Multistate health transition modeling using neural networks

This article proposes a new model that combines neural network with generalized linear (GLM) to estimate and predict health transition intensities. We introduce networks modeling incorporate socioeconomic lifestyle factors allow for nonlinear links between these variables. use transfer learning link the models different transitions improve estimation limited data. apply individual-level data from Chinese Longitudinal Healthy Longevity Survey 1998 2018. The results show our performs better in prediction than standalone GLM models. provide estimates of life expectancies range population subgroups. also describe wide possible applications further health-related research, including risk using claim mortality based on