A Comparison of Statistical Techniques in the Differential Diagnosis of Non-Toxic Goitre

Publisher Website
Google Scholar
Abstract
A modification of the likelihood approach to statistical diagnosis is described, whereby continuous observations are treated as such, rather than being divided into discrete classes. The method has been further extended to allow the inclusion of queried diagnoses, which approximates more closely to the clinical situation. This approach is compared with the discrete likelihood method, again modified to allow queried diagnoses, in an application to the differential diagnosis of non-toxic goiter. Both techniques are contrasted with the diagnoses made by clinical observers experienced in the field of thyroid disease.

This publication has 1 reference indexed in Scilit: