× Home Projects Partners FAQs Contacts
ihi-logo

EPILEPSY: Scientists develop tool for detecting epilepsy

May 9, 2023
EPILEPSY: Scientists develop tool for detecting epilepsy
A snip from The Lancet Digital Health journal which published Dr. Honorati Masanja and colleagues study. GRAPHIC | IFAKARA/KMC

A collaborative study by scientists from Europe and Africa has led to the development of a diagnostic tool for convulsive epilepsy in resource-poor settings within sub-Saharan Africa. The tool which can reduce diagnostic gaps was developed and tested on a large dataset of African individuals.

Confirming the new development, the scientists published in The Lancet Digital Health journal where they describe in more detail the diagnostic tool, its use and functionality.

Named the Epilepsy Diagnostic Companion (EDC), the tool is a predictive model and multilingual app used to identify convulsive seizures. Healthcare workers can use the tool to conduct simple screening on individuals as it enables earlier and more accurate diagnoses and leads to improved care and de-stigmatization of people with convulsive epilepsy. 

Ifakara Health Institute’s Dr. Honorati Masanja contributed to the study in collaboration with colleagues from Kenya, Uganda, Ghana, South Africa, the Netherlands and the UK including the Epilepsy Pathway Innovation in Africa (EPInA) study group. The lead authors are Gabriel Davis Jones and Arjune Sen from Oxford Epilepsy Research Group, UK.

“Using machine-learning techniques, we have developed and tested a region-specific questionnaire panel and predictive model to identify people who have had a convulsive seizure. We have developed a predictive model and free clinical app to screen people who present with potential symptoms of convulsive epilepsy.”

“The EDC provides a powerful tool to help diagnose convulsive epilepsy in resource-poor settings. The aim is not to replace health professionals, but rather aid in directing the individual to see the most appropriate clinician,” the scientists added.

To develop the predictive model and app, the scientists used data from the Studies of the Epidemiology of Epilepsy in Demographic Sites (SEEDS) database in Kenya, Uganda, Ghana, Tanzania, and South Africa. They analyzed epilepsy-specific data from 4,097 people of whom 48·5% had convulsive epilepsy.

The scientists applied a three-stage screening process to identify cases. The stages involved interviewing heads of households in each dwelling about convulsions, administering detailed questionnaires to individuals who screened positive in the first stage and finally assessing those who tested positive in the second stage with clinicians making a final diagnosis of epilepsy.

From the collected information, the scientists identified eight binary features to predict convulsive seizures. They then assessed several machine-learning algorithms to create a multivariate prediction model and validated the best-performing model with the internal dataset and a prospectively collected external validation dataset. 

Finally, the scientists used these features to develop a questionnaire-based predictive panel that they implemented into a multilingual app – the EDC – for healthcare workers in each geographical region.

“Although additional work is required, we hope that the app might help dispel some of the stigmatization of epilepsy by providing estimated outputs confirming an organic diagnosis and the opportunity for further education.”

According to the scientists, the identification of people who might have convulsive epilepsy has several benefits such as reducing the risk in individuals by enabling appropriate safety education and mitigation of potential provoking factors of seizures and increasing the appropriateness of onward referral streamlining the care pathway which can be cost saving at a societal level.

Link to publication: https://pubmed.ncbi.nlm.nih.gov/36963908/