Table 1.
Relevant studies involving ML methods for NEC diagnosis and prognosis.
Fig 1.
The flowchart of the proposed method.
Fig 2.
The structure of the used RQBSO algorithm.
Fig 3.
The pseudocode of the used RQBSO algorithm.
Fig 4.
(a) solutions generated by the first strategy, (b) solutions generated by the second strategy.
Table 2.
Main perinatal and clinical characteristics of two datasets.
Table 3.
Hyper-parameters used by RQBSO algorithm.
Fig 5.
Comparison of ROC and PRC curve of RQBSO and other algorithms.
(a, b) correspond to the ROC curve of dataset 1 and dataset 2. The numbers in parentheses indicate the AUROC value. The x-axis represents sensitivity, or true positive rate (TPR). The y-axis is 1-Specificity, or false positive rate (FPR). (c, d) represents the PRC curve of dataset 1 and dataset 2. The numbers in parentheses indicate the AUPRC value. The x-axis represents recall. The y-axis is precision.
Table 4.
The performance comparison of different feature selection models.
Table 5.
Feature importance ranking of dataset 1.
Table 6.
Feature importance ranking of dataset 2.
Fig 6.
Comparison of ROC and PRC curve of different classifiers.
(a, b) correspond to the ROC curve of dataset 1 and dataset 2. The numbers in parentheses indicate the AUROC value. The x-axis represents sensitivity, or true positive rate (TPR). The y-axis is 1-Specificity, or false positive rate (FPR). (c, d) represents the PRC curve of dataset 1 and dataset 2. The numbers in parentheses indicate the AUPRC value. The x-axis represents recall. The y-axis is precision.