Fig. 3

Performance of the optimal prediction model for treatment response in the discovery and validation cohorts. Visualization of regression performance of four RES-prediction models in a discovery cohort and b external validation cohort. The solid line represents the linear relationship between the scaled prediction value and scaled PANSS reduction rate. c, d illustrate the AUCs of four RES-prediction models. Solid lines in different colors represent receiver operating curves for different RES-prediction models, and the ribbons in different colors represent the confidence intervals of different models in the c discovery cohort and d external validation cohort. e Decision curves for four RES-prediction models in the discovery cohort (solid line) and external validation cohort (dashed line); lines in different colors represent different models. PANSS. Positive and negative syndrome scale; RES. Response; AUC. Area under the curve; C + P. Clinical information + PRS; C + G. Clinical information + GRS; C + M. Clinical information + proxyDNAm; C + PGM. Clinical information + PRS + GRS + proxyDNAm; PRS. Polygenic risk score; GRS. Genetic risk score