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

AI models in clinical neonatology: a review of modeling approaches and a consensus proposal for standardized reporting of model performance

Artificial intelligence (AI) is a rapidly advancing area with growing clinical applications in healthcare. The neonatal intensive care unit (NICU) produces large amounts of multidimensional data allowing AI and machine learning (ML) new avenues to improve early diagnosis, enhance monitoring, and provide highly-targeted treatment approaches. In this article, we review recent clinical applications of AI to important neonatal problems, including sepsis, retinopathy of prematurity, bronchopulmonary dysplasia, and others. For each clinical area, we highlight a variety of ML models published in the literature and examine the future role they may play at the bedside. While the development of these models is rapidly expanding, a fundamental understanding of model selection, development, and performance evaluation is crucial for researchers and healthcare providers alike. As AI plays an increasing role in daily practice, understanding the implications of AI design and performance will enable more effective implementation. We provide a comprehensive explanation of the AI development process and recommendations for a standardized performance metric framework. Additionally, we address critical challenges, including model generalizability, ethical considerations, and the need for rigorous performance monitoring to avoid model drift. Finally, we outline future directions, emphasizing the importance of collaborative efforts and equitable access to AI innovations.

Diagnosing Necrotizing Enterocolitis via Fine-Grained Visual Classification

Necrotizing Enterocolitis (NEC) is a devastating condition affecting prematurely born neonates. Reviewing Abdominal X-rays (AXRs) is a key step in NEC diagnosis, staging and treatment decision-making, but poses significant challenges due to the subtle, difficult-to-identify radiological signs of the disease. In this paper, we propose AIDNEC - AI Diagnosis of NECrotizing enterocolitis, a deep learning method to automatically detect and stratify the severity (surgical or medical) of NEC from no pathology in AXRs. The model is trainable end-to-end and integrates a Detection Transformer and Graph Convolution modules for localizing discriminative areas in AXRs, used to formulate subtle local embeddings. These are then combined with global image features to perform Fine-Grained Visual Classification (FGVC). We evaluate AIDNEC on our GOSH NEC dataset of 1153 images from 334 patients, achieving 79.7% accuracy in classifying NEC against No Pathology. AIDNEC outperforms the backbone by 2.6%, FGVC models by 2.5% and CheXNet by 4.2%, with statistically significant (two-tailed p 0.05) improvements, while providing meaningful discriminative regions to support the classification decision. Additional validation in the publicly available Chest X-ray14 dataset yields comparable performance to state-of-the-art methods, illustrating AIDNEC's robustness in a different X-ray classification task.

Construction and evaluation of a risk model for adverse outcomes of necrotizing enterocolitis based on LASSO-Cox regression

Objective

This study aimed to develop a nomogram to predict adverse outcomes in neonates with necrotizing enterocolitis (NEC).

Methods

In this retrospective study on neonates with NEC, data on perinatal characteristics, clinical features, laboratory findings, and x-ray examinations were collected for the included patients. A risk model and its nomogram were developed using the least absolute shrinkage and selection operator (LASSO) Cox regression analyses.

Results

A total of 182 cases of NEC were included and divided into a training set (148 cases) and a temporal validation set (34 cases). Eight features, including weight [p = 0.471, HR = 0.99 (95% CI: 0.98-1.00)], history of congenital heart disease [p < 0.001, HR = 3.13 (95% CI:1.75-5.61)], blood transfusion before onset [p = 0.757, HR = 0.85 (95%CI:0.29-2.45)], antibiotic exposure before onset [p = 0.003, HR = 5.52 (95% CI:1.81-16.83)], C-reactive protein (CRP) at onset [p = 0.757, HR = 1.01 (95%CI:1.00-1.02)], plasma sodium at onset [p < 0.001, HR = 4.73 (95%CI:2.61-8.59)], dynamic abdominal x-ray score change [p = 0.001, HR = 4.90 (95%CI:2.69-8.93)], and antibiotic treatment regimen [p = 0.250, HR = 1.83 (0.65-5.15)], were ultimately selected for model building. The C-index for the predictive model was 0.850 (95% CI: 0.804-0.897) for the training set and 0.7880.788 (95% CI: 0.656-0.921) for the validation set. The area under the ROC curve (AUC) at 8-, 10-, and 12-days were 0.889 (95% CI: 0.822-0.956), 0.891 (95% CI: 0.829-0.953), and 0.893 (95% CI:0.832-0.954) in the training group, and 0.812 (95% CI: 0.633-0.991), 0.846 (95% CI: 0.695-0.998), and 0.798 (95%CI: 0.623-0.973) in the validation group, respectively. Calibration curves showed good concordance between the predicted and observed outcomes, and DCA demonstrated adequate clinical benefit.

Conclusions

The LASSO-Cox model effectively identifies NEC neonates at high risk of adverse outcomes across all time points. Notably, at earlier time points (such as the 8-day mark), the model also demonstrates strong predictive performance, facilitating the early prediction of adverse outcomes in infants with NEC. This early prediction can contribute to timely clinical decision-making and ultimately improve patient prognosis.

Understanding necrotizing enterocolitis endotypes and acquired intestinal injury phenotypes from a historical and artificial intelligence perspective

Necrotizing enterocolitis (NEC) remains a devastating disease in preterm and term neonates. Despite significant progress made in understanding NEC pathogenesis over the last 50 years, the inability of current definitions to discriminate the various pathophysiological processes underlying NEC has led to an umbrella term that limits clinical and research progress. In this mini review, we provide a historical perspective on how NEC definitions and pathogenesis have evolved to our current understanding of NEC endotypes. We also discuss how artificial intelligence-based approaches are influencing our knowledge of risk-factors, classification and prognosis of NEC and other neonatal intestinal injury phenotypes.

Struggling to Understand the NEC Spectrum-Could the Integration of Metabolomics, Clinical-Laboratory Data, and Other Emerging Technologies Help Diagnosis?

Necrotizing enterocolitis (NEC) is the most prevalent and potentially fatal intestinal injury mainly affecting premature infants, with significant long-term consequences for those who survive. This review explores the scale of the problem, highlighting advancements in epidemiology, the understanding of pathophysiology, and improvements in the prediction and diagnosis of this complex, multifactorial, and multifaced disease. Additionally, we focus on the potential role of metabolomics in distinguishing NEC from other conditions, which could allow for an earlier and more accurate classification of intestinal injuries in infants. By integrating metabolomic data with other diagnostic approaches, it is hoped to enhance our ability to predict outcomes and tailor treatments, ultimately improving care for affected infants.

An investigation into augmentation and preprocessing for optimising X-ray classification in limited datasets: a case study on necrotising enterocolitis

PURPOSE

Obtaining large volumes of medical images, required for deep learning development, can be challenging in rare pathologies. Image augmentation and preprocessing offer viable solutions. This work explores the case of necrotising enterocolitis (NEC), a rare but life-threatening condition affecting premature neonates, with challenging radiological diagnosis. We investigate data augmentation and preprocessing techniques and propose two optimised pipelines for developing reliable computer-aided diagnosis models on a limited NEC dataset.

METHODS

We present a NEC dataset of 1090 Abdominal X-rays (AXRs) from 364 patients and investigate the effect of geometric augmentations, colour scheme augmentations and their combination for NEC classification based on the ResNet-50 backbone. We introduce two pipelines based on colour contrast and edge enhancement, to increase the visibility of subtle, difficult-to-identify, critical NEC findings on AXRs and achieve robust accuracy in a challenging three-class NEC classification task.

RESULTS

Our results show that geometric augmentations improve performance, with Translation achieving +6.2%, while Flipping and Occlusion decrease performance. Colour augmentations, like Equalisation, yield modest improvements. The proposed Pr-1 and Pr-2 pipelines enhance model accuracy by +2.4% and +1.7%, respectively. Combining Pr-1/Pr-2 with geometric augmentation, we achieve a maximum performance increase of 7.1%, achieving robust NEC classification.

CONCLUSION

Based on an extensive validation of preprocessing and augmentation techniques, our work showcases the previously unreported potential of image preprocessing in AXR classification tasks with limited datasets. Our findings can be extended to other medical tasks for designing reliable classifier models with limited X-ray datasets. Ultimately, we also provide a benchmark for automated NEC detection and classification from AXRs.

Machine learning-based analysis for prediction of surgical necrotizing enterocolitis in very low birth weight infants using perinatal factors: a nationwide cohort study

Early prediction of surgical necrotizing enterocolitis (sNEC) in preterm infants is important. However, owing to the complexity of the disease, identifying infants with NEC at a high risk for surgical intervention is difficult. We developed a machine learning (ML) algorithm to predict sNEC using perinatal factors obtained from the national cohort registry of very low birth weight (VLBW) infants. Data were collected from the medical records of 16,385 VLBW infants registered in the Korean Neonatal Network (KNN). Infants who underwent surgical intervention were identified with sNEC, and infants who received medical treatment, with medical NEC (mNEC). We used 38 variables, including maternal, prenatal, and postnatal factors that were obtained within 1 week of birth, for training. A total of 1085 patients had NEC (654 with sNEC and 431 with mNEC). VLBW infants showed a higher incidence of sNEC at a lower gestational age (GA) (p < 0.001). Our proposed ensemble model showed an area under the receiver operating characteristic curve of 0.721 for sNEC prediction.    Conclusion: Proposed ensemble model may help predict which infants with NEC are likely to develop sNEC. Through early prediction and prompt intervention, prognosis of sNEC may be improved. What is Known: • Machine learning (ML)-based techniques have been employed in NEC research for prediction, diagnosis, and prognosis, with promising outcomes. • While most studies have utilized abdominal radiographs and clinical manifestations of NEC as data sources, and have demonstrated their usefulness, they may prove weak in terms of early prediction. What is New: • We analyzed the perinatal factors of VLBW infants acquired within 7 days of birth and used ML-based analysis to identify which infants with NEC are vulnerable to clinical deterioration and at high risk for surgical intervention using nationwide cohort data.

An investigation into augmentation and preprocessing for optimising X-ray classification in limited datasets: a case study on necrotising enterocolitis

Purpose

Obtaining large volumes of medical images, required for deep learning development, can be challenging in rare pathologies. Image augmentation and preprocessing offer viable solutions. This work explores the case of necrotising enterocolitis (NEC), a rare but life-threatening condition affecting premature neonates, with challenging radiological diagnosis. We investigate data augmentation and preprocessing techniques and propose two optimised pipelines for developing reliable computer-aided diagnosis models on a limited NEC dataset.

Methods

We present a NEC dataset of 1090 Abdominal X-rays (AXRs) from 364 patients and investigate the effect of geometric augmentations, colour scheme augmentations and their combination for NEC classification based on the ResNet-50 backbone. We introduce two pipelines based on colour contrast and edge enhancement, to increase the visibility of subtle, difficult-to-identify, critical NEC findings on AXRs and achieve robust accuracy in a challenging three-class NEC classification task.

Results

Our results show that geometric augmentations improve performance, with Translation achieving +6.2%, while Flipping and Occlusion decrease performance. Colour augmentations, like Equalisation, yield modest improvements. The proposed Pr-1 and Pr-2 pipelines enhance model accuracy by +2.4% and +1.7%, respectively. Combining Pr-1/Pr-2 with geometric augmentation, we achieve a maximum performance increase of 7.1%, achieving robust NEC classification.

Conclusion

Based on an extensive validation of preprocessing and augmentation techniques, our work showcases the previously unreported potential of image preprocessing in AXR classification tasks with limited datasets. Our findings can be extended to other medical tasks for designing reliable classifier models with limited X-ray datasets. Ultimately, we also provide a benchmark for automated NEC detection and classification from AXRs.

Surgical prediction of neonatal necrotizing enterocolitis based on radiomics and clinical information

PURPOSE

To assess the predictive value of radiomics for surgical decision-making in neonatal necrotizing enterocolitis (NEC) when abdominal radiographs (ARs) do not suggest an absolute surgical indication for free pneumoperitoneum.

METHODS

In this retrospective study, we finally included 171 newborns with NEC and obtained their ARs and clinical data. The dataset was randomly divided into a training set (70%) and a test set (30%). We developed machine learning models for predicting surgical treatment using clinical features and radiomic features, respectively, and combined these features to build joint models. We assessed predictive performance of the different models by receiver operating characteristic curve (ROC) analysis and compared area under curve (AUC) using the Delong test. Decision curve analysis (DCA) was used to assess the potential clinical benefit of the models to patients.

RESULTS

There was no significant difference in AUC between the clinical model and the four radiomic models (P > 0.05). The XGBoost joint model had better predictive efficacy and stability (AUC, training set: 0.988, test set: 0.959). Its AUC in the test set was significantly higher than that of the clinical model (P < 0.05). DCA showed that the XGBoost joint model achieved higher net clinical benefit compared to the clinical model in the threshold probability range (0.2-0.6).

CONCLUSION

Radiomic features based on AR are objective and reproducible. The joint model combining radiomic features and clinical signs has good surgical predictive efficacy and may be an important method to help primary neonatal surgeons assess the surgical risk of NEC neonates.