Due to its low prevalence, high mortality and uniquely hidden intrapelvic position, ovarian cancer remains a subject of intense interest to researchers. Statistical calculation and new technology both have major roles to play in the effort to screen this cancer at an early stage. Advanced statistics, such as multivariate analysis, remain at the root of screening endeavors. Multivariate analysis has the power to combine many tests and to produce better results in terms high specificity and positive predictive value. Multivariate analysis techniques include Mahalanobis distance (D2), canonical stepwise discriminant function (Z) and Posterior Probability. These may have varied efficacy, but to date comparisons have not been conducted to determine which is best in the context of ovarian cancer screening.
Bose, Chinmoy K.
"Efficiency of Canonical Discriminant Function versus Mahalanobis Distance in Differentiating Groups: Screening Ovarian Cancer in a Multivariate System Analysis Using Enzyme Markers,"
Journal of Modern Applied Statistical Methods:
1, Article 29.
Available at: http://digitalcommons.wayne.edu/jmasm/vol8/iss1/29