Variational Disentanglement for Rare Event Modeling.

Combining the increasing availability and abundance of healthcare data and the current advances in machine learning methods have created renewed opportunities to improve clinical decision support systems. However, in healthcare risk prediction applications, the proportion of cases with the condition (label) of interest is often very low relative to the available sample size. Though very prevalent in healthcare, such imbalanced classification settings are also common and challenging in many other scenarios. So motivated, we propose a variational disentanglement approach to semi-parametrically learn from rare events in heavily imbalanced classification problems. Specifically, we leverage the imposed extreme-distribution behavior on a latent space to extract information from low-prevalence events, and develop a robust prediction arm that joins the merits of the generalized additive model and isotonic neural nets. Results on synthetic studies and diverse real-world datasets, including mortality prediction on a COVID-19 cohort, demonstrate that the proposed approach outperforms existing alternatives.

Asthma and the Risk of SARS-CoV-2 Infection Among Children and Adolescents.

OBJECTIVES: Over 6 million pediatric severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections have occurred in the United States, but risk factors for infection remain poorly defined. We sought to evaluate the association between asthma and SARS-CoV-2 infection risk among children. METHODS: We conducted a retrospective cohort study of children 5 to 17 years of age receiving care through the Duke University Health System and who had a Durham County, North Carolina residential address. Children were classified as having asthma using previously validated electronic health record-based definitions. SARS-CoV-2 infections were identified based on positive polymerase chain reaction testing of respiratory samples collected between March 1, 2020, and September 30, 2021. We matched children with asthma 1:1 to children without asthma, using propensity scores and used Poisson regression to evaluate the association between asthma and SARS-CoV-2 infection risk. RESULTS: Of 46 900 children, 6324 (13.5%) met criteria for asthma. Children with asthma were more likely to be tested for SARS-CoV-2 infection than children without asthma (33.0% vs 20.9%, P < .0001). In a propensity score-matched cohort of 12 648 children, 706 (5.6%) children tested positive for SARS-CoV-2 infection, including 350 (2.8%) children with asthma and 356 (2.8%) children without asthma (risk ratio: 0.98, 95% confidence interval: 0.85-1.13. There was no evidence of effect modification of this association by inhaled corticosteroid prescription, history of severe exacerbation, or comorbid atopic diseases. Only 1 child with asthma required hospitalization for SARS-CoV-2 infection. CONCLUSIONS: After controlling for factors associated with SARS-CoV-2 testing, we found that children with asthma have a similar SARS-CoV-2 infection risk as children without asthma.

Reduced pediatric urgent asthma utilization and exacerbations during the COVID-19 pandemic.

The COVID-19 pandemic has had a profound impact on healthcare access and utilization, which could have important implications for children with chronic diseases, including asthma. We sought to evaluate changes in healthcare utilization and outcomes in children with asthma during the COVID-19 pandemic. We used electronic health records data to evaluate healthcare use and asthma outcomes in 3959 children and adolescents, 5-17 years of age, with a prior diagnosis of asthma who had a history of well-child visits and encounters within the healthcare system. We assessed all-cause healthcare encounters and asthma exacerbations in the 12-months preceding the start of the COVID-19 pandemic (March 1, 2019-February 29, 2020) and the first 12 months of the pandemic (March 1, 2020-February 28, 2021). All-cause healthcare encounters decreased significantly during the pandemic compared to the preceding year, including well-child visits (48.1% during the pandemic vs. 66.6% in the prior year; p < .01), emergency department visits (9.7% vs. 21.0%; p < .01), and inpatient admissions (1.6% vs. 2.5%; p < .01), though there was over a 100-fold increase in telehealth encounters. Asthma exacerbations that required treatment with systemic steroids also decreased (127 vs. 504 exacerbations; p < .01). Race/ethnicity was not associated with changes in healthcare utilization or asthma outcomes. The COVID-19 pandemic corresponded to dramatic shifts in healthcare utilization, including increased telehealth use and improved outcomes among children with asthma. Social distancing measures may have also reduced asthma trigger exposure.

Development and Performance of a Clinical Decision Support Tool to Inform Resource Utilization for Elective Operations.

Importance: Hospitals ceased most elective procedures during the height of coronavirus disease 2019 (COVID-19) infections. As hospitals begin to recommence elective procedures, it is necessary to have a means to assess how resource intensive a given case may be. Objective: To evaluate the development and performance of a clinical decision support tool to inform resource utilization for elective procedures. Design, Setting, and Participants: In this prognostic study, predictive modeling was used on retrospective electronic health records data from a large academic health system comprising 1 tertiary care hospital and 2 community hospitals of patients undergoing scheduled elective procedures from January 1, 2017, to March 1, 2020. Electronic health records data on case type, patient demographic characteristics, service utilization history, comorbidities, and medications were and abstracted and analyzed. Data were analyzed from April to June 2020. Main Outcomes and Measures: Predicitons of hospital length of stay, intensive care unit length of stay, need for mechanical ventilation, and need to be discharged to a skilled nursing facility. These predictions were generated using the random forests algorithm. Predicted probabilities were turned into risk classifications designed to give assessments of resource utilization risk. Results: Data from the electronic health records of 42 199 patients from 3 hospitals were abstracted for analysis. The median length of stay was 2.3 days (range, 1.3-4.2 days), 6416 patients (15.2%) were admitted to the intensive care unit, 1624 (3.8%) received mechanical ventilation, and 2843 (6.7%) were discharged to a skilled nursing facility. Predictive performance was strong with an area under the receiver operator characteristic ranging from 0.76 to 0.93. Sensitivity of the high-risk and medium-risk groupings was set at 95%. The negative predictive value of the low-risk grouping was 99%. We integrated the models into a daily refreshing Tableau dashboard to guide decision-making. Conclusions and Relevance: The clinical decision support tool is currently being used by surgical leadership to inform case scheduling. This work shows the importance of a learning health care environment in surgical care, using quantitative modeling to guide decision-making.