Prioritising Follow-Up for People with Suspected Epilepsy Using a Digital EEG Biomarker.

Abstract

Lengthy waits for follow-up testing are common for people with suspected epilepsy. This delays diagnosis, prolongs uncertainty and increases seizure risk. Initial EEGs are frequently inconclusive, yet follow-ups are typically dictated by referral date, rather than risk. Here, we tested whether a digital EEG biomarker could help prioritise those most likely to have epilepsy for expedited EEG testing. We analysed 196 non-contributory (non-diagnostic) initial EEGs collected from six National Health Service (NHS) sites in England. From these recordings, we extracted eight computational features and derived a digital biomarker that quantifies the likelihood of the EEG was recorded from someone with active epilepsy. We used this information to reorder follow-up lists and compared outcomes against standard referral-based scheduling. We found that ordering for follow-up testing based upon the digital biomarker consistently prioritised people subsequently diagnosed with epilepsy. The diagnostic yield of each subsequent EEG performed was increased relative to orderings based on time of referral. Our study indicates that a routine EEG may furnish an objective risk metric that could accelerate second-line investigations and so reduce diagnostic delay whilst improving resource allocation in clinical practice.