A Predictive Model for Major Adverse Aortic Events in Patients with Abdominal Aortic Aneurysm Using Clinical and Biomarker Data

Figures

• 

  Figure 1. Circulating protein levels expressed as mean (standard deviation) for patients with vs. without abdominal aortic aneurysm (AAA). (A) Matrix metalloproteinase 3 (MMP-3), (B) human epididymal secretory protein 4 (HE4/WFDC2), (C) Chitinase 3-like-1, (D) Kallikrein 6/Neurosin, (E) B lymphocyte stimulator (BAFF/BLyS), (F) Cathepsin S, and (G) matrix metalloproteinase 1 (MMP-1).

• 

  Figure 2. Receiver operating characteristic curve for the random forest machine learning model in predicting 2-year major adverse aortic events on test set data. Area represents area under the receiver operating characteristic curve (AUROC); panel refers to 4-protein biomarker panel consisting of matrix metalloproteinase 3 (MMP-3), human epididymal secretory protein 4 (HE4/WFDC2), Chitinase 3-like-1, and Kallikrein 6/Neurosin.

• 

  Figure 3. Variable importance scores (gain) for the clinical characteristics and proteins used as input features for the random forest machine learning model for predicting 2-year major adverse aortic events. Higher score indicates greater importance in contributing to an overall prediction. Abbreviations: matrix metalloproteinase 3 (MMP3), human epididymal secretory protein 4 (HE4WFDC2), AAA (abdominal aortic aneurysm), smoking history (SmokingHx; includes both past and current smoking), coronary artery disease (CAD), peripheral artery disease (PAD), hypercholesterolemia (hyperchol).

• 

  Figure 4. Kaplan–Meier analysis of freedom from major adverse aortic events in patients predicted to be at low vs. high risk by the random forest machine learning model. The receiver operating characteristic curve (ROC) threshold used to classify patients into low vs. high risk was 0.39. Abbreviations: HR (hazard ratio), CI (confidence interval).

• 

  Figure 5. Receiver operating characteristic curve for the random forest machine learning model in predicting 2-year major adverse cardiovascular events on test set data. Area represents area under the receiver operating characteristic curve (AUROC); panel refers to 4-protein biomarker panel consisting of matrix metalloproteinase 3 (MMP-3), human epididymal secretory protein 4 (HE4/WFDC2), Chitinase 3-like-1, and Kallikrein 6/Neurosin.

• 

  Figure 6. Summary of predictive model development process. AAA (abdominal aortic aneurysm), AUROC (area under the receiver operating characteristic curve).