Fig. 2: Effects of changes in image preprocessing on AI-based racial identity prediction.

The average scores of the racial identity prediction model were computed for different window width and field of view values and compared to the default preprocessing used to train the model. The average scores were computed in a weighted fashion to equally weight each patient race across the test dataset (see “Methods”). The percent change in average score per race is plotted. A positive change (red) indicates an increase in the average score for the corresponding race and preprocessing combination across the entire test set. Results are shown separately for the CXP and MXR datasets. Source data are provided as a Source data file.