استخدام نموذج كوكس لتحليل العوامل المؤثرة على وقت البقاء للمرضي المصابين بسرطان الدم في مرحلة الطفولة دراسة حالة على المعهد القومي الأورام بمدينة مصراته من 2017 إلى 2023

The Cox regression model is a vital tool in survival data analysis, particularly for severe diseases like childhood leukemia. It enables researchers to estimate the effect of independent variables on survival time without needing to specify a distribution, allowing doctors and researchers to develop more effective treatment strategies and improve clinical outcomes. This study aims to analyze factors affecting the survival duration of children with leukemia using the Cox regression method and estimating the survival function using the Kaplan-Meier method. The study also seeks to determine the relationship between this type of cancer and various personal and clinical variables, as well as relevant medical analyses. The chi-square test was used as a statistical tool to verify these associations, and the log-rank test was employed to identify the different factors with significant effects. In this study, a multiple linear regression model was developed to predict the probability of patient death based on recorded medical characteristics and analyses for each case. The necessary data was collected from 50 recorded cases in the pediatric oncology department of the National Cancer Institute in Misurata during the period from 2017 to 2023. The survival analysis results showed that the survival duration of children with type B cancer was approximately 929 days, the longest compared to type M3. Other types did not record any deaths. The results also indicated that children diagnosed using Flow-cytometry had a significantly higher survival chance compared to those diagnosed using BMB+IHC, highlighting the importance of the diagnostic method in improving survival chances. Furthermore, the survival probability was much higher for children who did not experience disease recurrence or those who did not suffer from bleeding or skin rashes, while the survival probability was significantly lower for children with below-normal PLT levels. The multiple variable linear regression model showed that the most influential factors in patient mortality risk included the diagnostic method, bleeding, skin rash presence, disease recurrence, and PLT levels, reflecting the importance of studying these factors to gain a deeper understanding of their impact on the survival of children with leukemia. Keywords: Cox regression model, Survival function, Kaplan-Meier, Log-rank test