Skip to main content

View full-text article in PMC

. 2019 Jan 16;16(2):250. doi: 10.3390/ijerph16020250

Table 2.

Comparison of the area under the curve (AUC) when using only images and when adding demographic variables.

Models	Training				Test
	AUC			p Value	AUC			p Value
	I-CNN *	D-CNN **	Difference	p Value	I-CNN *	D-CNN **	Difference	p Value
VGG19	0.9570	0.9714	0.0144	<0.001	0.9075	0.9213	0.0138	0.049
InceptionV3	0.9523	0.9616	0.0093	0.014	0.8821	0.9045	0.0224	0.033
ResNet50	0.9219	0.9250	0.0031	0.434	0.8780	0.8955	0.0175	0.051
DenseNet121	0.9315	0.9472	0.0157	0.002	0.8605	0.8893	0.0288	0.011
InceptionResNetV2	0.9482	0.9455	0.0027	0.407	0.8851	0.8864	0.0013	0.888

* I-CNN: convolutional neural network only using images; ** D-CNN: I-CNN with demographic variables added.