Empowering Epilepsy Diagnosis in Sub-Saharan Africa: A Machine Learning Breakthrough

Diagnosing epilepsy in low-resource settings has long been a challenge. With few neurologists and limited access to advanced medical tools, millions go undiagnosed, facing unnecessary risks and stigma. Our recent study, published in *The Lancet Digital Health*, introduces a practical solution: the Epilepsy Diagnostic Companion (EDC), a smartphone app powered by machine learning to identify convulsive epilepsy. This tool is tailored to the unique cultural and clinical contexts of sub-Saharan Africa, where the burden of epilepsy is disproportionately high.

Using data from over 4,000 individuals across five African countries, we developed a predictive model that relies on just eight simple questions. These questions address key features of convulsive seizures, such as tongue biting and loss of consciousness, and were carefully selected for their clinical relevance and cultural appropriateness. The app demonstrated high sensitivity and specificity during testing, outperforming traditional diagnostic approaches. Unlike one-size-fits-all tools, the EDC is designed specifically for healthcare workers in low-resource settings, requiring minimal training to use effectively.

The impact of this app could be transformative. By bridging the diagnostic gap, the EDC enables earlier, more accurate identification of epilepsy, empowering community healthcare workers to provide timely education, safety advice, and referrals. This approach not only improves patient outcomes but also reduces the societal stigma surrounding epilepsy. With plans for iterative updates and expanded features, the EDC is a powerful example of how technology can address global health challenges, offering hope to millions in need.

Read the Lancet Digital Health article here.