A Data-Driven Algorithm Integrating Clinical and Laboratory Features for the Diagnosis and Prognosis of Necrotizing Enterocolitis

Abstract

BackgroundNecrotizing enterocolitis (NEC) is a major source of neonatal morbidity and mortality. Since there is no specific diagnostic test or risk of progression model available for NEC, the diagnosis and outcome prediction of NEC is made on clinical grounds. The objective in this study was to develop and validate new NEC scoring systems for automated staging and prognostic forecasting.Study designA six-center consortium of university based pediatric teaching hospitals prospectively collected data on infants under suspicion of having NEC over a 7-year period. A database comprised of 520 infants was utilized to develop the NEC diagnostic and prognostic models by dividing the entire dataset into training and testing cohorts of demographically matched subjects. Developed on the training cohort and validated on the blind testing cohort, our multivariate analyses led to NEC scoring metrics integrating clinical data.ResultsMachine learning using clinical and laboratory results at the time of clinical presentation led to two NEC models: (1) an automated diagnostic classification scheme; (2) a dynamic prognostic method for risk-stratifying patients into low, intermediate and high NEC scores to determine the risk for disease progression. We submit that dynamic risk stratification of infants with NEC will assist clinicians in determining the need for additional diagnostic testing and guide potential therapies in a dynamic manner.Algorithm availability http://translationalmedicine.stanford.edu/cgi-bin/NEC/index.pl and smartphone application upon request.

Current Research on the Epidemiology, Pathogenesis, and Management of Necrotizing Enterocolitis

“…A large multi-institutional study from the USA has attempted to allocate NEC infants into low, intermediate and high risk on the basis of routinely acquired clinical data, including radiological and laboratory findings. Their machine-learning model performs well, and the authors suggest that dynamic risk stratification will assist in determining the need for additional diagnostic testing and guide potential therapies [12]. …”

Early prediction of severity of necrotizing enterocolitis in preterm infants: A composite scoring tool

“…Published by Sciedu Press to be significant by Ji et al [26] This lack of association between respiratory distress and gastric residuals and severity of NEC observed by Ji et al could be related to inclusion of multiple factors in their study that may be collinear relationship to these clinical measures, and may explain the loss of statistical significance in multivariate analysis. Further, there is difference in the timing when the scores were performed.…”

“…Recently, Ji et al [26] devised a computer-based algorithm utilizing demographic, clinical, radiological, laboratory and feeding parameters to develop a risk stratification model in an attempt to predict disease severity at the presentation of NEC. Clinical features such as respiratory distress and gastric residuals, observed to be different between the 2 study groups in our study, and in prior reports [13] were not observed…”

Predicting Mortality or Intestinal Failure in Infants with Surgical Necrotizing Enterocolitis

“…Of note, the AUC of this model (0.75; 95%CI 0.74–0.77) was similar to the AUC of our hybrid model in the validation cohort for the composite outcome of death or IF (0.78; 95% CI 0.66–0.89). Another tool is the Stanford NEC scoring system, which has been used to guide staging of infants at initial presentation of NEC (24). The model uses variables, such as the presence of metabolic acidosis, portal venous gas, and abdominal wall discoloration, to predict the severity of NEC at the timing of onset of clinical symptoms.…”

“…One strength of this study is the identification and exclusion of patients with SIP, who are typically not at risk of IF and have a lower mortality rate than infants with NEC (24, 26). This is an important consideration as up to 20% of patients in national surgical NEC datasets may actually have SIP and thus reported outcomes may be biased (24).…”