A prediction algorithm method is developed by entering the characteristic features of the PPG signal into machine learning algorithms in order to measure the hemoglobin value non-invasively.

We have used the PPG signals in the prediction of hemoglobin levels.

As the machine learning algorithms, CART (classification and regression trees), LSR (least squares regression), GLR (generalized linear regression), MVLR (multivariate linear regression), PSLR (partial least squares regression), GRNN (generalized regression neural network), MLP (multilayer perceptron) and SVR (support vector regression) are used.

The methods of RELIEFF feature selection (RFS) and correlation-based feature selection (CFS) are used to decrease the number of features.