Table 3 Classification results of predicting postintervention MCID participants using six machine learning methods with three types of datasets

Model

Clinical Data Set

fNIRS Data Set

AUC

ACC

Sen

Spc

AUC

ACC

Sen

Spc

SVM

0.802 (0.095)

0.735 (0.085)

0.634 (0.247)

0.766 (0.109)

0.828 (0.086)

0.765 (0.089)

0.716 (0.195)

0.781 (0.102)

RF

0.752 (0.122)

0.745 (0.088)

0.431 (0.221)

0.843 (0.098)

0.661 (0.120)

0.709 (0.094)

0.376 (0.213)

0.813 (0.106)

ANN

0.803 (0.120)

0.757 (0.079)

0.502 (0.216)

0.836 (0.094)

0.773 (0.111)

0.735 (0.082)

0.521 (0.197)

0.803 (0.101)

LR

0.787 (0.102)

0.729 (0.098)

0.642 (0.233)

0.756 (0.122)

0.778 (0.093)

0.722 (0.092)

0.634 (0.182)

0.751 (0.118)

KNN

0.728 (0.104)

0.633 (0.111)

0.752 (0.215)

0.596 (0.133)

0.716 (0.112)

0.634 (0.105)

0.754 (0.175)

0.596 (0.129)

EN

0.763 (0.104)

0.687 (0.096)

0.654 (0.231)

0.698 (0.114)

0.762 (0.105)

0.713 (0.085)

0.711 (0.151)

0.715 (0.109)

Model

Clinical-fNIRS Data Set

AUC

ACC

Sen

Spc

SVM

0.849 (0.078)

0.805 (0.069)

0.764 (0.221)

0.818 (0.088)

RF

0.696 (0.114)

0.704 (0.102)

0.454 (0.221)

0.782 (0.119)

ANN

0.861 (0.087)

0.805 (0.082)

0.786 (0.218)

0.812 (0.098)

LR

0.823 (0.088)

0.783 (0.078)

0.692 (0.200)

0.813 (0.095)

KNN

0.717 (0.115)

0.615 (0.097)

0.794 (0.187)

0.564 (0.126)

EN

0.799 (0.101)

0.729 (0.089)

0.722 (0.019)

0.732 (0.109)