The using of COX regression and artificial neural networks in the survival analysis

The Cox Regression Model is the most commonly used regression model in survival analysis. In this study, Cox regression and artificial neural networks were introduced for survival analysis, and these models were exemplified using data from 1168 leukemia patients in the Republic of Yemen between January 2017 and February 2022. The aim of this study is to identify the most significant factors (risk factors) influencing the survival times of leukemia patients in the Republic of Yemen, dealing with one of the most common diseases (cancer). Additionally, the study compares the performance indicators between the Artificial Neural Networks model and the Cox Regression model in survival analysis. To achieve this analysis, the Cox regression model was used as a classic method in analyzing survival data in addition to artificial neural network as a modern method. The best model was chosen from the proposed Cox models and then the same data was analyzed using artificial neural networks. A comparison was made between the results of the Cox regression model and the results obtained from the use of artificial neural networks. The comparison between the two methods was carried out according to the criteria of the least mean squared error (MSE) and the mean absolute error (MAE). After selecting the best model from Cox regression models and the best model from the artificial neural network models and comparing them, the artificial neural networks outperformed the Cox regression model in analyzing survival data according to the previously mentioned criteria.