id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-001253-3jnkki5z	Mohammad, Fahim	Advantages and Limitations of Anticipating Laboratory Test Results from Regression- and Tree-Based Rules Derived from Electronic Health-Record Data	2014-04-14		.txt	text/plain	3850	178	55	We used four types of input-age, gender, diagnoses (three-digit ICD-9 codes), and results of laboratory tests on blood samples added to the record in the seven days before a given test was ordered-to build simple, robust models for whether the result of a test would be within the reference interval (''normal'') or outside of it in a given direction (''abnormal''), treating high and low results separately. To determine how well sendout and in-house test results can be anticipated based on basic information available in the medical record, we used two independent methods-generalized linear modeling (GLM) and classification and regression trees (CART)to build simple, robust test-result predictors and then evaluated the performance of these predictors according to the standard clinical metrics of positive predictive value (PPV) and negative predictive value (NPV), as well as sensitivity and specificity via the receiveroperator curve (ROC) area under the curve (AUC).	./cache/cord-001253-3jnkki5z.txt	./txt/cord-001253-3jnkki5z.txt