Figure 1

Hypothetical distribution patterns of continuous markers with different relative risks, and thresholds for risk prediction. Two normal distributions (mean = 0 and standard deviation = 0.5) are used to represent the distribution of a continuous marker in disease (solid curved line) and non-disease (dashed curved line) populations for six different ORs. The locations of the means for the disease population are consistent with the logit model Pr(D = 1|X) = α + βX, in which one unit increase corresponds to the OR shown in the figure. The three vertical bars (solid, dotted, and dashed) correspond to different thresholds (cut off value ‘c’) for positive-negative calls of a disease with a continuous distribution marker. Specifically, the solid bar represents the threshold value c such that the sensitivity is kept for 0.95 for various OR values in the plot; the dotted bar represents the threshold value c such that the sensitivity and specificity are equal for various OR values in the plots; and the dashed bar represents the threshold value c such that the specificity is kept for 0.95 for various OR values in the plot. The examples of using the continuous marker for disease classification or prevention are shown in Figure 2; and corresponding sensitivity, specificity and PAR% of various thresholds (three bars in this Figure) are shown in Table 2 and Figure 2 (the cross, circle, and triangle in Figure 2 correspond to solid, dashed, and solid vertical bars in Figure 1).