Forecasting regional trends and burden of diabetes mellitus in Ethiopia using deep learning

Diabetes mellitus, a chronic disease, is a major public health concern impacting both industrialized and developing nations. Hence this study presents trends, and forecasting of diabetes mellitus for 10 years and associated factors across region in Ethiopia a deep learning analysis.

This study explores data from the Global Burden of Diseases 2021 study, during the period from 1990 to 2021.The analysis focuses forecasting of age-standardized incidence, mortality and disability-adjusted life years for diabetes mellitus. A Long Short-Term Memory (LSTM) model is used for predictions of future trends from 2022 to 2031, quantified based on the estimated annual percentage change.

In Ethiopia, the overall estimated incidence was 7464 cases (95% UI: 5835–9237), with a higher burden in males (5022 cases) compared to females (2442 cases). The ASIR was also higher in males (1092 per 100,000) than in females (843 per 100,000), reflecting a significant gender disparity in disease burden. Death rates decreased significantly (-2.74% annually), with sharper declines among males (-3.06%) than females (-2.22%). The prediction of DALYs and YLLs after 2025–2028, diabetes-related deaths and years lived with disability are projected to rise sharply through 2031. The forecast predicts a sharp increase in deaths, reaching over 55,000 by 2031, indicating a concerning upward trend in diabetes-related mortality.

This study provides critical insights into Ethiopia diabetes burden, future predictions of DM, revealing persistent increase in prevalence and mortality despite slight decline in incidence. This analysis utilized RNN and LSTM models to forecast incidence, prevalence and disability in Ethiopia. The application of LSTM model showed superior predictive accuracy over traditional methods. Key findings highlight the influence of risk factors, aging populations and regional disparities in healthcare access.