Development and Validation of a Magnetic Resonance Imaging-Based Machine Learning Model for TMJ Pathologies

Table 4

Evaluation for diagnostic performance by four indicators for mandibular condyle: precision, recall, F1-score, and support in the training set. “1” indicates normal. “2” indicates degenerative joint diseases.

		Indicators	KNN	SVM	XGBoost	RF	LR	DT
Training	1	Precision	0.83	0.79	0.91	0.95	0.81	0.88
		Recall	0.92	1.00	0.98	1.00	0.98	0.98
		F1-score	0.88	0.88	0.95	0.97	0.89	0.94
		Support	53	53	53	53	53	53
	2	Precision	0.50	0.00	0.90	1.00	0.67	0.80
		Recall	0.29	0.00	0.64	0.79	0.14	0.74
		F1-score	0.36	0.00	0.75	0.88	0.24	0.76
		Support	14	14	14	14	14	14
Testing	1	Precision	0.81	0.78	0.81	0.78	0.78	0.70
		Recall	0.93	1.00	0.93	1.00	1.00	0.86
		F1-score	0.87	0.88	0.87	0.88	0.88	0.83
		Support	14	14	14	14	14	14
	2	Precision	0.50	0.00	0.50	0.00	0.00	0.33
		Recall	0.25	0.00	0.25	0.00	0.00	0.25
		F1-score	0.33	0.00	0.33	0.00	0.00	0.29
		Support	4	4	4	4	4	4

LR: logistic regression; RF: random forest; DT: decision tree; KNN: -nearest neighbors; XGBoost; SVM: support vector machine.