Improving mortality forecasting: integrating the lee-carter model with neural networks / Hana Natasya Abd Hamid, Khairunnisa Khairul Hizam and Nurul Aityqah Yaacob

This research integrates the Lee-Carter (LC) model with neural network (NN) methods to enhance mortality forecasting. The LC model, widely used for demographic forecasting, has limitations in capturing complex and nonlinear mortality trends. To address these limitations, we incorporate NN methods, namely a multilayer feed-forward neural network (MFFNN), to identify patterns within mortality data. The study evaluates the performance of the LC and LC-NN models across five countries: Germany, Japan, Malaysia, South Korea, and the United States of America (USA). Findings indicate that the LC-NN model outperforms the LC model, as demonstrated by lower Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) values. This integration significantly improves forecasting accuracy, providing more reliable insights into future mortality trends. The results have significant implications for public health planning and policymaking, offering a robust tool for forecasting demographic changes and their impact on healthcare systems. This integration advances mortality forecasting, leading to better public health outcomes.