Fig. 4
From: Polygenic risk score improves the accuracy of a clinical risk score for coronary artery disease
Change in predicted probabilities and risk reclassification. A Difference between 10-year risk by PCE and PRS-enhanced PCE. B PCE + PRS 10-year risk reclassification. C Net reclassification improvement and integrated discrimination improvement results. A Change in the predicted probabilities of the recalibrated pooled cohort equations (PCE) model after the addition of polygenic risk scores (PRSs) for CAD. The x-axis shows the predicted probability from the baseline PCE model. The y-axis is the difference in 10-year risk probabilities of a CAD event between the PRS-enhanced model and the baseline PCE model. The scatterplot has a random draw of 1% of the participants shown. The histogram x- and y-axes are based on the full population. B Reclassification table of predicted probabilities by PCE and PRS-enhanced PCE models at 7.5% threshold. Rows indicating an improved classification with the PRS-enhanced PCE model are marked by a plus sign while rows indicating a deteriorated classification are marked by a minus sign. C Table of net reclassification improvement (NRI) and integrated discrimination improvement (IDI). NRIa is defined in the continuous case as the sum of proportions of cases and noncases with improved combined score minus the sum of proportions with a deteriorated combined score. In the categorical case, NRI is defined by change at a 7.5% threshold predicted probability. A positive NRI indicates a better combined score overall. IDIb measures the difference of average probabilities of an event in cases and noncases. A larger IDI indicates more discrimination in the combined score. aNRI = P(up|case) − P(down|case) − P(up|noncase) + P(down|noncase). bIDI = PPCE+PRS(case) − PPCE+PRS(noncase) − PPCE(case) + PPCE(noncase)