Youden index

The Youden index (also Youden's J ) is a measure that can be used to determine which threshold value is best suited to differentiate between two groups in a measurement using a receiver operating characteristic . The Youden Index was developed by William J. Youden in 1950 and is calculated as follows:

{\ displaystyle J = {\ text {Sensitivity}} + {\ text {Specificity}} - 1}

where sensitivity and specificity are measures for assessing a classifier for group differentiation.

According to Youden, the value range is between 0 and 1. The separation of the groups works best when the Youden index is the largest. The index is independent of the basic rate (prevalence) of a disorder. It is different, for example, with the overall hit rate (English. Accuracy, i.e. the probability of diagnosing correctly). The case numbers of the two groups to be separated are included here, which can lead to a bias if the value is dominated by the error rate of the larger group. The Youden index must be distinguished from the measures of total hit rate and detection rate. In the Youden Index, false negative classifications are weighted in the same way as false positive ones. For example, a test with a sensitivity of 0.25 and a specificity of 0.9 would give the same value for the Youden index as a test with a sensitivity of 0.9 and a specificity of 0.25. In practice, however, a classification error of one type can be more serious than a classification error of the other type, which is why it can be useful to consider these classification errors separately. For example, you could choose a threshold value at which some healthy people are falsely categorized as sick, but less really sick people are overlooked.

Individual evidence

↑ Lothar Kreienbrock, Iris Pigeot, Wolfgang Ahrens: Epidemiological Methods . Springer-Verlag, 2012, ISBN 978-3-8274-2334-4 , pp. 171 ( google.de ).
^ ^A ^b William J. Youden: Index for rating diagnostic tests . In: Cancer . 3, No. 1, 1950, pp. 32-35. doi : 10.1002 / 1097-0142 (1950) 3: 1 <32 :: aid-cncr2820030106> 3.0.co; 2-3 .
↑ ^a ^b Helfried Moosbrugger, Augustin Kelava: Test theory and questionnaire construction . Springer, 2011, ISBN 978-3-642-20072-4 , pp. 186–187 ( limited preview in Google Book search).
^ ^A ^b Heinrich Tröster: Early detection in children and adolescents: strategies for developmental, learning and behavioral disorders . Hogrefe Verlag, 2009, ISBN 978-3-8409-2078-3 , pp. 94 ( limited preview in Google Book search).
↑ Martin Schumacher, Gabriele Schulgen-Kristiansen: Methodology of clinical studies: methodological principles of planning, implementation and evaluation . Springer-Verlag, 2008, ISBN 978-3-540-85136-3 ( google.de [accessed on February 17, 2019]).
^ Noel S. Weiss, Thomas D. Koepsell: Epidemiologic Methods: Studying the Occurrence of Illness . Oxford University Press, 2014, ISBN 978-0-19-938778-6 ( google.de [accessed February 16, 2019]).

[1] Lothar Kreienbrock, Iris Pigeot, Wolfgang Ahrens: Epidemiological Methods . Springer-Verlag, 2012, ISBN 978-3-8274-2334-4 , pp. 171 ( google.de ).

[:2-2] A ^b William J. Youden: Index for rating diagnostic tests . In: Cancer . 3, No. 1, 1950, pp. 32-35. doi : 10.1002 / 1097-0142 (1950) 3: 1 <32 :: aid-cncr2820030106> 3.0.co; 2-3 .

[:0-3] Helfried Moosbrugger, Augustin Kelava: Test theory and questionnaire construction . Springer, 2011, ISBN 978-3-642-20072-4 , pp. 186–187 ( limited preview in Google Book search).

[:1-4] A ^b Heinrich Tröster: Early detection in children and adolescents: strategies for developmental, learning and behavioral disorders . Hogrefe Verlag, 2009, ISBN 978-3-8409-2078-3 , pp. 94 ( limited preview in Google Book search).

[5] Martin Schumacher, Gabriele Schulgen-Kristiansen: Methodology of clinical studies: methodological principles of planning, implementation and evaluation . Springer-Verlag, 2008, ISBN 978-3-540-85136-3 ( google.de [accessed on February 17, 2019]).

[6] Noel S. Weiss, Thomas D. Koepsell: Epidemiologic Methods: Studying the Occurrence of Illness . Oxford University Press, 2014, ISBN 978-0-19-938778-6 ( google.de [accessed February 16, 2019]).