Added value of systemic inflammation markers for monitoring response to neoadjuvant chemotherapy in breast cancer patients

Figure 1 The flow chart of patient selection and data process. NACT: Neoadjuvant chemotherapy; ROC: Receiver operating characteristic.

Figure 2 Statistical analysis of features included in machine learning based models. A: Heatmap representing the correlation between candidate variables included in predictive models using Spearman’s correlation coefficient; B: Scaled importance rank of all features included in predictive models for identifying risk of response after neoadjuvant chemotherapy (rNACT) in breast cancer patients; C-J: Box and jitter plots showing distribution of continuous features included in predictive models between rNACT and non-rNACT groups. BMI: Body mass index; PR: Partial response; NLR: Neutrophil-to-lymphocyte ratio; LMR: Lymphocyte-to-monocyte ratio; PLR: Platelet-to-lymphocyte ratio; A/G: Albumin-to-globulin ratio; LDL: Low-density lipoprotein; HDL: High-density lipoprotein; TC: Total cholesterol; TG: Triglyceride.

Figure 3 Validation and comparison of the predictive model. A: Area under the receiver operating characteristic curve (AUC) to assess the performance of response after neoadjuvant chemotherapy (rNACT) risk prediction of machine learning based models; B: AUC to assess the performance of rNACT risk prediction of generalized linear model (GLM); C: Discriminative evaluation of support vector machine in predicting rNACT; D: Discriminative evaluation of GLM in predicting rNACT. SVM: Support vector machine; GLM: Generalized linear model.

Figure 4 Prediction performance of support vector machine model via clinical impact curve. A: Training set; B: Validation set. The green line predicts the probability of poor response after neoadjuvant chemotherapy (rNACT), and the blue line shows how many patients will be at high risk of non-rNACT.