Artificial Neural Network for miRNA and BMD Integration

For BMD-correlated EIs, the expression values of miR-194-5p and the bone mass parameter BMD were normalized into a 0 to 1 number before further model integration was performed as previously described^{32 (link)}. The Intelligent Problem Solver (IPS) tool in the software STATISTICA Neural Networks (SNN, Release 4.0E; Statsoft, Tulsa, OK, USA) was applied to construct a RBF-ANN model to investigate the effect of miRNA and EIs integration on the spine BMD association. The RBF-ANN model was named as ANN I in this study. The holdout cross-validation method was applied for preliminary validation of the model, as IPS randomly divided all the participants into three sets (training set, verification set, and testing set) in a 2:1:1 ratio. Thus, one-quarter of the participants was not included in model building and was used for model testing. The IPS calculated correlation coefficients for the training set (R_Tr) and the testing set (R_Te). The two correlation coefficients measured the correlation between model output and spine BMD. Similar values of R_Tr and R_Te indicate good generalization ability of the model. For model type comparison, the SPSS statistical software version 19.0 (IBM Corp., New York city, NY, USA) was applied to integrate miR-194-5p and BMD-correlated EIs into a MLR model.