Table 2 Predictive performance for HER2 state of single data source models based on a variety of classifiers, in the training, the validation, and the test set.

From: Intra- and peritumoral radiomics features based on multicenter automatic breast volume scanner for noninvasive and preoperative prediction of HER2 status in breast cancer: a model ensemble research

Single data source models	Classifiers	Data sets	AUC (95% CI)	Cutoff	Sensitivity	Specificity
Tumor	RFC	Training	0.973 (0.954, 0.922)	0.4966	1.000	0.832
		Validation	0.598 (0.393, 0.804)	0.4258	0.545	0.250
		Test	0.559 (0.443, 0.675)	0.3001	0.694	0.548
	ETC	Training	0.977 (0.973, 0.985)	0.4755	1.000	0.957
		Validation	0.555 (0.371, 0.739)	0.4055	1.000	0.292
		Test	0.516 (0.392.0.640)	0.3497	1.000	0.016
	ABC	Training	1.000 (1.000, 1.000)	0.5024	1.000	0.000
		Validation	0.549 (0.336, 0.763)	0.4841	0.636	0.625
		Test	0.531 (0.410, 0.652)	0.4265	0.944	0.177
Clinical	ETC	Training	1.000 (1.000, 1.000)	0.5996	1.000	0.000
		Validation	0.803 (0.660, 0.947)	0.3358	1.000	0.625
		Test	0.695 (0.583, 0.807)	0.2803	0.861	0.419
	RFC	Training	1.000 (1.000, 1.000)	0.5706	1.000	0.000
		Validation	0.754 (0.595, 0.913)	0.4457	0.909	0.667
		Test	0.705 (0.601, 0.809)	0.3522	0.889	0.468
	LGBM	Training	1.000 (1.000, 1.000)	0.5302	1.000	0.000
		Validation	0.667 (0.479, 0.855)	0.0608	0.909	0.417
		Test	0.722 (0.613, 0.831)	0.0373	0.972	0.258
R5mm	GBC	Training	1.000 (1.000, 1.000)	0.5003	1.000	0.000
		Validation	0.625 (0.430, 0.820)	0.0058	1.000	0.250
		Test	0.465 (0.346, 0.585)	0.000	1.000	0.000
	ETC	Training	1.000 (1.000, 1.000)	0.5667	1.000	0.000
		Validation	0.625 (0.409, 0.841)	0.1939	1.000	0.125
		Test	0.472 (0.352, 0.593)	0.0582	1.000	0.032
	ABC	Training	1.000 (1.000, 1.000)	0.5046	1.000	0.000
		Validation	0.606 (0.372, 0.840)	0.4815	0.727	0.500
		Test	0.472 (0.349, 0.596)	0.3547	1.000	0.016
R3mm	LGBM	Training	1.000 (1.000, 1.000)	0.5293	1.000	0.000
		Validation	0.619 (0.393, 0.845)	0.4159	0.545	0.833
		Test	0.487 (0.368, 0.606)	0.000	1.000	0.000
	RFC	Training	1.000 (1.000, 1.000)	0.5064	1.000	0.000
		Validation	0.614 (0.414, 0.813)	0.3730	1.000	0.292
		Test	0.494 (0.374, 0.614)	0.1290	1.000	0.048
	ABC	Training	0.998 (0.996, 1.000)	0.4945	1.000	0.968
		Validation	0.606 (0.422, 0.790)	0.4707	1.000	0.250
		Test	0.483 (0.368, 0.599)	0.4088	1.000	0.032

Bold characters represented the classifier with the highest sum of AUCs in the validation set and the test set.
R3mm, model based on peritumoral 3 mm ring of breast tumor; R5mm, model based on peritumoral 5 mm ring of breast tumor; Tumor, model based on radiomics features of the tumor; ABC, Ada Boosting Classifier; AUC, the area under the curve; CI, confidence interval; Clinical, model based on clinical, ABVS, and serology features of breast tumor; ETC, Extra Tree Classifier; GBC, Gradient Boosting Classifier; LGBM, Light Gradient Boosting Machine; RFC, Random Forest Classifier.

Back to article page

Table 2 Predictive performance for HER2 state of single data source models based on a variety of classifiers, in the training, the validation, and the test set.

Search

Quick links