学科服务 更多>>

关键词： Data analytics   Data-driven decision making   Patient no-show prediction   Probabilistic risk scores   Web-based decision support system  

摘要： Patients who miss their appointments (no-shows) reduce revenues and impair the delivery of quality healthcare. Much research has been devoted to identifying "no-show patients". In this study, we build a Tree Augmented Naive Bayes (TAN)-based, probabilistic data driven methodology that consists of five steps. After data acquisition and preparation in the first step, the second step is dedicated to selecting important variables through purely data-driven wrapper methods such as Extreme Gradient Boosting (XGB), Particle Swarm Optimization (PSO) and Genetic Algorithms (GA). Then, in the third step, the Synthetic Minority Oversampling Technique (SMOTE) and Random under Sampling (RUS) are employed to overcome the data imbalance issue that exists in the dataset. In the fourth step, the conditional interrelations among the predictors are obtained via TAN model, and Bayesian-belief based posterior probabilities are calculated. It should be noted that, the parameters of the TAN model are tuned in a way that there does not have to exist a conditional interrelation between every single predictor. The consistency and reliability of these posterior probabilities are then justified/investigated via the proposed algorithm. The patients are clustered into five risk groups, and a web-based decision support system graphical user interface is built as a proof of concept. Results show that interesting conditional inter-relations exist between the seven variables that are commonly selected by GA and PSO. In addition, an overall area under the receiver operating characteristic curve (ROC) score of 0.828 was achieved with a sensitivity score of 0.785. The reliable, TAN-based posterior probabilities and conditional relationships among the predictors for such a parsimonious model that has a fairly high sensitivity in detecting the minority samples, can be adopted by primary care facilities to improve the decision-making process in managing the no-show problem. (c) 2020 Elsevier Ltd. All righ

关键词： colorectal   disease category   readmission  

摘要： Rationale The aim of this study was to identify temporal readmission patterns according to baseline disease categories to provide opportunities for targeted interventions. Methods Retrospective analysis of consecutive adult (>= 18 years) patients who underwent elective colorectal resections (2011-2017) at Mayo Clinic Rochester, MN. A prospective administrative database including patient demographics, procedure characteristics, discharge information and specifics on 30-day readmissions (to index facility) including timing and reasons was utilized. The ICD-9 codes were regrouped into the main pathologies Cancer, Crohn's disease (CD)/chronic ulcerative colitis (CUC), and diverticular disease. Results In total, 521 (7.2%) out of 7245 patients undergoing inpatient colorectal surgery were readmitted. In all increments of time from discharge (0-2 days: 31.3% of all readmissions, 3-7 days: 32.4% of all readmissions, 8-14 days: 18% of all readmissions, and 15-30 days: 18.3% of all readmissions), reasons for readmission differed significantly (all P < 0.001). Across all disease categories, early readmissions (within 2 days of discharge) were most likely due to ileus/obstruction (53.4% of early readmissions), whereas with 42.5%, infection was the most common cause for late readmissions (>7 days). Patients with home discharge were more likely to be readmitted earlier within the 30-day observation period (P = 0.099), whereas patients with a longer length of index hospital stay (>7 days) were readmitted later (P = 0.080). Conclusions Reasons for readmission appear to be universal across different disease categories. Targeted educational and collaborative measures may help to mitigate the burden of hospital readmissions to index facilities.

关键词： Aquatic insect   Current velocity   TITAN2   Flow reduction   Water flow categories  

摘要： The LIFE Index can be useful to determine the effect of changes in river flow on the community of aquatic macroinvertebrates and as a tool for the implementation of environmental flows. This study demonstrates how to classify aquatic macroinvertebrates into a flow category in order to adjust the LIFE Index to Costa Rica. A panel of experts was surveyed to classify the most common genera into a water velocity category, based on their experience. Also, for one-year, aquatic macroinvertebrates were collected in the low and middle basin of the Naranjo River under different velocities, and by using the TITAN2 package, their respective thresholds for current velocity were determined. Variation was observed in the responses of the expert panel;however, several taxa overlapped by more than 60% between the expert classification and the TITAN2 test results. The TITAN2 test assigned a velocity threshold to 32 genera, with the inflection point being 0.1 m/s. The expert panel served as a tool to assign a category to those genera that the TITAN2 test did not contemplate. Organisms without morphological adaptations to survive fast flowing conditions decreased in frequency above 0.1 m/s, while genera related to moderate and high velocities increased in frequency. Through the panel of experts and the TITAN2 test, it was possible to assign the most common genera in the country to a current velocity category and therefore adjust the LIFE index. (C) 2020 European Regional Centre for Ecohydrology of the Polish Academy of Sciences. Published by Elsevier B.V. All rights reserved.

关键词： Machine learning   Radiography   Artificial intelligence  

摘要： Objectives The goal of the present study was to classify the most common types of plain radiographs using a neural network and to validate the network's performance on internal and external data. Such a network could help improve various radiological workflows. Methods All radiographs from the year 2017 (n = 71,274) acquired at our institution were retrieved from the PACS. The 30 largest categories (n = 58,219, 81.7% of all radiographs performed in 2017) were used to develop and validate a neural network (MobileNet v1.0) using transfer learning. Image categories were extracted from DICOM metadata (study and image description) and mapped to the WHO manual of diagnostic imaging. As an independent, external validation set, we used images from other institutions that had been stored in our PACS (n = 5324). Results In the internal validation, the overall accuracy of the model was 90.3% (95%CI: 89.2-91.3%), whereas, for the external validation set, the overall accuracy was 94.0% (95%CI: 93.3-94.6%). Conclusions Using data from one single institution, we were able to classify the most common categories of radiographs with a neural network. The network showed good generalizability on the external validation set and could be used to automatically organize a PACS, preselect radiographs so that they can be routed to more specialized networks for abnormality detection or help with other parts of the radiological workflow (e.g., automated hanging protocols;check if ordered image and performed image are the same). The final AI algorithm is publicly available for evaluation and extension.

关键词： Cytomegalovirus   hematopoietic stem cell transplant   infection   primary immunodeficiency   serostatus  

摘要： Background Cytomegalovirus (CMV) serostatus of recipient (R) and donor (D) influences hematopoietic stem cell transplant (HSCT) outcome. However, it is not a reliable indicator of CMV infection in primary immunodeficiency disorder (PIDD) recipients who are unable to make adequate antigen-specific immunoglobulin (Ig) or who receive intravenous Ig (IVIg) prior to testing. Objective Since no data exist on PIDD with unknown CMV serostatus, we aimed to evaluate the relationship between pre-HSCT recipient and donor serostatus and incidence of CMV infection in recipients with unknown serostatus. Methods A retrospective analysis of all pediatric PIDD HSCTs (2007-2018) was performed at University of California San Francisco. Recipients were separated into categories based on pre-transplant serostatus: 1) seropositive (R(+)), 2) seronegative (R(-)), and 3) unknown serostatus (R(x)). Patients with pre-HSCT active CMV viremia were excluded. Results A total of 90 patients were included, 69% male. The overall incidence of CMV infection was 20%, but varied in R(+), R(-), and R(x) at 80%, 0%, and 14%, (P-value = .0001). Similarly, 5-year survival differed among groups, 60% R(+), 100% R(-), and 90% of R(x) (P-value = .0045). There was no difference in CMV reactivation by donor serostatus (P-value = .29), however, faster time to clearance of CMV was observed for R(x)/D(+) group (median 9.5 days (IQR 2.5-18), P-value = .024). Conclusion We identify a novel group of recipients, R(x), with an intermediate level of survival and CMV incidence post-HSCT, when compared to seropositive and seronegative recipients. No evidence of CMV transmission from D(+) in R(-) and R(x) was observed. We believe R(x) should be considered as a separate category in future studies to better delineate recipient risk status.

关键词： Boolean functions   Spectral analysis   Reservoir modeling   Categorical simulation  

摘要： This work introduces a new method for simulating facies distribution with two categories based on Fourier analysis of Boolean functions. According to this method, two categories of facies distributed along vertical wells are encoded as Boolean functions taking two values. The subsequent simulation process is divided into three consecutive steps. First, Boolean functions of the well data are decomposed into a binary version of a Fourier series. Decomposition coefficients are then simulated over the two-dimensional area as stationary random fields. Finally, synthetic data in the interwell space are reconstructed from simulated coefficients. The described method was implemented experimentally in software and tested on a case of a real oil field and on a case of a synthetic oil field model. Simulations on the synthetic model were used to test the performance of the method for two different bases in the Fourier expansion (Walsh functions and Haar wavelets). The simulation results were compared to those obtained on the same synthetic model via the classical sequential indicator simulation. It was shown that, for both bases, the new method reproduces statistical parameters of the well data better than sequential indicator simulation.

关键词： building‐   specific loss assessment   collapse risk   long‐   duration ground motions   seismic risk   steel moment resisting frames  

摘要： This paper investigates the effect of ground motion duration on risks to life safety and costs associated with low-to-medium-rise reference steel moment resisting frame buildings classified as risk categories II and IV. The ductile steel resisting frame buildings are designed in the downtown area of Seattle, Washington, where long-duration earthquakes are more likely to occur. For this purpose, two sets of spectrally equivalent long- and short-duration ground motions are employed to perform nonlinear dynamic analyses for two-dimensional nonlinear structural models of the steel frame buildings. It is shown that the consideration of the effects of long-duration ground motion significantly increases the collapse risk of the steel frame buildings, by a factor of about 5 on average, in comparison to the short-duration set. The destabilizing P-Delta effects are found to increasingly aggravate the influence of long-duration earthquakes on the risk to life safety of the medium-rise steel frame buildings. For frequently occurring ground motion intensities, reparable damage to nonstructural components holds most of the total expected economic seismic loss in the reference buildings, regardless of the ground motion set and assigned risk category. The influence of ground motion duration on the economic seismic loss in the steel frame buildings starts to become important for strong ground motions associated with design-level earthquakes or higher.

摘要： We will show a centrally free action of an amenable rigid C*-tensor category on a properly infinite von Neumann algebra has the Rohlin property. Our main result is the classification of centrally free cocycle actions of an amenable rigid C*-tensor category up to approximate inner conjugacy on properly infinite von Neumann algebras. This is regarded as a generalization of classification of amenable discrete groups due to A. Connes, V. Jones and A. Ocneanu. We have the following two applications: a classification of centrally free actions of amenable discrete quantum groups of Kac type on von Neumann algebras and another proof of S. Popa's celebrated classification result of amenable subfactors. As another application of the Rohlin property, we will prove the fullness of the crossed product of a full factor by a minimal action of a compact group.

关键词： Intimate partner violence (IPV)   psychological abuse/ violence   emotional abuse/ violence   physical abuse/ violence   sexual abuse/ violence   combined abuse/ violence   measurement   health  

摘要： Background: Intimate partner violence (IPV) damages health and is costly to families and society. Individuals experience different forms and combinations of IPV;better understanding of the respective health effects of these can help develop differentiated responses. This study explores the associations of different categories of IPV on women's mental and physical health. Methods: Using data from the World Health Organization (WHO) Multi-Country Study on Women's Health and Domestic Violence, multilevel mixed effects logistic regression modelling was used to analyse associations between categories of abuse (physical IPV alone, psychological IPV alone, sexual IPV alone, combined physical and psychological IPV, and combined sexual with psychological and/or physical IPV) with measures of physical and mental health, including self-reported symptoms, suicidal thoughts and attempts, and nights in hospital. Results: Countries varied in prevalence of different categories of IPV. All categories of IPV were associated with poorer health outcomes;the two combined abuse categories were the most damaging. The most common category was combined abuse involving sexual IPV, which was associated with the poorest health [attempted suicide: odds ratio (OR): 10.78, 95% confidence interval (CI) 8.37-13.89, thoughts of suicide: 8.47, 7.03-10.02, memory loss: 2.93, 2.41-3.56]. Combined psychological and physical IPV was associated with the next poorest outcomes (attempted suicide: 5.67, 4.23-7.60, thoughts of suicide: 4.41, 3.63-5.37, memory loss: 2.33, 1.88-2.87-). Conclusions: Understanding the prevalence and health impact of different forms and categories of IPV is crucial to risk assessment, tailoring responses to individuals and planning services. Previous analyses that focused on singular forms of IPV likely under-estimated the more harmful impacts of combined forms of abuse.

摘要： Purpose While infections are common following left ventricular assist device (LVAD) implant, the relationship between timing and category of first infection and mortality is less well understood. Methods STS Intermacs patients (>19 yrs) receiving a primary, durable continuous flow LVAD from 4/2012 to 5/2017 were included. The primary exposure was defined in 3 ways: any infection, timing of first post-implant infection (early: <90d, intermediate: 90-180d, late: >180d), and first infection category (VAD-specific, VAD-related, non-VAD). Cox proportional-hazards regression models were used to estimate the association between first infection and all-cause mortality, adjusting for comorbidities & post-implant stroke. Results A total of 12,957 patients (228,742 person-months of follow-up) were implanted (DT: 47.4%, BTT: 41.2%) at 166 centers. First infections were most often non-VAD (54.2%), followed by VAD-specific (25.4%) & VAD-related (20.4%). Rates of first infection (per 100 person-months) were highest in the early interval (10.7) vs intermediate (3.7) or late (2.7), p<0.001. Relative to patients without infection, patients with any infection had a significantly higher hazard of death (HR 3.46). First infection in the intermediate interval following implant was associated with the largest increase in hazard of death (HR 3.97), followed by late (HR 3.65) and early intervals (HR: 3.31). VAD-related infections were associated with the largest increase in hazard of death (HR 4.04) followed by non-VAD (HR 3.34), and VAD-specific infections (HR 3.23). Conclusion First infections were most often non-VAD-related and rates were highest early after implant. VAD-related infections and infections occurring in the intermediate interval were associated with the largest increase in hazard of death. A mixed-methods evaluation could explore these findings at the institutional level and why these patterns occur.