Modelling and Forecasting Age-Specific Cancers Mortality Rate Using Lee-Carter Model

Abstract

In 1992, Lee and Carter proposed a method which combines demography and stochastic to model and forecast the mortality rates, which became the reference and a leading statistical model. In this study we identified Cancer as characterized by out-of-control cell growth and it is the second leading cause of death after Ischemic heart disease. The problem of this study is that the cancer has the highest death rate among other diseases and it's treatment required financial resources that strain the state treasury, more over the absence of data reduced the performance of the model . The importance of this study is to help the governments ,voluntary organizations and health sector to make plans and researches scientifically. According to this importance the aims of this study is to use original Lee-Carter model to model and forecast age-specific cancer mortality rate for three types of cancer (Oral, Lung and Colon ) for period 2015 to 2020. The model's parameters estimated by Singular value Decomposition (SVD) and Maximum Likelihood Estimation (MLE), and used Auto Regressive Integrated Moving Average (ARIMA) Random Walk with drift (0,1,0) to forecast mortality index for Egyptian male and female based on five-year data aggregation that obtained from World Health Organization (WHO) for the period 2001-2014. The results obtained by using different statistic packages R ,ilc, Demography and forecast packages. Our findings showed that the SVD is better for male with error (ME=0.00016, MSE=25208 ), while for female the SVD is better with error (ME=0.02856, MSE=0.32310) for oral cancer. while the MLE is better for male with error (ME=0.00714, MSE=0.12385), and the SVD is better for female with error (ME=0.00523, MSE=0.08022) for lung cancer. while the MLE is better for male with error (ME=0.00506, MSE=0.11065), and the SVD is better for female with error (ME=-0.00401, MSE=0.13561) for colon cancer. Also the results showed that the lung cancer has highest mortality rate and it is 76.27 per 100.000 in year 2020 in age-group (70-74) for male then colon and it is 27.91 in year 2020 in age-group (70-74) for male, after that oral cancer rate and it is 3.11 per 100.000 in year 2015 in age-group (70-74) for male. The study came out with numbers of recommendations from them the most importance are to apply Lee-Carter method to modeling and forecasting age-specific mortality rate and SVD to estimate the model's parameters, and to have care and accuracy when registering data. Health sector must make plans and programs to reduce the cancer mortality rate especially for male.