Fig. 6
From: Artificial intelligence for the prevention and prediction of colorectal neoplasms
Prediction model performance for the best models for men and women: 0 S-SKDE-4E-GP and 0 S-SKDE-3E-GP, respectively. S stands for the polyp threshold size, SKDE is the sigmoid-like KDE transformation, E is the exponent applied on the KDE transformation and GP is Gaussian process classifier. a, d The KDE of the predicted scores for polyp incidence. Each distribution of yes/no is for the true incidence. b, e Violin plots of the predicted scores for true positive (TP), true negative (TN), false positive (FP), and false negative (FN). c, f Normalized confusion matrix