Table 2.

Confusion matrix of neural network model trained and tested on only appointment records without missing information (mean and 90% confidence interval over 10-fold cross-validation are shown).

	Real shows	Real no-shows
Predicted shows	21441.0 [21405.5,21469.1]	1001.0 [969.9,1033.1]
Predicted no-shows	1172.8 [1148.7,1204.2]	4905.7 [4871.8,4937.4]
Confusion matrix (row sum ratio/confidence interval over 10-fold cross-validation).
Predicted shows	0.96 [0.95,0.96]	0.04 [0.04,0.05]
Predicted no-shows	0.19 [0.19,0.20]	0.81 [0.81,0.81]