Table 1.
Characteristics of heart failure patients in the study cohort (n = 10,757).
Table 2.
Variables selected by the statistical method (t-test and chi-squared test) for predicting 30-day HF readmission or death (sorted in ascending order of p-value).
Table 3.
Variables selected by the wrapper-based machine learning techniques for predicting 30-day HF readmission or death.
Fig 1.
Performance of the multi-layer perceptron (MLP) prediction model on variables generated by principal component analysis.
Table 4.
Performance of the multi-layer perceptron prediction model for 30-day HF readmission or death in HF patients.