Partial area index of the ROC curve for diagnostic tests

• Autores: Manuel Franco Nicolás , Juana María Vivo Molina
• Localización: XXXI Congreso Nacional de Estadística e Investigación Operativa ; V Jornadas de Estadística Pública: Murcia, 10-13 de febrero de 2009 : Libro de Actas, 2009, ISBN 978-84-691-8159-1
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • The area under receiver operating characteristic (ROC) curve is used in medical decision making to express a summary measure for the e ectiveness of screening and diagnostic tests. Nevertheless, it is not a meaningful index of clinical diagnostic performance when clinical tasks demand a restricted range of sensitivity values. Jiang et al. (Radiology, 201:745-750, 1996) developed a ROC partial area index, which measures clinical diagnostic performance more meaningfully for highly sensitive tests. However, this index might still nd some drawbacks to compare two diagnostic performances under a speci c region.

    Here, we derive tighter bounds for the partial area of an ROC curve, and then attempts to improve the partial area index of an ROC curve are discussed. Thus, we propose a modi ed partial area index for any restricted range of sensitivity values. In addition, we derive this improved partial area index for diagnostic test when high sensitivity must be clinically maintained.