Fig. 5: Machine learning (ML, XGBoost) algorithm to identify biomarkers and their interaction predicting recurrence in patients with stage III melanoma. | Nature Communications

Fig. 5: Machine learning (ML, XGBoost) algorithm to identify biomarkers and their interaction predicting recurrence in patients with stage III melanoma.

From: Influence of microbiota-associated metabolic reprogramming on clinical outcome in patients with melanoma from the randomized adjuvant dendritic cell-based MIND-DC trial

Fig. 5: Machine learning (ML, XGBoost) algorithm to identify biomarkers and their interaction predicting recurrence in patients with stage III melanoma.

A ML model summary. Features are clinical parameters and metabolomics (MB) + metagenomics (MG) monitored in serum and feces, respectively, at T1 (n = 88 patients). SHapley Additive exPlanations (SHAP) values for each feature per patient are positive when the value of the feature increases the prediction of recurrence, negative otherwise. Each dot represents one patient and the color represents the value of each feature. The importance of the feature is depicted with the number on the left column. B ML performance using Boruta feature selection algorithm based on XGboost for 2Y-R prediction. Representation of the Area Under the ROC Curve (AUC) values for each treatment arm and feature (clinical, MB or MGS parameters) according to T1, T2 and T2-T1 slope of the trajectory. ROC: receiver operating characteristic. C Circosplot indicating correlations between common features described in (A, B), thickness of lines indicating an increasing positive (pink) or negative (blue) correlation.

Back to article page