Life expectancy and lifespan disparity forecasting: a long short-term memory approach

Abstract

After the World War II, developed countries experienced a constant decline in mortality. As a result, life expectancy has never stopped increasing, despite an evident deceleration in developed countries, e.g. England, USA and Denmark. In this paper, we propose a new approach for forecasting life expectancy and lifespan disparity based on the recurrent neural networks with a long short-term memory. This type of neural network leads to predicting future values of longevity indexes while maintaining the significant influence of the past trend, but at the same time adequately reproducing the recent trend into forecasting. The model was applied to five countries for two fitting periods focusing on the forecasting life expectancy and lifespan disparity, both independently and simultaneously at birth and age 65. The results were compared to the projections obtained by four different models, namely, the Double Gap, ARIMA, CoDa and Lee-Carter in the independent case and the first-order Vector Autoregression model in the simultaneous case. Our predictions seem to be coherent with historical trends and biologically reasonable, providing a more accurate portrait of the future life expectancy and lifespan disparity.