Comparing Machine Learning Models and Statistical Models for Predicting Heart Failure Events: A Systematic Review and Meta-Analysis

Figure 1. Literature selection procedure.

Figure 2. Performance distribution of statistical and ML models. (A) Pooled C-index of meta-analysis with respect to tasks and model types. (B) Task specific comparison result. (C) Model specific comparison result. In subplots (B,C), the inner pie indicates the number of pairwise comparisons between ML models and statistical models. The middle pie indicates the number of pairs in which the ML model achieved better performance, while the outer pie indicates the number of pairs in which the superiority of ML models reached statistical significance. ML, Machined Learning; DL, Deep Learning; DT, Decision Tree; SVM, Support Vector Machine; MLP, Multi-Layer Perceptron; RF, Random Forest.

Figure 3. Result of risk of bias analysis.