Reassessing acquired neonatal intestinal diseases using unsupervised machine learning

BACKGROUND

Acquired neonatal intestinal diseases have an array of overlapping presentations and are often labeled under the dichotomous classification of necrotizing enterocolitis (which is poorly defined) or spontaneous intestinal perforation, hindering more precise diagnosis and research. The objective of this study was to take a fresh look at neonatal intestinal disease classification using unsupervised machine learning.

METHODS

Patients admitted to the University of Florida Shands Neonatal Intensive Care Unit January 2013-September 2019 diagnosed with an intestinal injury, or had imaging findings of portal venous gas, pneumatosis, abdominal free air, or had an abdominal drain placed or exploratory laparotomy during admission were included. Congenital gastroschisis, omphalocele, intestinal atresia, malrotation were excluded. Data was collected via retrospective chart review with subsequent hierarchal, unsupervised clustering analysis.

RESULTS

Five clusters of intestinal injury were identified: Cluster 1 deemed the "Low Mortality" cluster, Cluster 2 deemed the "Mature with Inflammation" cluster, Cluster 3 deemed the "Immature with High Mortality" cluster, Cluster 4 deemed the "Late Injury at Full Feeds" cluster, and Cluster 5 deemed the "Late Injury with High Rate of Intestinal Necrosis" cluster.

CONCLUSION

Unsupervised machine learning can be used to cluster acquired neonatal intestinal injuries. Future study with larger multicenter datasets is needed to further refine and classify types of intestinal diseases.

IMPACT

Unsupervised machine learning can be used to cluster types of acquired neonatal intestinal injury. Five major clusters of acquired neonatal intestinal injury are described, each with unique features. The clusters herein described deserve future, multicenter study to determine more specific early biomarkers and tailored therapeutic interventions to improve outcomes of often devastating neonatal acquired intestinal injuries.

Case Reports of Cow's Milk Protein Allergy Presenting as Delayed Passage of Meconium With Early Onset Infant Constipation

A cellular proliferation to milk allergens has been found in the cord blood cells of neonates. While this reflects a sensitivity during the fetal life, its clinical significance and disease, particularly its unconventional presentations, have remained largely unrecognized by care providers. Here, we report three cases of infants whose mothers consumed dairy products during pregnancy, who developed a severely constipated pre- and postnatal bowel. The passage of meconium was significantly delayed with subsequent early-onset infant constipation that was intractable to conventional therapies but remitted when milk proteins were withheld, recurred when milk proteins were reintroduced, and resolved again when switched to an extensively hydrolyzed or amino acid-based infant formula. Based on this and other observations, it is believed that these infants must have initiated and/or developed cow's milk protein allergy prenatally during fetal life. We suggest that a 2-week trial of cow's milk protein avoidance be applied to these neonate infants with early-onset constipation before an unnecessary invasive work-up for Hirschsprung disease and others is initiated per the current guidelines.

Hourly Kinetics of Critical Organ Dysfunction in Extremely Preterm Infants

Rationale: Use of severity of illness scores to classify patients for clinical care and research is common outside of the neonatal ICU. Extremely premature (<29 weeks' gestation) infants with extremely low birth weight (<1,000 g) experience significant mortality and develop severe pathology during the protracted birth hospitalization. Objectives: To measure at high resolution the changes in organ dysfunction that occur from birth to death or discharge home by gestational age and time, and among extremely preterm infants with and without clinically meaningful outcomes using the neonatal sequential organ failure assessment score. Methods: A single-center, retrospective, observational cohort study of inborn, extremely preterm infants with extremely low birth weight admitted between January 2012 and January 2020. Neonatal sequential organ failure assessment scores were calculated every hour for every patient from admission until death or discharge. Measurements and Main Results: Longitudinal, granular scores from 436 infants demonstrated early and sustained discrimination of those who died versus those who survived to discharge. The discrimination for mortality by the maximum score was excellent (area under curve, 0.91; 95% confidence intervals, 0.88-0.94). Among survivors with and without adverse outcomes, most score variation occurred at the patient level. The weekly average score over the first 28 days was associated with the sum of adverse outcomes at discharge. Conclusions: The neonatal sequential organ failure assessment score discriminates between survival and nonsurvival on the first day of life. The major contributor to score variation occurred at the patient level. There was a direct association between scores and major adverse outcomes, including death.

Using machine learning analysis to assist in differentiating between necrotizing enterocolitis and spontaneous intestinal perforation: A novel predictive analytic tool

PURPOSE

Necrotizing enterocolitis (NEC) and spontaneous intestinal perforation (SIP) are devastating diseases in preterm neonates, often requiring surgical treatment. Previous studies evaluated outcomes in peritoneal drain placement versus laparotomy, but the accuracy of the presumptive diagnosis remains unknown without bowel visualization. Predictive analytics provide the opportunity to determine the etiology of perforation and guide surgical decision making. The purpose of this investigation was to build and evaluate machine learning models to differentiate NEC and SIP.

METHODS

Neonates who underwent drain placement or laparotomy NEC or SIP were identified and grouped definitively via bowel visualization. Patient characteristics were analyzed using machine learning methodologies, which were optimized through areas under the receiver operating characteristic curve (AUROC). The model was further evaluated using a validation cohort.

RESULTS

40 patients were identified. A random forest model achieved 98% AUROC while a ridge logistic regression model reached 92% AUROC in differentiating diseases. When applying the trained random forest model to the validation cohort, outcomes were correctly predicted.

CONCLUSIONS

This study supports the feasibility of using a novel machine learning model to differentiate between NEC and SIP prior to any intended surgical interventions.

LEVEL OF EVIDENCE

level II TYPE OF STUDY: Clinical Research Paper.

17-Month-Old Girl With Severe, Prolonged Lethargy and Somnolence

A 17-month-old girl arrived at the pediatric ED with decreased responsiveness. She was lethargic, localizing only to noxious stimuli with vital signs significant for fever of 103.8 °F, heart rate of 185 beats/min, respiratory rate of 12 breaths/min, blood pressure of 100/59 mmHg, and oxygen saturation level of 88% on room air. She was admitted to the pediatric intensive care unit (PICU) due to concerns of septic meningitis with altered mental status and respiratory distress, and was treated with antibiotics. A respiratory viral panel (RVP) was positive for adenovirus, resulting in all antibiotics being discontinued. She remained lethargic until day nine of illness, when she had improved almost completely to her baseline. Polymerase chain reaction (PCR) of her cerebral spinal fluid returned positive for adenovirus serotype A, thus confirming our case of transient adenovirus encephalopathy. This case illustrates the importance of keeping adenovirus in the differential for encephalopathy versus a neurologic abnormality or other malignant or infectious etiology.

Maximum vasoactive-inotropic score and mortality in extremely premature, extremely low birth weight infants

OBJECTIVE

To determine the relationship between maximum vasoactive-inotropic (VIS^max) and mortality in extremely premature (<29 weeks completed gestation), extremely low birth weight (ELBW, <1000 g) infants.

STUDY DESIGN

Single center, retrospective, and observational cohort study.

RESULTS

We identified 436 ELBW, <29 week, inborn infants cared for during the study period. Compared to infants with VIS^max of 0, the frequency of mortality based on VIS^max ranged from 3.3-fold to 46.1-fold. VIS^max > 30 was associated with universal mortality. Multivariable modeling that included gestational age, birth weight, and VIS^max revealed significant utility to predict mortality with negative predictive value of 87.0% and positive predictive value of 84.8% [adjusted AUROC: 0.90, (0.86-0.94)] among patients that received vasoactive-inotropic treatment.

CONCLUSION

VIS^max is an objective measure of hemodynamic/cardiovascular support that was directly associated with mortality in extremely premature ELBW infants. The VIS^max represents an important step towards neonatal precision medicine and risk stratification of extremely premature ELBW infants.

Abstract TP92: Prevalence of Aortic Aneurysms in Patients With Intracranial Aneurysms: A Retrospective Analysis

Introduction: Intracranial (IA) and aortic aneurysms (AA) share genetic and environmental risks. In a retrospective review of clinical data, we identified those presenting with IA (+/- subarachnoid hemorrhage), who also had abdominal or thoracic imaging.

Method: In the University of Virginia (UVA) Clinical Data Repository we searched for patients with IA by ICD-9 diagnosis codes and CPT treatment codes. We used three strategies to identify those with aortic imaging: 1) ICD-9 diagnosis codes for AA, 2) CPT and billing codes for abdominal or thoracic imaging (aortogram, ultrasound, angiography, magnetic resonance, and computerized tomography), and 3) CPT and billing codes for AA treatment (endovascular, stent, graft, wrap, or screening). We reviewed all charts to identify those with confirmed IA and imaging. We performed a multivariable logistic regression analysis accounting for sex, age, hypertension, diabetes mellitus, smoking status, and IA size, multiplicity, and location to look for associations with AA.

Results: Among individuals seen at UVA hospital from 2004 to 2015, we identified 13245 cases with a possible IA, of whom 1017 potential aortic imaging. Our review of charts revealed that 720 did not having an IA. Of the remaining 287 cases, 51 lacked appropriate aortic imaging. We recorded information from the remaining 236 cases - 94/236 (39.8%) had an AA identified. In our multivariable analysis, only female sex was significantly associated with a co-prevalent AA (Odds Ratio: 0.20, 95% confidence interval 0.09 - 0.45). In the table, we provide descriptive statistics for those with and without AA as well as the subsets with thoracic aortic aneurysms (TAA) and abdominal aortic aneurysms (AAA).

Conclusion: The co-prevalence of AA in a clinical population known to have IA and with aortic imaging was 39.8%. Men had higher risk for AA. This indicates a shared risk for aneurismal disease in the aorta and brain warranting prospective investigation on the mechanisms of risk.