Framework for feature selection of predicting the diagnosis and prognosis of necrotizing enterocolitis

Abstract

Neonatal necrotizing enterocolitis (NEC) occurs worldwide and is a major source of neonatal morbidity and mortality. Researchers have developed many methods for predicting NEC diagnosis and prognosis. However, most people use statistical methods to select features, which may ignore the correlation between features. In addition, because they consider a small dimension of characteristics, they neglect some laboratory parameters such as white blood cell count, lymphocyte percentage, and mean platelet volume, which could be potentially influential factors affecting the diagnosis and prognosis of NEC. To address these issues, we include more perinatal, clinical, and laboratory information, including anemia—red blood cell transfusion and feeding strategies, and propose a ridge regression and Q-learning strategy based bee swarm optimization (RQBSO) metaheuristic algorithm for predicting NEC diagnosis and prognosis. Finally, a linear support vector machine (linear SVM), which specializes in classifying high-dimensional features, is used as a classifier. In the NEC diagnostic prediction experiment, the area under the receiver operating characteristic curve (AUROC) of dataset 1 (feeding intolerance + NEC) reaches 94.23%. In the NEC prognostic prediction experiment, the AUROC of dataset 2 (medical NEC + surgical NEC) reaches 91.88%. Additionally, the classification accuracy of the RQBSO algorithm on the NEC dataset is higher than the other feature selection algorithms. Thus, the proposed approach has the potential to identify predictors that contribute to the diagnosis of NEC and stratification of disease severity in a clinical setting.

State of the art review on machine learning and artificial intelligence in the study of neonatal necrotizing enterocolitis

“…One cluster of peptides classified as fibrinogen A were most useful and when developing a linear discriminate analysis (LDA) model using both clinical parameters and urine peptide biomarkers, the model was able to correctly classify 100% of the infants as either surgical NEC or medical NEC, while the model using only clinical features was unable to classify 39% of the patients (46). Song et al designed an algorithm with the intent of determining features that would be important to distinguish NEC diagnosis from feeding intolerance (FI) and predicting whether infants with NEC will require surgery (60). In their model distinguishing NEC from FI, seven features from their original set of 119 were important for diagnosis and their model achieved a high AUROC score (60).…”

“…Song et al designed an algorithm with the intent of determining features that would be important to distinguish NEC diagnosis from feeding intolerance (FI) and predicting whether infants with NEC will require surgery (60). In their model distinguishing NEC from FI, seven features from their original set of 119 were important for diagnosis and their model achieved a high AUROC score (60). With a similar AUROC score, the model predicting NEC prognosis also had high performance and weighted five of the features as being most important for prediction (60).…”

“…To understand more about complex models' decision-making process, separate ML models can be developed like Lin et al who used a RF model to determine what taxonomic features from the microbiome data were important for the MIL model (54). Others used unsupervised ML through hierarchical clustering to narrow down the features that were used in the final model such as in Sylvester et al and Rusconi et al (46,60). Requiring a secondary model to understand the model being developed adds another layer of complexity to the ML process and can make interpretability difficult for the eventual end users, the clinicians, who have varying levels of understanding of ML (65,66).…”

Determining important features for dengue diagnosis using feature selection methods

“…Feature selection methods are often used to minimize the number of input variables that are considered to be the most significant to a machine learning model to improve the model performance [9,10]. In recent years, numerous publications focus on the implementation of feature selection methods for disease prediction [9][10][11][12][13]. In the classification stage, most researchers use machine learning techniques such as BayesNet [9,10,13], support vector machine [9,11] and tree-based classifiers [9,10,13].…”

“…In recent years, numerous publications focus on the implementation of feature selection methods for disease prediction [9–13]. In the classification stage, most researchers use machine learning techniques such as BayesNet [9,10,13], support vector machine [9,11] and tree-based classifiers [9,10,13].…”