Forecasting the prevalence of epilepsy in low- and middle-income countries to 2050 using a hybrid deep neural network-transformer modeling framework: insights from the Global Burden of Disease Study 2023.

Background: Epilepsy is a major public health challenge affecting individuals of all ages, especially in low- and middle-income countries (LMICs). Reliable prevalence projections are critical for healthcare planning and resource allocation. This study aimed to forecast the prevalence of epilepsy and its trends in LMICs by age, sex, year, and income level by 2050. Methods: Using data from the Global Burden of Disease Study (GBD) 2023, we projected the prevalence and number of idiopathic and secondary epilepsy cases in LMICs from 2024 to 2050. We developed a hybrid deep neural network (DNN)-Transformer framework that integrates Poisson regression and Autoregressive Integrated Moving Average (ARIMA) models for prevalence projection. Decomposition analysis was applied to quantify the contributions of population growth, aging, and prevalence change to the increase in epilepsy cases. Dementia-attributable epilepsy was independently projected to address secondary causes not included in GBD 2023. Results: By 2050, the age-standardized prevalence rate (ASPR) of epilepsy in LMICs was projected to reach 907.22 per 100,000 [95% uncertainty interval (UI) 731.01-1083.56], a 33.32% increase from 2023, with cases rising to 72.04 million (95% UI 57.86-86.23), a 58.68% increase. The ASPRs of idiopathic and secondary epilepsy were estimated at 323.11 and 584.10 per 100,000 in 2050, respectively, with the increase in secondary epilepsy being more than 7-fold that of idiopathic epilepsy since 2023. The ASPR of secondary epilepsy due to neonatal disorders was projected to rise by 65.76%. Model validation demonstrated good predictive performance (root mean squared error <0.001). From 2023 to 2050, the increases in idiopathic and secondary epilepsy cases were forecast to be highest in low-income countries (LICs; 76.12% and 241.50%, respectively), with growth declining as income levels increased. Population growth (21.40%) primarily drove the increase in idiopathic epilepsy cases, whereas changes in prevalence (59.89%) predominantly drove the rise in secondary epilepsy cases. Dementia-attributable secondary epilepsy was projected to reach 3.40 million cases by 2050. Conclusions: We forecast a continuous increase in the prevalence and number of epilepsy cases in LMICs through 2050, with secondary epilepsy increasing more rapidly than idiopathic epilepsy. LICs may exhibit the greatest increases over the next three decades, necessitating targeted interventions and further investigation.