Mortality forecasting is important for population policy and public health, particularly in the context of pandemics and aging populations. This study addresses this need by proposing a novel hybrid modeling framework that combines generalized estimating equations based on quasi-likelihood method and autoregressive integrated moving average models to predict mortality rates across Organization for Economic Cooperation and Development countries. The suggested method provides for parameter estimate in correlated panel data, notably longitudinal data, and precise temporal prediction of important predictors. The study, which examined the drivers of particular mortality, non-communicable diseases, accidents and other external causes of death, employed a large dataset covering 33 years from 1980 to 2013. The prediction performance of the model was high with an average determination coefficient R² of 0.928 across countries, indicating an excellent model fit. The approach was shown to be reliable out of sample with forecasts produced for the years 2014 and 2023. The methodology offers a statistically interpretable alternative to black-box machine learning methods, and fills a gap in the literature where generalized estimating equations and autoregressive integrated moving average models have been utilized mostly in isolation.
In this research we combine the two methodologies to provide a clear and powerful method for predicting mortality rates. It may be a useful tool for governments, insurance companies, and healthcare planners to prepare for demographic shift. This can allow people to make informed decisions and prepare for what lies ahead.