Abstracts

Clinical diagnosis of epilepsy is based on expert analysis of the likelihood of further seizures, a decision that considers multiple factors including a person's medical history and analysis of routine diagnostic tests such as scalp EEG. Due to the unpredictable nature of seizures, epilepsy is often difficult to diagnose and treat. In the UK, 125,000 people are referred to first seizure clinics per annum with suspected epilepsy of which 40,000 receive a confirmed diagnosis of epilepsy.

Mathematical and computational analysis of background EEG has led to several candidate biomarkers. A recent Phase II multi-site, retrospective study (N=281) on clinically non-informative EEGs validated a subset of these candidate biomarkers in a single statistical classifier (‘BioEP’; Tait et al. 2024). By offering potential decision support in clinically non-informative EEGs, these methods might contribute to reduced diagnostic delay and misdiagnosis rates.

A prospective single site diagnostic belief updating study to examine the utility of BioEP in supporting clinicians’ decision-making was conducted. Adults with suspected seizures attending a first seizure clinic were asked take part by a Consultant Nurse for the Epilepsies (CNE) at the Royal Wolverhampton NHS Trust (UK), with a target of N = 88 subjects. For each patient, the clinician rated the probability of having another epileptic seizure on a 7-point scale ranging from ‘virtually certain’ to ‘exceptionally unlikely’. This rating was conducted before (initial belief) and after (updated belief) the presentation of the BioEP results. Recruitment took place over 1 year, with patients receiving routine standard of care.

A total of N = 91 individual patients suspected of having epilepsy were asked to participate, with a 100% uptake. Four patients were withdrawn during the study, leading to a final total of N = 87 subjects (F = 40; mean age = 42.9 yrs, std = 19.6 yrs). No EEG files were excluded (i.e. all inclusion criteria met). The CNE reported a generally positive attitude towards the BioEP score as measured through self-reported confidence.

Our study considers the impact on self-reported clinical confidence in decision-making of the use of EEG-derived computational biomarkers. Future work should assess the change in diagnostic yield and time to diagnosis when utilising these biomarkers in carefully designed prospective multi-site studies.