Prognostic Modeling of Brain Tumor Survival

Abstract

Brain tumors, including gliomas and meningiomas, vary in prognosis depending on tumor class, location, volume, and patient-specific factors such as health, gender, and treatment. Using an open-source dataset from the Masaryk Memorial Cancer Institute in Brno, we analyze survival outcomes of patients treated with stereotactic radiotherapy between 2004 and 2011. Models considered include the Cox proportional hazards model, the exponential parametric model, the log-logistic model, and the random survival forest. Comparative results highlight differences in fit, predictive accuracy, and interpretability, offering guidance for biostatistical approaches to brain tumor survival analysis.