Development and utility assessment of a machine learning bloodstream infection classifier in pediatric patients receiving cancer treatments.

BACKGROUND: Objectives were to build a machine learning algorithm to identify bloodstream infection (BSI) among pediatric patients with cancer and hematopoietic stem cell transplantation (HSCT) recipients, and to compare this approach with presence of neutropenia to identify BSI. METHODS: We included patients 0-18 years of age at cancer diagnosis or HSCT between January 2009 and November 2018. Eligible blood cultures were those with no previous blood culture (regardless of result) within 7 days. The primary outcome was BSI. Four machine learning algorithms were used: elastic net, support vector machine and two implementations of gradient boosting machine (GBM and XGBoost). Model training and evaluation were performed using temporally disjoint training (60%), validation (20%) and test (20%) sets. The best model was compared to neutropenia alone in the test set. RESULTS: Of 11,183 eligible blood cultures, 624 (5.6%) were positive. The best model in the validation set was GBM, which achieved an area-under-the-receiver-operator-curve (AUROC) of 0.74 in the test set. Among the 2236 in the test set, the number of false positives and specificity of GBM vs. neutropenia were 508 vs. 592 and 0.76 vs. 0.72 respectively. Among 139 test set BSIs, six (4.3%) non-neutropenic patients were identified by GBM. All received antibiotics prior to culture result availability. CONCLUSIONS: We developed a machine learning algorithm to classify BSI. GBM achieved an AUROC of 0.74 and identified 4.3% additional true cases in the test set. The machine learning algorithm did not perform substantially better than using presence of neutropenia alone to predict BSI.

Heart rate variability as an indicator for morphine pharmacokinetics and pharmacodynamics in critically ill newborn infants.

Morphine is the commonest drug used for analgesia in newborn infants. It is a natural opioid that acts as an agonist at the mu and kappa receptors, which are receptors for analgesia and sedation. Morphine pharmacokinetics and pharmacodynamics (PKPD) for the newborn infant population are not well understood. The objective of this study is to use morphine PKPD parameters to estimate morphine plasma concentrations to be correlated with heart rate variability in the neonatal population.

Neonatal short bowel syndrome: a cohort study.

BACKGROUND: To date, our knowledge of morbidity and mortality in neonatal short bowel syndrome (SBS) is based on individual case series. Shortcomings of the published literature include long patient recruitment time, selection bias, variable SBS definitions, failure to account for gestational age, and incomplete follow-up. By applying more rigorous methodology, our aim was to determine outcomes of SBS neonates compared with a control group of neonates without SBS. METHODS: A cohort study of all neonates with abdominal pathology requiring laparotomy between January 1, 1997, and December 31, 1998, with observation through July 1, 2001. Short bowel syndrome was defined as patients requiring parenteral nutrition for more than 42 days or residual small bowel length of less than 25% predicted by gestational age. Student's t test, Mann-Whitney U test, and chi2 were used where appropriate. Kaplan-Meier curves were used to determine cumulative survival. Covariates important in the development of SBS were examined using forward step-wise logistic regression. RESULTS: There were 175 patients (with SBS = 40, without SBS = 135) with a mean gestational age of 30.7 +/- 4.6 weeks vs 35.9 +/- 4.8 weeks, respectively (P < .0005). The patients with SBS suffered significantly more morbidity than the group without SBS in all categories of investigation (surgical complications, septic events, central venous line complications, duration to adaptation and parenteral nutrition independence, cholestasis and liver failure, and duration of hospitalization). The case fatality rate was 37.5% in patients with SBS vs 13.3% in patients without SBS (P = .001). Most of the deaths were caused by liver failure or sepsis and occurred within 1 year from the date of surgery. Presence of an ileostomy (exp(B) = 12.29; P < .0005) and a residual small bowel length less than 50% of the original length (exp(B) = 26.84; P < .0005) were the only 2 variables in a logistic regression analysis found to be independently associated with the development of SBS. CONCLUSION: This cohort study clearly illustrates the tremendous morbidity experienced by infants with SBS relative to other surgical neonates. Accurate estimates of the morbidity associated with SBS enables clinicians to appropriately counsel parents, allocate resources and initiate therapeutic trials.

Neonatal short bowel syndrome: population-based estimates of incidence and mortality rates.

BACKGROUND: Congenital or acquired neonatal short bowel syndrome (SBS) carries significant morbidity and mortality rates. No accurate population estimates of incidence and mortality exist because of differences in definition, follow-up, and regional referral patterns. METHODS: A retrospective cohort study was performed involving 175 surgical neonates admitted to our institution from January 1, 1997 to December 31, 1999 and followed up until July 1, 2001. Institution and population-based estimates of incidence and mortality were performed using postcensal population figures (1997) from Statistics Canada. RESULTS: The overall incidence of SBS was 22.1 per 1,000 neonatal intensive care unit (NICU) admissions (95% CI = 15.3, 28.9) and 24.5 per 100,000 live births (95% CI = 12.1, 36.9). The incidence was much greater in premature infants (less than 37 weeks). The SBS case fatality rate was 37.5% (95% CI = 22.5, 52.5) and the cause-specific and proportional mortality rates (for children less than 4 years old) were 2.0 of 100,000 population per year (0.4 to 3.6/100,000/year) and 1.4% (0.3% to 2.6%), respectively. CONCLUSIONS: Patients with neonatal SBS pose a complex management challenge and are responsible for a significant cost to the health care system. To our knowledge, this study represents the first population-based estimates for neonatal SBS incidence and mortality rates. Accurate estimates will assist clinicians in counseling parents, allocating resources, and planning clinical trials.