Factors driving CT utilisation in tertiary hospitals: a decomposition analysis using linked administrative data in Western Australia

ObjectivesWhile CT scanning plays a significant role in healthcare, its increasing use has raised concerns about inappropriate use. This study investigated factors driving the changing use of CT among people admitted to tertiary hospitals in Western Australia (WA).Design and settingA repeated cross-sectional study of CT use in WA in 2003–2005 and 2013–2015 using linked administrative heath data at the individual patient level.ParticipantsA total of 2 375 787 tertiary hospital admissions of people aged 18 years or older.Main outcome measureRate of CT scanning per 1000 hospital admissions.MethodsA multivariable decomposition model was used to quantify the contribution of changes in patient characteristics and changes in the probability of having a CT over the study period.ResultsThe rate of CT scanning increased by 112 CT scans per 1000 admissions over the study period. Changes in the distribution of the observed patient characteristics were accounted for 62.7% of the growth in CT use. However, among unplanned admissions, changes in the distribution of patient characteristics only explained 17% of the growth in CT use, the remainder being explained by changes in the probability of having a CT scan. While the relative probability of having a CT scan generally increased over time across most observed characteristics, it reduced in young adults (−2.8%), people living in the rural/remote areas (−0.8%) and people transferred from secondary hospitals (−0.8%).ConclusionsOur study highlights potential improvements in practice towards reducing medical radiation exposure in certain high risk population. Since changes in the relative probability of having a CT scan (representing changes in scope) rather than changes in the distribution of the patient characteristics (representing changes in need) explained a major proportion of the growth in CT use, this warrants more in-depth investigations in clinical practices to better inform health policies promoting appropriate use of diagnostic imaging tests.

931Work productivity trajectories in a representative sample of Australians living with multiple sclerosis- group-based modelling

Background Studies have documented loss of work capacity and work productivity loss in multiple sclerosis (MS). Little is known about the longitudinal trajectories of work productivity in MS. Objectives To explore trajectories of work productivity in people living with multiple sclerosis (PwMS) and examine the baseline factors linked with assignment to the trajectories group. Methods Study participants were from the Australian MS Longitudinal Study (AMSLS) from 2015 to 2018 who were employed with ≥2 datapoints (n = 1205). We used group-based trajectory modelling to identify unique work productivity trajectories in PwMS. Multinomial logistic regression was used to assess associations with the work productivity trajectories. Results We identified three distinct trajectories of work productivity: ‘moderately worsened’ (16.7%) with a mean work productivity of 47.6% in 2015, ‘mildly worsened’ (50.1%) with a mean work productivity of 86.3% in 2015 and ‘normal’ (33.2%) with a mean work productivity of 99.7% in 2015. On multivariable analysis, the relative probability of being in moderately or mildly worsened work productivity trajectory were higher for baseline factors such as higher education level, longer disease duration, higher disability score, and high MS symptom severity. For example, the relative probability of being in ‘moderately worsened’ rather than ‘normal’ work productivity trajectory increased by 94% (RRR:1.94 ; 95% confidence interval:1.68 - 2.25) for each unit increase in ‘fatigue and cognitive symptoms’ cluster. Conclusion Education level, disability and MS symptom severity increased the probability of following low work productivity trajectory. Key messages Work productivity interventions should target MS symptom severity and disability reduction.

Northern bobwhite select for shrubby thickets interspersed in grasslands during fall and winter

Resource selection is a key component in understanding the ecological processes underlying population dynamics, particularly for species such as northern bobwhite (Colinus virginianus), which are declining across their range in North America. There is a growing body of literature quantifying breeding season resource selection in bobwhite; however, winter information is particularly sparse despite it being a season of substantial mortality. Information regarding winter resource selection is necessary to quantify the extent to which resource requirements are driving population change. We modeled bobwhite fall and winter resource selection as a function of vegetation structure, composition, and management from traditionally (intensively) managed sites and remnant (extensively managed) grassland sites in southwest Missouri using multinomial logit discrete choice models in a Bayesian framework. We captured 158 bobwhite from 67 unique coveys and attached transmitters to 119 individuals. We created 671 choice sets comprised of 1 used location and 3 available locations. Bobwhite selected for locations which were closer to trees during the winter; the relative probability of selection decreased from 0.45 (85% Credible Interval [CRI]: 0.17–0.74) to 0.00 (85% CRI: 0.00–0.002) as distance to trees ranged from 0–313 m. The relative probability of selection increased from near 0 (85% CRI: 0.00–0.01) to 0.33 (85% CRI: 0.09–0.56) and from near 0 (85% CRI: 0.00–0.00) to 0.51 (85% CRI: 0.36–0.71) as visual obstruction increased from 0 to 100% during fall and winter, respectively. Bobwhite also selected locations with more woody stems; the relative probability of selection increased from near 0.00 (85% CRI: 0.00–0.002) to 0.30 (85% CRI: 0.17–0.46) and near 0.00 (85% CRI: 0.00–0.001) to 0.35 (85% CRI: 0.22–0.55) as stem count ranged from 0 to 1000 stems in fall and winter, respectively. The relative probability of selection also decreased from 0.35 (85% CRI: 0.20–0.54) to nearly 0 (85% CRI: 0.00–0.001) as percent grass varied from 0 to 100% in fall. We suggest that dense shrub cover in close proximity to native grasslands is an important component of fall and winter cover given bobwhite selection of shrub cover and previously reported survival benefits in fall and winter.

Risk factors for adverse outcomes during mechanical ventilation of 1152 COVID-19 patients: a multicenter machine learning study with highly granular data from the Dutch Data Warehouse

Background The identification of risk factors for adverse outcomes and prolonged intensive care unit (ICU) stay in COVID-19 patients is essential for prognostication, determining treatment intensity, and resource allocation. Previous studies have determined risk factors on admission only, and included a limited number of predictors. Therefore, using data from the highly granular and multicenter Dutch Data Warehouse, we developed machine learning models to identify risk factors for ICU mortality, ventilator-free days and ICU-free days during the course of invasive mechanical ventilation (IMV) in COVID-19 patients. Methods The DDW is a growing electronic health record database of critically ill COVID-19 patients in the Netherlands. All adult ICU patients on IMV were eligible for inclusion. Transfers, patients admitted for less than 24 h, and patients still admitted at time of data extraction were excluded. Predictors were selected based on the literature, and included medication dosage and fluid balance. Multiple algorithms were trained and validated on up to three sets of observations per patient on day 1, 7, and 14 using fivefold nested cross-validation, keeping observations from an individual patient in the same split. Results A total of 1152 patients were included in the model. XGBoost models performed best for all outcomes and were used to calculate predictor importance. Using Shapley additive explanations (SHAP), age was the most important demographic risk factor for the outcomes upon start of IMV and throughout its course. The relative probability of death across age values is visualized in Partial Dependence Plots (PDPs), with an increase starting at 54 years. Besides age, acidaemia, low P/F-ratios and high driving pressures demonstrated a higher probability of death. The PDP for driving pressure showed a relative probability increase starting at 12 cmH2O. Conclusion Age is the most important demographic risk factor of ICU mortality, ICU-free days and ventilator-free days throughout the course of invasive mechanical ventilation in critically ill COVID-19 patients. pH, P/F ratio, and driving pressure should be monitored closely over the course of mechanical ventilation as risk factors predictive of these outcomes.

Easy Computation of the Bayes Factor to Fully Quantify Occam’s Razor: Least-squares Fitting and Policy Decisions

The Bayes factor is the gold-standard figure of merit for comparing fits of models to data, for hypothesis selection and parameter estimation. However, it is little-used because it has been considered to be subjective, and to be computationally very intensive. A simple computational method has been known for at least 30 years, but has been dismissed as an approximation. We show here that all three criticisms are misplaced. The method should be used with all least-squares fitting, because it can give very different, better outcomes than classical methods. It can discriminate between models with equal numbers of parameters and equally good fits to data. It quantifies the Occam’s Razor injunction against over-fitting, and it demands that physically-meaningful parameters rejected by classical significance testing be included in the fitting, to avoid spurious precision and incorrect values for the other parameters. It strongly discourages the use of physically-meaningless parameters, thereby satisfying the Occam’s Razor injunction to use existing entities for explanation rather than multiplying new ones. More generally, being a relative probability, the Bayes factor combines naturally with other quantitative information to guide action in the absence of certain knowledge.

Deformation Parameters and Collective Temperature Changes in Photofission Mass Yields of Actinides Within the Systematic Statistical Scission Point Model

The photofission fragment mass yields of actinides are evaluated using a systematic statistical scission point model. In this model, all energies at the scission point are presented as a linear function of the mass numbers of fission fragments. The mass yields are calculated with a new approximated relative probability for each complementary fragment. The agreement with the experimental data is quite good, especially with a collective temperature Tcol of 2 MeV at intermediate excitation energy and Tcol = 1 MeV for spontaneous fission. This indicates that the collective temperature is greater than the value obtained by the initial excitation energy. The generalized superfluid model is applied for calculating the fragment temperature. The deformation parameters of fission fragments have been obtained by fitting the calculated results with the experimental values. This indicates that the deformation parameters decrease with increasing excitation energy. Also, these parameters decrease for fissioning systems with odd mass numbers.

Differences in the Associations Between Sociodemographic and Lifestyle Factors with Dietary Patterns in Mexican Men and Women.

Background. Diet is one of the leading risk factors for developing non-communicable diseases and is related to sociodemographic and lifestyle factors, including sex. We aimed to investigate the associated factors of dietary patterns among adults living in Mexico City by sex. Methods. We used data from a city-wide representative survey conducted between May and June 2015 in Mexico City (n=1,142). Self-reported information about sociodemographic and lifestyle variables was collected. Dietary information was collected using a semi-quantitative food frequency questionnaire. Dietary patterns were constructed by cluster analysis. We used sex-specific multivariable multinomial logistic models to assess the association of demographic and lifestyle factors with dietary patterns using. Results. Three dietary patterns were identified: basic, prudent and fast food. Among men and women, higher school attainment was associated with a lower relative probability of having a basic rather than prudent dietary pattern (women: RRR= 0.8, 95% CI: 0.8, 0.9; men: RRR= 0.8, 95% CI:0.7, 0.9). Compared to single men, divorced or separated men (RRR=3.8, 95% CI: 1.3, 11.2) and those living with a partner (RRR=2.6, 95% CI: 1.1, 6.0) had a higher relative probability of consuming a fast food dietary pattern than the prudent one. Men living with a partner (RRR=3.0, 95% CI:1.1, 8.6) or working long shifts (RRR=3.8, 95% CI: 1.3, 11.1) had a higher probability of consuming a basic pattern rather than a prudent one compared to peers. Conclusion. Differences by sex in the associations between sociodemographic factors and dietary patterns may be due to gender roles. Public policies and programs should consider the gender perspective to accomplish positive results in both men and women.

Medical Fraud and Abuse Detection System Based on Machine Learning

It is estimated that approximately 10% of healthcare system expenditures are wasted due to medical fraud and abuse. In the medical area, the combination of thousands of drugs and diseases make the supervision of health care more difficult. To quantify the disease–drug relationship into relationship score and do anomaly detection based on this relationship score and other features, we proposed a neural network with fully connected layers and sparse convolution. We introduced a focal-loss function to adapt to the data imbalance and a relative probability score to measure the model’s performance. As our model performs much better than previous ones, it can well alleviate analysts’ work.

Merging different resolution rainfall products to support landslide prediction over Nepal

<p>Landslides within Nepal result both from human interventions, intensive rainfall and tectonic activity. This work presents the steps taken towards the development of a Territorial landslide early warning system (Te-LEWSs) for predicting the relative probability of the occurrence of precipitation driven landslides in the west of Nepal. Since precipitation triggers may be dominated by intense short periods of rainfall focus is given to testing the use of relationships between high resolution local observed precipitation, satellite data and Numerical Weather Models output in the development of the forecasting model. Our results show the relative importance of these alongside the significance of human activity when the model is compared against observed data sets.</p>