Fig. 2: Confidence score distribution and accuracy of the classification performance and exposure assessment evaluation.

The proportion (%) of codes in a confidence score interval is denoted by the bars in the histogram. The mean accuracy (%) and standard deviation of codes (n = 4 independent accuracy scores) in an interval are respectively represented by the dotted and vertical lines in the continuous graph. Here, the dots denote the individual accuracy of each classification model in an interval. The grey area represents the accuracy range over the intervals. Colors correspond to the different classification models. a Results of the classification performance evaluation. b Results of the exposure assessment evaluation of all individuals and c exposed individuals.