BACKGROUND AND OBJECTIVES: The decline in hospital mortality in children hospitalized with sepsis has increased the number of survivors. These survivors are at risk for adverse long-term outcomes, including readmission and recurrent or unresolved infections. We described the epidemiology of 90-day readmissions after sepsis hospitalization in children. We tested the hypothesis that a sepsis hospitalization increases odds of 90-day readmissions. METHODS: Retrospective cohort analysis of the Nationwide Readmissions Database. We included index unplanned admissions of non-neonatal pediatric patients and described the proportion of readmissions, including those involving infection or sepsis. We performed multivariable analysis to determine the odds of readmission after a sepsis and nonsepsis admission and compared costs of readmission after sepsis and nonsepsis admissions. RESULTS: Of 562 817 pediatric admissions, 7634 (1.4%) and 555 183 (98.6%) were discharged alive after admissions with and without sepsis. The rate of 90-day readmission after sepsis was 21.4%: 7.2% and 25.5% in previously healthy and chronically ill patients. The adjusted mean cost during readmission was $7385. Half of readmissions (52.9%) involved recurrent infection or sepsis. Sepsis admissions were associated with higher odds of readmission at 90 days compared with nonsepsis admissions (adjusted odds ratio 1.15, 95% confidence interval 1.08–1.23). The results remained unchanged for 30-day and 6-month readmissions. CONCLUSIONS: Readmissions occur after 1 in 5 pediatric sepsis hospitalizations and increase health care costs. Sepsis hospitalization increased odds of readmission and commonly involved recurrent infection or sepsis. Clinicians caring for these patients should consider surveillance for recurrent or unresolved infection, and researchers should explore underlying mechanisms and potential interventions to reduce readmissions.
Supercomputers are complex systems producing vast quantities of performance data from multiple sources and of varying types. Performance data from each of the thousands of nodes in a supercomputer tracks multiple forms of storage, memory, networks, processors, and accelerators. Optimization of application performance is critical for cost effective usage of a supercomputer and requires efficient methods for effectively viewing performance data. The combination of supercomputing analytics and 3D gaming visualization enables real-time processing and visual data display of massive amounts of information that humans can process quickly with little training. Our system fully utilizes the capabilities of modern 3D gaming environments to create novel representations of computing hardware which intuitively represent the physical attributes of the supercomputer while displaying real-time alerts and component utilization. This system allows operators to quickly assess how the supercomputer is being used, gives users visibility into the resources they are consuming, and provides instructors new ways to interactively teach the computing architecture concepts necessary for efficient computing.
2 August 2023 in Journal of Intensive Care Medicine
Introduction: Sepsis is more common in males than females, but whether outcomes differ by sex in various pediatric age groups is unclear. The Life After Pediatric Sepsis Evaluation (LAPSE) was a multicenter prospective cohort study that evaluated health-related quality of life (HRQL) in children after community-acquired septic shock. In this secondary analysis, we evaluated whether male children are at increased risk of mortality or long-term decline in HRQL than female children by age group. Methods: Children (1 month–18 years) with community-acquired septic shock were recruited from 12 pediatric intensive care units in the U.S. Data included sex, age group (<1 year, 1–<13 years, 13–18 years), acute illness severity (acute organ dysfunction and inflammation), and longitudinal assessments of HRQL and mortality. Persistent decline in HRQL was defined as a 10% decrease in HRQL comparing baseline to 3 months following admission. Male and female children were stratified by age group and compared to evaluate the difference in the composite outcome of death or persistent decline in HRQL using the Cochran–Mantel–Haenszel test. Results: Of 389 children, 54.2% ( n = 211) were male. Overall, 10% (21/211) of males and 12% (22/178) of females died by 3 months ( p = 0.454). Among children with follow-up data, 41% (57/138) of males and 44% (48/108) of females died or had persistent decline in HRQL at 3 months ( p = 0.636), with no observed difference by sex when stratified by age group. There was no significant difference in acute illness severity between males and females overall or stratified by age group. Conclusions: In this secondary analysis of the LAPSE cohort, HRQL, and mortality were not different between male and female children when stratified by age group. There were no significant differences by sex across multiple measures of illness severity or treatment intensity.
Artificial Intelligence/Machine Learning applications require the training of complex models on large amounts of labelled data. The large computational requirements for training deep models have necessitated the development of new methods for faster training. One such approach is the data parallel approach, where the training data is distributed across multiple compute nodes. This approach is simple to implement and supported by most of the commonly used machine learning frameworks. The data parallel approach leverages MPI for communicating gradients across all nodes. In this paper, we examine the effects of using different physical hardware interconnects and network-related software primitives for enabling data distributed deep learning. We compare the effect of using GPUDirect and NCCL on Ethernet and OmniPath fabrics. Our results show that using Ethernet-based networking in shared HPC systems does not have a significant effect on the training times for commonly used deep neural network architectures or traditional HPC applications such as Computational Fluid Dynamics.