recognition algorithms were developed to identify VAVFL, objectively. Methods The algorithms adhered to the European Medicines Agency-approved protocol for the detection of VAVFL (Three Zone Age Corrected Full Field 135 Screening Test [FF135] and the Central C30-2 Threshold Test [C30-2T] with the Humphrey Field Analyzer) . Each algorithm compared the similarity of the measured field from each eye to that of modelled reference patterns of VAVFL, matched for equivalent severity, and objectively derived from a previously described case series of 123 adults. The algorithms were augmented by the optional inclusion of symmetrisation, a signal-to-noise enhancement technique based upon the between-eye mirror image symmetry of VAVFL. Utility of the algorithms for identifying VAVFL was evaluated against a case series of 89 consecutively identified individuals stratified across six diagnostic categories including homonymous and glaucomatous losses. Results The algorithms exhibited excellent agreement with a ‘gold standard’ clinical interpretation (sensitivity and specificity: FF135, 22/23; 30/30 ; C30-2T, 17/18; 48/51) . Symmetrisation was particularly useful in identifying VAVFL when perimetric learning or fatigue influenced the outcome for one eye and for visualisation in the presence of concomitant homonymous loss. Conclusion The directly-interpretable machine learning outcome correctly identified VAVFL and could assist patient management in community (neuro-)ophthalmology. WHAT IS ALREADY KNOWN Deep learning Artificial Intelligence t

FAQs

Why is identifying vigabatrin-associated visual field loss (VAVFL) clinically challenging?

VAVFL can be hard to interpret due to unfamiliar defect patterns and because reliable perimetry is difficult in epilepsy patients, particularly those with cognitive limitations.

How do the proposed machine-learning algorithms identify VAVFL?

Each algorithm matches the similarity between an eye’s measured field and severity-equivalent model reference patterns derived from a prior case series, using an EMA-approved perimetric protocol. Symmetrisation can be optionally applied to enhance signal-to-noise.

How well did the algorithms agree with clinical interpretation, and when was symmetrisation most helpful?

The algorithms demonstrated excellent agreement with a gold-standard clinical interpretation. Symmetrisation was especially useful when perimetric learning or fatigue affected one eye and for visualisation when concomitant homonymous loss was present.

Aug 5, 2026

Upload by Levi