Validity of Machine Learning in Detecting Complicated Appendicitis in a Resource-Limited Setting: Findings from Vietnam

Table 3

Characteristics of machine learning models to classify complicated appendicitis.

Model and parameters used	Description	Imbalance unadjusted data		Imbalance adjusted data
		AUC	Accuracy	AUC	Accuracy
		SVM
Linear	Liner function	0.711	0.752	0.730	0.655
Gaussian	Radial basis function	0.699	0.798	0.754	0.666
Sigmoid	Sigmoid function	0.627	0.726	0.637	0.596
Polynomial	Polynomial function	0.728	0.805	0.791	0.726
DT
Gini	Gini impurity	0.574	0.689	0.719	0.719
Entropy	Information gain - entropy	0.609	0.697	0.758	0.758
Logistic
NCG	Newton -CG	0.711	0.805	0.737	0.675
LBFGS	Limited-memory Broyden-Fletcher-Goldfarb-Shanno	0.711	0.805	0.737	0.675
Lib	Liblinear	0.714	0.803	0.789	0.729
SAG	Stochastic average gradient descent	0.711	0.805	0.737	0.675
SAGA	Stochastic average gradient accelerated	0.714	0.802	0.789	0.726
KNN
Uniform	Uniform weights	0.680	0.779	0.815	0.734
Distance	Distance weights	0.672	0.776	0.831	0.741
ANN
GD_a	Gradient descent, identity	0.732	0.807	0.809	0.743
GD_b	Gradient descent, logistic	0.713	0.802	0.803	0.737
GD_c	Gradient descent, tanh	0.734	0.805	0.81	0.742
GD_d	Gradient descent, ReLU	0.748	0.807	0.826	0.766
SGD_a	Stochastic gradient descent, identity	0.692	0.776	0.690	0.632
SGD_b	Stochastic gradient descent, logistic	0.681	0.752	0.614	0.535
SGD_c	Stochastic gradient descent, tanh	0.695	0.774	0.691	0.633
SGD_d	Stochastic gradient descent, ReLU	0.688	0.752	0.689	0.631
LBFGS_a	Limited-memory BFGS, identity	0.744	0.802	0.811	0.750
LBFGS_b	Limited-memory BFGS, logistic	0.641	0.687	0.820	0.755
LBFGS_c	Limited-memory BFGS, tanh	0.644	0.704	0.853	0.776
LBFGS_d	Limited-memory BFGS, ReLU	0.707	0.776	0.838	0.754
GB
GB1	Friedman MSE, sqrt, deviance	0.741	0.812	0.887	0.823
GB2	Friedman MSE, log, deviance	0.753	0.810	0.890	0.820
GB3	Friedman MSE, deviance	0.746	0.802	0.887	0.825
GB4	Friedman MSE, sqrt, AdaBoost	0.771	0.815	0.888	0.818
GB5	Friedman MSE, log, AdaBoost	0.750	0.810	0.894	0.821
GB6	Friedman MSE, AdaBoost	0.756	0.810	0.887	0.820
GB7	MSE, sqrt, deviance	0.745	0.807	0.89	0.821
GB8	MSE, log, deviance	0.743	0.810	0.886	0.814
GB9	MSE, deviance	0.740	0.803	0.887	0.824
GB10	MSE, sqrt, AdaBoost	0.752	0.809	0.889	0.808
GB11	MSE, log, AdaBoost	0.745	0.812	0.89	0.818
GB12	MSE, AdaBoost	0.755	0.812	0.887	0.820