Please join the Biometry Hub Virtual StatsPD meeting with a presentation by Charles Caraguel from the School of Animal and Veterinary Sciences. The presentation will be based on a recent overview of diagnostic likelihood ratio, co-authored by Associate Professor Caraguel.

‘Diagnostic likelihood ratio – often ignored, usually useful’

Misclassification is not only a common and expected concerns in scientific research, it is also a common occurrence in the diagnosis of medical patients, be it for humans or animals. When evaluating and reporting the accuracy of patients’ classification towards a disease, the diagnostic sensitivity (classification accuracy among diseased patients) and the diagnostic specificity (classification accuracy among non-diseased patients) of a test are conventionally used since the mid-1940s. However, these measures of classification accuracy have their limitations when use in isolation, in particular when selecting, among several, the best suited test. The diagnostic likelihood ratio, first introduced in 1980s and primarily used in human medicine, appears to alleviate some of these limitations but remains relatively unused.

Charles Caraguel will provide an overview of what the diagnostic likelihood ratio is and how it can surpass the diagnostic sensitivity and specificity using examples from serological tests for bovine tuberculosis.