Figure 1.

The flow diagram of classification and prediction. A total of 70 HCs and 70 MWOA were used to construct the classification model and the top 3,650 discriminative features could discriminate MWoA with the highest accuracy. Then, the 3,650 discriminative features were used to construct predicting model and 70 predictive features could predict the reduction of migraine attack frequency with the highest correlation coefficient. HCs, healthy controls; MWOA, migraine without aura; LOOCV, leave-one-out cross-validation; SVM, support vector machine; SVR, support vector regression.