Optimal Diagnostic Strategy

Thursday, October 18, 2012 - 4:00pm
Ben Haaland, National University of Singapore
140 Bardeen

Developing a diagnostic strategy is a common problem in medicine.
Issues and examples in clinical diagnosis are explored. A few methods of
developing a diagnostic strategy are illustrated and discussed. Measures of
the quality of diagnostic tests and strategies, including simplicity,
sensitivity, specificity, receiver operating characteristics,
positive/negative predictive values, positive/negative likelihood ratios,
costs, and risks, are examined. Techniques for comparing tests are
illustrated. Rules-of-thumb for choosing a single test, the threshold for a
score-based test, and a sequence of tests are developed and illustrated. An
optimal diagnostic strategy in terms of net benefit is presented. Several
examples from clinical practice highlight the discussed methods.