Fig. 3: Generation of Yale Mixed Modality Model (YMMM) for the prediction of best overall response to immunotherapy in patients with advanced melanoma. | npj Precision Oncology

Fig. 3: Generation of Yale Mixed Modality Model (YMMM) for the prediction of best overall response to immunotherapy in patients with advanced melanoma.

From: Models that combine transcriptomic with spatial protein information exceed the predictive value for either single modality

Fig. 3

a Identification of the optimal number of predictors for final model inclusion. Area under the curve (AUC) and 95% confidence intervals based on the number of predictors included in the model; AUC curve peaks when 8 predictors are included in the model. b Heatmap showing the most frequently appearing predictors for any given number of variables included in the model (K); calculations are based on the top ten highest AUC models for different K values. For K = 8, selected predictors are CCNO, ID4, IER3, IL2RB, MGMT, NRDE2, TNFAIP6, and MSH2 in s100/HMB45.

Back to article page