Search

This study investigates the safety and efficacy of a large-dose, needle-based epidural technique in obstetric anesthesia. The technique differs from a standard, catheter-based approach in that the anesthetic dose is administered through an epidural needle prior to insertion of the epidural catheter. Using a data-driven informatics and machine learning approach, our findings show that the needle-based technique is faster and more dose-effective in achieving sensory level. We also find that injecting large doses in the epidural space through the epidural needle is safe, with complication rates similar to those reported in published literature for catheter-based technique. Further, machine learning reveals that if the needle dose is kept under 18 ml, the resulting hypotension rate will be significantly lower than published results. The machine learning framework can predict the incidence of hypotension with 85% accuracy. The findings from this investigation facilitate delivery improvement and establish an improved clinical practice guideline for training and for dissemination of safe practice.

The ability to predict how different individuals will respond to vaccination and to understand what best protects individuals from infection greatly facilitates developing next-generation vaccines. It facilitates both the rapid [...]

Introduction: CSF levels of established Alzheimer's disease (AD) biomarkers remain stable despite disease progression, and non-amyloid non-tau biomarkers have the potential of informing disease stage and progression. We previously identified complement 3 (C3) to be decreased in AD dementia, but this change was not found by others in earlier AD stages. We hypothesized that levels of C3 and associated factor H (FH) can potentially distinguish between mild cognitive impairment (MCI) and dementia stages of AD, but we also found their levels to be influenced by age and disease status. Results: We developed a biochemical/bioinformatics pipeline to optimize the handling of complex interactions between variables in validating biochemical markers of disease. We used data from the Alzheimer's Disease Neuroimaging Initiative (ADNI, n = 230) to build parallel machine learning models, and objectively tested the models in a test cohort (n = 73) of MCI and mild AD patients independently recruited from Emory University. Whereas models incorporating age, gender, APOE ε4 status, and CSF amyloid and tau levels failed to reliably distinguish between MCI and mild AD in ADNI, introduction of CSF C3 and FH levels reproducibly improved the distinction between the two AD stages in ADNI (p < 0.05) and the Emory cohort (p = 0.014). Within each AD stage, the final model also distinguished between fast vs. slower decliners (p < 0.001 for MCI, p = 0.007 for mild AD), with lower C3 and FH levels associated with more advanced disease and faster progression. Conclusions: We propose that CSF C3 and FH alterations may reflect stage-associated biomarker changes in AD, and can complement clinician diagnosis in diagnosing and staging AD using the publically available ADNI database as reference. [ABSTRACT FROM AUTHOR]

When we encounter an unexpected critical health problem, a hospital's emergency department (ED) becomes our vital medical resource. Improving an ED's timeliness of care, quality of care, and operational efficiency while reducing avoidable readmissions, is fraught with difficulties, which arise from complexity and uncertainty. In this paper, we describe an ED decision support system that couples machine learning, simulation, and optimization to address these improvement goals. The system allows healthcare administrators to globally optimize workflow, taking into account the uncertainties of incoming patient injuries and diseases and their associated care, thereby significantly reducing patient length of stay. This is achieved without changing physical layout, focusing instead on process consolidation, operations tracking, and staffing. First implemented at Grady Memorial Hospital in Atlanta, Georgia, the system helped reduce length of stay at Grady by roughly 33 percent. By repurposing existing resources, the hospital established a clinical decision unit that resulted in a 28 percent reduction in ED readmissions. Insights gained from the implementation also led to an investment in a walkin center that eliminated more than 32 percent of the nonurgent-care cases from the ED. As a result of these improvements, the hospital enhanced its financial standing and achieved its target goal of an average ED length of stay of close to seven hours. ED and trauma efficiencies improved throughput by over 16 percent and reduced the number of patients who left without being seen by more than 30 percent. The annual revenue realized plus savings generated are approximately $190 million, a large amount relative to the hospital's $1.5 billion annual economic impact. The underlying model, which we generalized, has been tested and implemented successfully at 10 other EDs and in other hospital units. The system offers significant advantages in that it permits a comprehensive analysis of the entire patient flow from registration to discharge, enables a decision maker to understand the complexities and interdependencies of individual steps in the process sequence, and ultimately allows the users to perform system optimization. [ABSTRACT FROM AUTHOR]

Classification is concerned with the development of rules for the allocation of observations to groups, and is a fundamental problem in machine learning. Much of previous work on classification models investigates two-group discrimination. Multi-category classification is less-often considered due to the tendency of generalizations of two-group models to produce misclassification rates that are higher than desirable. Indeed, producing “good” two-group classification rules is a challenging task for some applications, and producing good multi-category rules is generally more difficult. Additionally, even when the “optimal” classification rule is known, inter-group misclassification rates may be higher than tolerable for a given classification model. We investigate properties of a mixed-integer programming based multi-category classification model that allows for the pre-specification of limits on inter-group misclassification rates. The mechanism by which the limits are satisfied is the use of a reserved judgment region, an artificial category into which observations are placed whose attributes do not sufficiently indicate membership to any particular group. The method is shown to be a consistent estimator of a classification rule with misclassification limits, and performance on simulated and real-world data is demonstrated. [ABSTRACT FROM AUTHOR]