![Feature impact scores, performance metrics, and validation of machine learning model 2 for predicting sMNs. (A) Bar plot of the feature impact scores for model 2, demonstrating the relative contribution of each feature to the model’s performance. (B) Multivariate regression (R) scores and P values of variables used in model 2, with asterisks denoting statistical significance (P < .05). (C) Summary of model 2 performance metrics, including sensitivity (0.84), specificity (0.73), and an AUC of 0.82. (D) Confusion matrix displaying true positives (36 [13.0%]), true negatives (169 [61.5%]), false positives (62 [22.5%]), and false negatives (8 [2.9%]). (E) AUC-ROC curve illustrating the model’s predictive performance with an AUC of 0.82, indicating strong discriminatory ability for identifying sMN risk. ATG, antithymocyte globulin.](https://ash.silverchair-cdn.com/ash/content_public/journal/bloodneoplasia/2/4/10.1016_j.bneo.2025.100163/1/bneo_neo-2025-000694-gr2.jpeg?Expires=1765532018&Signature=dxjVIRd1zWrA94-cxBbdSJQnl8~qgflBiTDvAmSVgpRLhN1Zn--Td9jd~U4Anew5TQWTvO8vH2H2bFchzhIfhI9suB~VNX5Iefo~a0StXiMvKeKHgoECZOJZrm6ndnw1LYHJzHAMWxC0OLRwI6WElK0jTRpz~p8FDaM20XGnTxzFMmSlDtTH13BKunDoqUDOI~aki~E9Dy4bOIWRvC3poMpLr9tjYXzVUW6M2djyEJsBrpOl6THGZK23mfNSPZvglMBvkETXUSDV0v6lOmA67ycVGlDZTD6CxKm~dwFZhcnWxVYU-S8kmEkw1jdXdZKR3BOm8PP9XtW19n0MYnGDuQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Feature impact scores, performance metrics, and validation of machine learning model 2 for predicting sMNs. (A) Bar plot of the feature impact scores for model 2, demonstrating the relative contribution of each feature to the model’s performance. (B) Multivariate regression (R) scores and P values of variables used in model 2, with asterisks denoting statistical significance (P < .05). (C) Summary of model 2 performance metrics, including sensitivity (0.84), specificity (0.73), and an AUC of 0.82. (D) Confusion matrix displaying true positives (36 [13.0%]), true negatives (169 [61.5%]), false positives (62 [22.5%]), and false negatives (8 [2.9%]). (E) AUC-ROC curve illustrating the model’s predictive performance with an AUC of 0.82, indicating strong discriminatory ability for identifying sMN risk. ATG, antithymocyte globulin.
