Lack of survival advantage among re-resected elderly glioblastoma patients: a SEER-Medicare study

Survival prediction of glioblastoma patients using modern deep learning and machine learning techniques

In this study, we utilized data from the Surveillance, Epidemiology, and End Results (SEER) database to predict the glioblastoma patients' survival outcomes. To assess dataset skewness and detect feature importance, we applied Pearson's second coefficient test of skewness and the Ordinary Least Squares method, respectively. Using two sampling strategies, holdout and five-fold cross-validation, we developed five machine learning (ML) models alongside a feed-forward deep neural network (DNN) for the multiclass classification and regression prediction of glioblastoma patient survival. After balancing the classification and regression datasets, we obtained 46,340 and 28,573 samples, respectively. Shapley additive explanations (SHAP) were then used to explain the decision-making process of the best model. In both classification and regression tasks, as well as across holdout and cross-validation sampling strategies, the DNN consistently outperformed the ML models. Notably, the accuracy were 90.25% and 90.22% for holdout and five-fold cross-validation, respectively, while the corresponding R2 values were 0.6565 and 0.6622. SHAP analysis revealed the importance of age at diagnosis as the most influential feature in the DNN's survival predictions. These findings suggest that the DNN holds promise as a practical auxiliary tool for clinicians, aiding them in optimal decision-making concerning the treatment and care trajectories for glioblastoma patients.

Prognostic nomogram models for predicting survival probability in elderly glioblastoma patients

PURPOSE

To investigate the prognostic factors of survival and develop a predictive nomogram model for elderly GBM patients.

METHODS

Elderly patients (> = 65 years) with histologically diagnosed GBM were extracted from the SEER database. Survival analysis of overall survival (OS) was performed by Kaplan-Meier analysis. Univariate and multivariate Cox regression analyses were used to determine independent prognostic factors and these factors were used to further construct the nomogram model.

RESULTS

A total of 9068 elderly GBM patients (5122 males and 3946 females) were included, with a median age of 72 years (65-96 years). All patients were divided randomly into the training group (n = 6044) and the validation group (n = 3024) by a ratio of 2:1. Cox regression analyses on OS showed eight independent prognostic factors (race, age, tumor side, tumor size, metastasis, surgery, radiotherapy, and chemotherapy) in the training cohort. Also, seven variables (except for race) were identified on CSS in the training group. By comprising these variables, the nomogram models on OS and CSS for predicting the 6-month, 1-year, and 2-year survival probability were constructed and exhibited moderate consistency, respectively. Then, they could be validated well in the validation cohort and by C-index, time-dependent ROC curve, calibration plot, and DCA curve.

CONCLUSIONS

Nomogram models on OS and CSS could provide an applicable tool to predict the survival probability and provide clinical references regarding treatment strategies and prognosis.