Estimation of Spatial Autoregressive Model to Demonstrate the Spatial Dependence of Cancer Patients

Recent years have seen a rise in interest among statisticians in spatial data analysis, which is unsurprising given the detrimental effects on results and information loss that can occur from ignoring the spatial dimension in statistical analyses. Because the components of the phenomenon under study are spatially dependent, researchers have utilized spatial regression models to examine the impact of the explanatory variables on the dependent variable. The simulation method is used when challenges or difficulties are complex to address numerically. It entails creating a system for the actual model and then performing experiments on this model. Consequently, the researchers estimated the spatial autoregressive model (SAR) using the maximum likelihood (MLE) method. The SAR included the researchers' proposed weight matrix, which was built using the Rock adjacency criterion and the Euclidean distance, as well as a modified spatial weight matrix built using the Rock adjacency criterion. The suggested weight matrix was appropriate for model estimation after comparing the spatial weight matrices using the mean absolute relative error (MAPE) criterion. By utilizing the maximum likelihood approach (MLE) with the modified spatial weight matrix and the proposed spatial weight matrix, we were able to examine the relationship between the dependent variable, which represents the number of patients in each governorate, and the selected explanatory variables, which include tumor size, age average, and the number of areas contaminated with uranium in the governorate. This study utilized the spatial model to examine real-world cancer data. Based on the pollution levels caused by uranium and other pollutants from oil refineries and other industries, the governorates of Basra and Baghdad had the highest number of cancer patients, indicating a clear spatial dependence between the location of the governorate and the increase in cancer cases.