Predictive Value of Blood Markers in Differentiating Healthy and Gbm Cases Using Artificial Neural Network.

Introduction: Glioblastoma multiform is the most common primary brain tumor that releases inflammatory mediators and changes in blood cells. This study answers the question of whether GBM patients can be distinguished from healthy individuals using changes in blood markers. Methods: This study was performed retrospectively using the records of newly diagnosed GBM patients and healthy individuals between February 2016 and February 2019. Blood markers including RBCs, WBCs, neutrophils, lymphocytes, monocytes, platelets, NLR, LMR, PLR, dNLR are examined in two groups of tumor patients and healthy individuals. Changes in blood markers were examined separately by statistical analysis and an artificial neural network (ANN). Statistically significant differences between blood markers in the two groups are investigated and the correlation between blood markers and GBM is determined. In this study, blood markers were specifically studied by artificial neural network. The accuracy of the artificial neural network in differentiating GBM and healthy individuals are determined. Results: In this study, 124 GBM patients and 124 healthy individuals were studied. In the statistical analysis, except for monocytes, other blood markers were significantly different in the two groups. WBC, neutrophils, NLR, PLR and dNLR were positively correlated with a GBM tumor. RBCs, lymphocytes, platelets, and LMR had a negative correlation with the presence of the tumor. The best predictive value among blood markers was related to dNLR (AUC=0.796) In the data analysis using an artificial neural network, the accuracy of differentiation between healthy and sick patients was 84.2%. Based on the characteristics of the artificial neural network used in this study, the placement accuracy and the predictive value of the artificial neural network in differentiating healthy and GBM cases were appropriate (MCC = 0.685, AUC = 0.875). Conclusion: All blood markers except monocytes are significantly different in healthy and GBM cases. The artificial neural network can differentiate between the two groups with appropriate accuracy and can be used clinically in predicting the presence of GBM. [ABSTRACT FROM AUTHOR]