Skip to main content
. 2022 Feb 5;26:100901. doi: 10.1016/j.conctc.2022.100901

Fig. 4.

Fig. 4

Bayesian Logistic Regression ED-EWOC dose recommendation process after completion of the fourth dose cohort. Left panel: (top) predictions of individual AUC estimates after the first cycle from the population PK model at each dose evaluated. Dots correspond to patients' data with black dots and red triangles respectively representing patients without and with DLT respectively, plain line is for the median AUC estimate, and dashed lines represent the 95% prediction interval. (bottom) Distribution of predicted population AUC for a given dose (here, for d = 2.75 mg). Middle panel: (top) Estimated posterior probability of DLT, defined as a function of exposure (AUC over first cycle). Dots correspond to patients' data, plain line is for the median DLT rate estimate and dashed lines represent the 95% prediction interval. (bottom) Histograms of the estimated distribution of DLT rates for different exposures. Right panel: Boxplot of the posterior integrated probability of DLT for each candidate dose according to the ED-EWOC approach. The DLT rate was assessed by Monte Carlo integration using the distribution of AUCs, as indicated. The pre-specified intervals of interest which PUD, PT, and POD relied upon and defining decision rules for the recommendation have been overlaid. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)