likelihood ratio

(redirected from Likelihood-ratio test)
Also found in: Encyclopedia, Wikipedia.

likelihood ratio

usually preceded by "maximum" (that is, maximum likelihood ratio), this ratio maximizes the probability that the parameters in the ratio agree with the empirically observed data.

like·li·hood ra·ti·o

(līk'lē-hud rā'shē-ō)
The ratio of the probability of a test result among patients with a certain disease or disorder to the probability of that same test result among patients who do not have the targeted disease or disorder.

likelihood ratio

(līk′lē-hood″),

LR

A statistical tool used to help determine the usefulness of a diagnostic test for including or excluding a particular disease. An LR = 1 suggests that the test ordered neither helps to diagnose the disease in question nor helps to rule it out. Higher LRs increase the probability that the disease will be present; LRs < 1.0 decrease the probability that the disease is present.

A positive LR can be thought of as the probability that someone with a suspected condition will, accurately, have a positive test result, divided by the probability that a healthy person will, inaccurately, test positive for the disease. Mathematically this can be represented by the following equation: LR+ = sensitivity of the test/ (1− specificity of the test). A negative LR is the probability that a sick person will fail to be detected by the test, divided by the probability that a healthy person will be accurately shown by the test to have no sign of disease. Mathematically: LR− = (1 − sensitivity of the test) / specificity of the test.

likelihood ratio

The percentage of ill people with a given test result divided by the percentage of well people with the same result. Ratios near unity should not influence decisions. This useful guide to refining clinical diagnosis is little used mainly because of its complexity; The Fagan nomogram can simplify the matter.
References in periodicals archive ?
The REML solution is obtained at the maximum of the restricted likelihood at which point a likelihood-ratio test statistic (LRT) of the form -2 ln([L.
o] ([alpha] = 0), can be determined via a Likelihood-ratio test.
Obviously, the likelihood-ratio test does not call for constrained ML estimation.
To find the significance of all or a subset of coefficients, likelihood-ratio tests can be performed based on the log-likelihood function values.
Likelihood-ratio tests were therefore carried out to determine whether the model allowing [beta] to vary freely yielded significantly better fits than one with three parameters and [beta] fixed at 3.
59 ([+ or -] 11%) ([+ or -] 40%) Table 4 Likelihood-ratio tests comparing differences in the fits between weight-length models which allow [beta] to vary and those which kept [beta] = 3.