Predicting the antioxidant status of human plasma by artificial neural network analysis

Introduction and Objectives: Various biomarkers represent antioxidant condition in human plasma. Measuring all these biomarkers would be a time-consuming process. This study developed an artificial neural network (ANN) analysis to predict antioxidant status by a ranked importance of biochemical parameters in human plasma. Materials and Methods: Blood samples with normal hematological and biochemical parameters were collected. Ferric reducing ability of plasma (FRAP) was measured as antioxidant assay of blood plasma. Hb oxidized derivatives and protein carbonyl (PCO) levels in plasma and erythrocytes were considered as oxidative markers in blood samples. ANN analysis was developed as multilayer feed forward architecture for predicting the parameters which correlated with antioxidant power of plasma. The best model was performed by multilayer perceptron method (MLP) with hyperbolic tangent and identity activation functions for hidden and output layers, respectively. Oxidant and antioxidant parameters were ranked based on the calculated importance. Results: Our results showed that BUN, creatinine, uric acid, oxyHb, and Hb absorbance from 420 to 560 nm were the important parameters with the normalized importance more than 50%. Conclusion: The results of this study demonstrated the ability of ANN analysis to predict oxidative parameters in human plasma and erythrocytes. Identification of important parameters can eliminate less important parameters from the clinical procedures leading to a cheaper and faster diagnosis.

کلیدواژه ها:

Antioxidant status ، Artificial neural network ، Human plasma