Info

Notes: sensitivity=0.9; specificity=0.95.

Notes: sensitivity=0.9; specificity=0.95.

number of cases of SARS that the case detection algorithm classified correctly (90) to the total number of patients with SARS (100), or 0.9. Sensitivity is the probability that the case detection algorithm will "detect" a patient with SARS. Specificity is the ratio of individuals without SARS that the case detection algorithm classified correctly as not having SARS (95) to the total number of individuals without SARS (100), or 0.95. The evalua-tor can also compute the ratio of individuals without SARS (5) who were incorrectly labeled by the case detection algorithms as SARS (5) over the total number of patients without SARS (100), or 0.05. This quantity is called the false positive rate (or false alarm rate), and it is the probability that a case detection algorithm will incorrectly classify a patient as SARS-infected. Note that specificity and false positive rate (also known as the false alarm rate) sum to one because those two rates account for all of the patients that did not have SARS (the ones who were classified negative for SARS by the algorithm and the ones who were classified positive for SARS). For this reason, specificity and false positive rate (false alarm rate) are really two sides of the same coin, and, since each conveys the same information about the classifier, evaluators will use them interchangeably depending on whether they want to stress the number of false alarms that the case detection algorithm will generate, or its specificity. Sensitivity and specificity of classification are fundamental properties of a classifier for a condition of interest (in this case the disease SARS). These two numbers completely characterize the detection characteristics of the case detection algorithm. (Note that these methods are also the basis for measuring data accuracy, which we discuss in Chapter 37, and for studying case definitions, which can be understood as a type of classifier that is used for case detection and during outbreak investigations. Case definitions are discussed in Chapters 3 and 13.)

Was this article helpful?

0 0

Post a comment