A Generalized Exponential Regression Model for Predicting High-Grade Glioma Growth in Paediatric Patients

Abstract

High-grade gliomas (HGG) are invasive brain tumours characterized by abnormal growth patterns and poor prognoses. Aware and precise prediction of tumour growth helps improve both treatment protocols and patient medical outcome. The quick replicating and diverse nature of HGGs in children makes their predictive progression highly difficult to determine. This research utilized a generalized exponential regression approach to study glioma progression in children's brains with predicted accuracy reaching 73.68% for all tumours but elevating to 77.7% for small tumours under 100 mm³. Statistical analyses revealed significant negative correlations between tumour growth and tumour size, along with pre-radiotherapy performance status (PS Before RT), as determined by Kendall’s Tau test.  The Mann-Whitney U and Kruskal-Wallis H tests were employed for bivariate analysis of categorical data, demonstrating a significant association (p < .05) among tumour growth rate, the extent of surgical resection, and survival status.  The child's age, the occurrence of headaches, and edema were independently associated with the progression of tumour growth. These findings enhance the understanding of paediatric HGGs development, facilitating more accurate prognostic evaluations and improving personalized treatment strategies.