A trend-residual decomposition-based modeling approach for Türkiye's total healthcare expenditure forecasting

Abstract

Accurate forecasting of health expenditures is a fundamental issue for the sustainability of health systems and policies. In this study, a trend -residual decomposition -based model is proposed to forecast T & uuml;rkiye's total healthcare expenditure with high accuracy. The proposed model has a two -stage forecasting procedure. In the first stage, the trend of the health expenditure time series is determined using polynomial regression. In the second stage, a residual model with optimized linear parameters by least squares estimation method and non-linear parameters by neural network algorithm is proposed to model the detrending part of the time series. The performance of the proposed model using healthcare expenditure data for the years 1999-2021 are compared with grey models, regression models, exponential smoothing models and ARIMA models. The results obtained by using the years 1999-2017 for training and the years 2018-2021 for test demonstrate that the proposed model has better modeling and forecasting performance than other models. Therefore, T & uuml;rkiye's total healthcare expenditure for the years 2022-2030 has been forecasted with the proposed model and it is predicted that it will reach 2.2 trillion TL in 2030.