Discrimination of indoor versus outdoor environmental state with machine learning algorithms in myopia observational studies

Table 2 Location of the testing group A predicted by Model A, the dataset B predicted by Model A and the testing group B predicted by Model B

Model	Data sets	Real location	Predicted		Total
Model	Data sets	Real location	Outdoor	Indoor	Total
A	A	Outdoor	15,502	59^a	15,561
		Indoor	43	7282	7325
		Total	15,545	7341	22,886
A	B	Outdoor	9799	3788^b	13,587
		Indoor	291	9661	9952
		Total	10,090	13,449	23,539
B	B	Outdoor	4386	495^c	4881
		Indoor	39	2142	2181
		Total	4425	2637	7062

^a59 outdoor locations were mistaken as indoors, and 43 indoor locations were mistaken as outdoors Kappa = 0.990, p < 0.001
^b3788 outdoor locations were mistaken as indoors, and 291 indoor locations were mistaken as outdoors. Kappa = 0.692, p < 0.001
^c495 outdoor locations were mistaken as indoors, and 39 indoor locations were mistaken as outdoors. Kappa = 0.821, p < 0.001

ISSN: 1479-5876