Michael C Spaeder1,2, J Randall Moorman3,4,5,6,7, Christine A Tran8, Jessica Keim-Malpass3,9, Jenna V Zschaebitz10, Douglas E Lake3,5,11, Matthew T Clark3,4. 1. Department of Pediatrics, Division of Pediatric Critical Care, University of Virginia School of Medicine, Charlottesville, VA, USA. ms7uw@virginia.edu. 2. Center for Advanced Medical Analytics, University of Virginia, Charlottesville, VA, USA. ms7uw@virginia.edu. 3. Center for Advanced Medical Analytics, University of Virginia, Charlottesville, VA, USA. 4. Advanced Medical Predictive Devices, Diagnostics and Displays Inc., Charlottesville, VA, USA. 5. Department of Medicine, Division of Cardiovascular Medicine, University of Virginia School of Medicine, Charlottesville, VA, USA. 6. Department of Biomedical Engineering, University of Virginia, Charlottesville, VA, USA. 7. Department of Molecular Physiology, University of Virginia, Charlottesville, VA, USA. 8. University of Virginia School of Medicine, Charlottesville, VA, USA. 9. University of Virginia School of Nursing, Charlottesville, VA, USA. 10. Department of Pediatrics, Division of Pediatric Critical Care, University of Virginia School of Medicine, Charlottesville, VA, USA. 11. Department of Statistics, University of Virginia, Charlottesville, VA, USA.
Abstract
BACKGROUND: Early recognition of patients at risk for sepsis is paramount to improve clinical outcomes. We hypothesized that subtle signatures of illness are present in physiological and biochemical time series of pediatric-intensive care unit (PICU) patients in the early stages of sepsis. METHODS: We developed multivariate models in a retrospective observational cohort to predict the clinical diagnosis of sepsis in children. We focused on age as a predictor and asked whether random forest models, with their potential for multiple cut points, had better performance than logistic regression. RESULTS: One thousand seven hundred and eleven admissions for 1425 patients admitted to a mixed cardiac and medical/surgical PICU were included. We identified, through individual chart review, 187 sepsis diagnoses that were not within 14 days of a prior sepsis diagnosis. Multivariate models predicted sepsis in the next 24 h: cross-validated C-statistic for logistic regression and random forest were 0.74 (95% confidence interval (CI): 0.71-0.77) and 0.76 (95% CI: 0.73-0.79), respectively. CONCLUSIONS: Statistical models based on physiological and biochemical data already available in the PICU identify high-risk patients up to 24 h prior to the clinical diagnosis of sepsis. The random forest model was superior to logistic regression in capturing the context of age.
BACKGROUND: Early recognition of patients at risk for sepsis is paramount to improve clinical outcomes. We hypothesized that subtle signatures of illness are present in physiological and biochemical time series of pediatric-intensive care unit (PICU) patients in the early stages of sepsis. METHODS: We developed multivariate models in a retrospective observational cohort to predict the clinical diagnosis of sepsis in children. We focused on age as a predictor and asked whether random forest models, with their potential for multiple cut points, had better performance than logistic regression. RESULTS: One thousand seven hundred and eleven admissions for 1425 patients admitted to a mixed cardiac and medical/surgical PICU were included. We identified, through individual chart review, 187 sepsis diagnoses that were not within 14 days of a prior sepsis diagnosis. Multivariate models predicted sepsis in the next 24 h: cross-validated C-statistic for logistic regression and random forest were 0.74 (95% confidence interval (CI): 0.71-0.77) and 0.76 (95% CI: 0.73-0.79), respectively. CONCLUSIONS: Statistical models based on physiological and biochemical data already available in the PICU identify high-risk patients up to 24 h prior to the clinical diagnosis of sepsis. The random forest model was superior to logistic regression in capturing the context of age.