TITLE:
Integrative Machine Learning Analysis Identifies a Robust Mitoxyperiosis-Related Prognostic Signature in Glioma
AUTHORS:
Zexing Xie, Xinkai Dou, Jiaze Lu, Xiangyin Liu, Zijian Zhou
KEYWORDS:
Mitoxyperiosis, Glioma, Machine Learning, Prognostic Signature, Integrative Analysis
JOURNAL NAME:
Journal of Biosciences and Medicines,
Vol.14 No.1,
January
23,
2026
ABSTRACT: Background: Glioma is the most common malignancy of the central nervous system with an extremely poor prognosis. Mitoxyperiosis, a recently defined mode of programmed cell death, has unclear prognostic value and molecular landscape in glioma. Methods: We integrated transcriptomic and survival data from the CGGA-657 (training set) and CGGA-313 (validation set) cohorts. Biological interactions among 58 candidate genes were revealed through protein-protein interaction (PPI) network analysis. An integrative machine learning framework comprising 10 algorithms was utilized to identify the optimal prognostic model after exhaustively screening 117 combinations. Results: An optimal risk model consisting of 15 genes was identified using StepCox[forward] + Enet[α = 0.1]. This model demonstrated excellent predictive accuracy in both training and validation sets (3-year AUC: 0.832 and 0.849, respectively), with Kaplan-Meier curves showing significant prognostic differences between risk groups (p Conclusion: This study developed and validated an integrative machine learning model based on mitoxyperiosis-related genes. The model provides high accuracy and clinical potential for predicting glioma prognosis, offering a novel biomarker for precision medicine.