BUPNN: Manifold Learning Regularizer-Based Blood Usage Prediction Neural Network for Blood Centers

Figure 6

The scatter plot the prediction of BUPNN for COM-Mea schemes, the missing values are filled with the median value. The vertical coordinate is used to distinguish different samples, and the horizontal coordinate indicates the predicted value of the BUPNN model for the samples. A predicted value of less than 0.5 indicates that no blood is needed, and a value greater than 0.5 indicates that blood is needed. The color of the scatter indicates whether the prediction is correct or not.