BioEP: From prototype to clinical evaluation

Background 125,000 people attend a first seizure clinic annually with suspected epilepsy. 40,000 eventually receive an epilepsy diagnosis following an average delay of 1-year and with a misdiagnosis rate of 30%.

The Electroencephalogram is a primary clinical tool for revealing brain abnormalities indicative of epilepsy, however 70% of diagnostic EEGs appear clinically normal.

To address this, we have developed a novel computer BIOmarker of EPilepsy: BioEP, that can reveal seizure risk from apparently normal EEG.

Aims and Objectives The main aim of this phase II project is to develop our existing BioEP prototype to a minimum viable product (MVP) and to secure CE marking.

The key objectives are to: • use EEG and clinical meta-data (including diagnosis) to develop machine learning methods to further increase accuracy of the BioEP seizure risk score; • develop the BioEP platform in collaboration with clinical users and people with lived experience; • evaluate the clinical utility of the technology through user testing; • gather necessary evidence for CE marking.

Project plan and methods used Our project is split into six worpackages spanning algorithm development, platform development, clinical evaluation, regulation and compliance, commercialisation and reporting and dissemination.

The algorithms we will develop utilise a number of machine learning approaches including random forest, automatic relevance determination (ARD), Bayesian shrinkage methods, least absolute shrinkage and selection operator (LASSO) regression and K-means clustering. The BioEP platform will be developed utilising Microsoft's .Net framework.

Timelines for delivery We will complete the project in 24 months.

Anticipated Impact and Dissemination We will deliver an MVP with a detailed product specification document and clinical evaluation file.

Research findings and clinical evaluation studies will be communicated in peer-reviewed publications and at scientific meetings.

We will work closely with key stakeholders, including those with lived experience, to disseminate findings through relevant networks.