Comparative evaluation of machine learning models versus TIMI score in ST-segment-elevation myocardial infarction patients

Figures

• 

  Fig. 1. Central illustration showing the overview of the methods used for data selection, use of machine learning (ML) algorithms, validation and the performance of ML model as well as TIMI score (AUC: Area Under the Curve; ML: machine learning; MI: myocardial infarction; STEMI:ST-Segment Elevation Myocardial Infarction; TIMI: thrombolysis in myocardial infarction; mo: months; HF: heart failure; MR: mitral regurgitation; CAD: coronary artery disease; LVEF: left ventricular ejection fraction).

• 

  Fig. 2. ROC curve on a training data (A), Unseen data (B) set for six different machine learning models for in-hospital mortality.

• 

  Fig. 3. SHAP based on Random Forest in-hospital mortality. Each point on a feature line is a SHAP value of a sample. The y-axis represents feature ranking in descending order. The x-axis represents SHAP value. The right-side (+ve SHAP value) of the central line indicates the mortality class. A greater + ve SHAP value indicates higher impact on the prediction of the mortality class. Features having red color on the left side (high BMI, ejection fraction more than 35 %) are in favor of survival. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

• 

  Fig. 4. ROC curve on a training data (A), Unseen data (B) set for six different machine learning models for 30 days mortality.

• 

  Fig. 5. SHAP based on Random Forest 30 days mortality. Each point on a feature line is a SHAP value of a sample. The y-axis represents feature ranking in descending order. The x-axis represents SHAP value. The right-side (+ve SHAP value) of the central line indicates the mortality class. A greater + ve SHAP value indicates higher impact on the prediction of the mortality class. Features having red color on the left side (aspirin, beta blocker, ACEi/ARBs, ejection fraction more than 40 %) are in favor of survival. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)