Table 2. Area under the ROC curve obtained in the disease classification experiment using patient data represented with original descriptors (“RawFeat”) and pre-processed by principal component analysis (“PCA”) and three-layer stacked denoising autoencoders (“DeepPatient”).

Time Interval = 1 year (76,214 patients)
Disease	Area under the ROC curve
	RawFeat	PCA	DeepPatient
Diabetes mellitus with complications	0.794	0.861	0.907
Cancer of rectum and anus	0.863	0.821	0.887
Cancer of liver and intrahepatic bile duct	0.830	0.867	0.886
Regional enteritis and ulcerative colitis	0.814	0.843	0.870
Congestive heart failure (non-hypertensive)	0.808	0.808	0.865
Attention-deficit and disruptive behavior disorders	0.730	0.797	0.863
Cancer of prostate	0.692	0.820	0.859
Schizophrenia	0.791	0.788	0.853
Multiple myeloma	0.783	0.739	0.849
Acute myocardial infarction	0.771	0.775	0.847