Search

Evidence-based guidelines suggest computed tomography without magnetic resonance imaging (MRI) is sufficient to rule out clinically significant cervical spine injury in obtunded adult blunt trauma patients. This study evaluated MRI utilization over time to investigate the impact of the 2015 Eastern Association for the Surgery of Trauma guidelines suggesting cervical collar clearance with computed tomography alone in this population. We hypothesized that MRI utilization would decrease following the guidelines.We performed a retrospective cross-sectional study of the National Trauma Data Bank from 2007 to 2018 using multivariable logistic regression of the likelihood of spinal MRI utilization. Blunt trauma patients 18 y and older with a Glasgow Coma Scale (GCS) of 8 or less, Abbreviated Injury Score head of 4 or greater, intubated for at least 72 h were included.The sample consisted of 76,450 patients from 567 trauma centers. Controlling for age, gender, race/ethnicity, insurance status, injury mechanism, Injury Severity Score, GCS, GCS motor, hospital teaching status and trauma center level, patients seen after 2015 had a higher odds ratio (OR) of undergoing spinal MRI relative to those seen before 2015 (OR 1.77, 95% CI 1.49-2.09; P0.001). Each year was associated with a significantly increased OR of undergoing spinal MRI compared to the year prior (OR 1.10, 95% CI 1.05-1.15; P0.001).Spinal MRI use has been increasing in obtunded adult blunt trauma patients including after the release of the Eastern Association for the Surgery of Trauma guidelines in 2015. Future work should identify whether this is driven by improper MRI utilization and, if so, strategies to promote guideline adherence.

Recently, convolutional neural networks (CNNs) have exhibited commendable performance for hyperspectral image (HSI) classification. Generally, an important number of samples are needed for each class to properly train CNNs. However, existing HSI data sets suffer from a significant class imbalance problem, where many classes do not have enough samples to characterize the spectral information. The performance of existing CNN models is biased toward the majority classes, which possess more samples for the training. This article addresses this issue of imbalanced data in HSI classification. In particular, a new 3D-HyperGAMO model is proposed, which uses generative adversarial minority oversampling. The proposed 3D-HyperGAMO automatically generates more samples for minority classes at training time, using the existing samples of that class. The samples are generated in the form of a 3-D hyperspectral patch. A different classifier from the generator and the discriminator is used in the 3D-HyperGAMO model, which is trained using both original and generated samples to determine the classes of newly generated samples to which they actually belong. The generated data are combined classwise with the original training data set to learn the network parameters of the class. Finally, the trained 3-D classifier network validates the performance of the model using the test set. Four benchmark HSI data sets, namely, Indian Pines (IP), Kennedy Space Center (KSC), University of Pavia (UP), and Botswana (BW), have been considered in our experiments. The proposed model shows outstanding data generation ability during the training, which significantly improves the classification performance over the considered data sets. The source code is available publicly at https://github.com/mhaut/3D-HyperGAMO.

Endmember estimation consists of two tasks, that is, determining the number of pure spectral constituents (endmembers) and extracting their spectral signatures. We present a new geometric distance-based method for endmember estimation from hyperspectral images (HSIs), which does not need to know the number of endmembers in advance. Our strategy optimizes the widely used maximum distance analysis (MDA) method from two viewpoints. First, the traditional MDA method performs endmember estimation by computing the maximum distances between any pixel and one specific pixel, line, plane, or affine hull (AH) composed by the endmembers that have been formerly extracted. Instead, our new strategy only requires computing the maximum distance between any pixel and one specific AH. This operation provides a simpler way than MDA to estimate endmembers. Second, our strategy exploits a new distance computation between any pixel and an AH and just needs the normal vector (compared to the traditional MDA method, which uses the normal vector and offset). The new distance computation in our method is much more efficient than that in the traditional MDA method.

Hyperspectral imaging (HSI) is a competitive remote sensing technique in several fields, from Earth observation to health, robotic vision, and quality control. Each HSI scene contains hundreds of (narrow) contiguous spectral bands. The amount of data generated by HSI devices is often both a solution and a problem for a given application. Extracting information from HSI data cubes is a complex and computationally demanding problem. To tackle this challenge, convolutional neural networks (CNNs) have been widely applied to HSI classification. Despite their success, CNNs are computationally demanding algorithms with high memory requirements due to their large number of internal parameters. The recent interest in using HSI devices in mobile and embedded systems for air and spaceborne platforms turned the attention to computationally lightweight CNN architectures with good classification accuracy. In this article, we present a contribution in that direction. The proposed method combines the ghost-module architecture with a CNN-based HSI classifier to reduce the computational cost and, simultaneously, achieves an efficient classification method with high performance. Our new method is evaluated against nine standard HSI classifiers, and five improved deep-CNN architectures, over five commonly used HSI data sets for algorithm benchmarking. Conducted experiments show that the proposed method exhibits similar or better performance than the other classifiers, achieving top values in the considered performance metrics--even for very limited training sets--and, most importantly, with a fraction of the computational cost. Our novel approach for HSI classification is a strong candidate for implementation on systems with limited computational resources.

BACKGROUND: There are substantial racial and socioeconomic disparities underlying endovascular abdominal aortic aneurysm repair (EVAR) in the United States. To date, race-based variations in reinterventions following elective EVAR have not been studied. Here, we aim to examine racial disparities associated with reinterventions following elective EVAR in a real-world cohort. MATERIALS AND METHODS: We used the Vascular Quality Initiative (VQI) EVAR dataset to identify all patients undergoing elective EVAR between January 2009 and December 2018 in the United States. We compared the association of race with reinterventions after EVAR and all-cause mortality using Welch two-sample t-tests, multivariate logistic regression, and Cox proportional hazards analyses adjusting for baseline differences between groups. RESULTS: At median follow-up of 1.1±1.1 years (1.3±1.4 years Black, 1.1±1.1 years White; P=0.02), a total of 1,164 of 42,481 patients (2.7%) underwent reintervention after elective EVAR, including 2.7% (n = 1,096) White vs. 3.2% (n = 68) Black (P = 0.21). Black patients requiring reintervention were more frequently female, more frequently current or former smokers, and less frequently insured by Medicare/Medicaid (P

PURPOSE Hereditary cancer syndromes infer high cancer risks and require intensive surveillance. Identification of high-risk individuals among patients with colorectal cancer (CRC) needs improvement. METHODS Three thousand three hundred ten unselected adults who underwent surgical resection for primary invasive CRC were prospectively accrued from 51 hospitals across Ohio between January 1, 2013, and December 31, 2016. Universal Tumor screening (UTS) for mismatch repair (MMR) deficiency was performed for all, and pathogenic germline variants (PGVs) were identified using multigene panel testing (MGPT) in those who met at least one inclusion criterion: MMR deficiency, diagnosed < 50 years, multiple primary tumors (CRC or endometrial cancer), or with a first-degree relative with CRC or endometrial cancer. RESULTS Five hundred twenty-five patients (15.9%) had MMR deficiency. Two hundred thirty-four of 3,310 (7.1%; 16% of the 1,462 who received MGPT) had 248 PGVs in cancer susceptibility genes. One hundred forty-two (4.3%) had a PGV in an MMR gene, and 101 (3.1%) had a PGV in a non-MMR gene. Ten with Lynch syndrome (LS) also had a non-MMR PGV and were included in both groups. Two (0.06%) had constitutional MLH1 hypermethylation. Of unexplained MMR-deficient patients, 88.4% (76 of 86) had double somatic MMR mutations. Testing for only MMR genes in MMR-deficient patients would have missed 18 non-MMR gene PGVs (7.3% of total PGVs identified). Had UTS been the only method used to screen for hereditary cancer syndromes, 38.6% (91 of 236) would have been missed, including 6.3% (9 of 144) of those with LS. These results have treatment implications as 5.3% (175 of 3,310) had PGVs in genes with therapeutic targets. CONCLUSION UTS alone is insufficient for identifying a large proportion of CRC patients with hereditary syndromes, including some with LS. At a minimum, 7.1% of individuals with CRC have a PGV and pan-cancer MGPT should be considered for all patients with CRC.

Background: This study evaluated the association between eligible patients not proceeding with resective epilepsy surgery and various demographic, disease-specific, and epilepsy-evaluation variables. Methods: This retrospective case-control study included patients identified as candidates for resective epilepsy surgery at the Montefiore Medical Center between January 1, 2009 and June 30, 2017. Chi-squared, two-tailed, independent sample t-test, Mann-Whitney U test and logistic regression were utilized to identify variables associated with patients not proceeding with surgery. Results: Among the 159 potential surgical candidates reviewed over the 8.5-year study period, only 53 ultimately proceeded with surgery (33%). Eighty-seven (55%) out of these 159 patients were identified as appropriate for resective epilepsy surgery during the study period. Thirty-four (39%) of these 87 patients did not proceed with surgery. Variables independently correlated (either positively or negatively) with the patient not proceeding with surgery were: being employed [Odds Ratio (OR) 4.2, 95% confidence interval (CI) 1.12-15.73], temporal lobe lesion on MRI (OR 0.35, 95% CI 0.14-0.84), temporal lobe EEG ictal onsets (OR 0.21, 95% CI 0.07-0.62), and temporal lobe epileptogenic zone (OR 0.19, 95% CI 0.07-0.55). Conclusion: The novel finding in this study is the association between employment status and whether the patient had epilepsy surgery: employed patients were 4.2 times more likely to not proceed with surgery compared to unemployed patients. In addition, patients with a temporal lobe lesion on MRI, temporal lobe EEG ictal onsets, and/or a temporal epileptogenic zone were more likely to proceed with surgery. Future work will be needed to evaluate these findings prospectively, determine if they generalize to other patient populations, explore the decision whether or not to proceed with epilepsy surgery from a patient-centered perspective, and suggest strategies to reduce barriers to this underutilized treatment.

The opioid epidemic continues to have devastating consequences for children and families across the United States with rising prevalence of opioid use and abuse. Given the ease of access to these medications, accidental ingestion and overdose by children are becoming increasingly more common. The recognition of opioid-induced neurotoxicity and the associated life-threatening complication of acute cerebellar cytotoxic edema are crucial, as are the high morbidity and mortality without timely intervention. We discuss an infant with acute cytotoxic cerebellar edema following mucosal exposure to a transdermal fentanyl patch.

Background Previous US-based studies have shown that a trauma center designation of level 1 is associated with improved patient outcomes. However, most studies are cross-sectional, focus on volume-related issues and are direct comparisons between levels. This study investigates the change in patient characteristics when individual trauma centers transition from level 2 to level 1 and whether the patients have similar outcomes during the initial period of the transition. Study Design We performed a retrospective cohort study that analyzed hospital and patient records included in the National Trauma Data Bank from 2007 to 2016. Patient characteristics were compared before and after their hospitals transitioned their trauma level. Mortality; complications including acute kidney injury, acute respiratory distress syndrome, cardiac arrest with CPR, deep surgical site infection, deep vein thrombosis, extremity compartment syndrome, surgical site infection, osteomyelitis, pulmonary embolism, and so on; ICU admission; ventilation use; unplanned returns to the OR; unplanned ICU transfers; unplanned intubations; and lengths of stay were obtained following propensity score matching, comparing posttransition years with the last pretransition year. Results Sixteen trauma centers transitioned from level 2 to level 1 between 2007 and 2016. One was excluded due to missing data. After transition, patient characteristics showed differences in the distribution of race, comorbidities, insurance status, injury severity scores, injury mechanisms, and injury type. After propensity score matching, patients treated in a trauma center after transition from level 2 to 1 required significantly fewer ICU admissions and had lower complication rates. However, significantly more unplanned intubations, unplanned returns to the OR, unplanned ICU transfers, ventilation use, surgical site infections, pneumonia, and urinary tract infections and higher mortality were reported after the transition. Conclusions Trauma centers that transitioned from level 2 to level 1 had lower overall complications, with fewer patients requiring ICU admission. However, higher mortality and more surgical site infections, pneumonia, urinary tract infections, unplanned intubations, and unplanned ICU transfers were reported after the transition. These findings may have significant implications in the planning of trauma systems for administrators and healthcare leaders.

As a newly emerging technology, deep learning (DL) is a very promising field in big data applications. Remote sensing often involves huge data volumes obtained daily by numerous in-orbit satellites. This makes it a perfect target area for data-driven applications. Nowadays, technological advances in terms of software and hardware have a noticeable impact on Earth observation applications, more specifically in remote sensing techniques and procedures, allowing for the acquisition of data sets with greater quality at higher acquisition ratios. This results in the collection of huge amounts of remotely sensed data, characterized by their large spatial resolution (in terms of the number of pixels per scene), and very high spectral dimensionality, with hundreds or even thousands of spectral bands. As a result, remote sensing instruments on spaceborne and airborne platforms are now generating data cubes with extremely high dimensionality, imposing several restrictions in terms of both processing runtimes and storage capacity. In this article, we provide a comprehensive review of the state of the art in DL for remote sensing data interpretation, analyzing the strengths and weaknesses of the most widely used techniques in the literature, as well as an exhaustive description of their parallel and distributed implementations (with a particular focus on those conducted using cloud computing systems). We also provide quantitative results, offering an assessment of a DL technique in a specific case study (source code available: https://github.com/mhaut/cloud-dnn-HSI ). This article concludes with some remarks and hints about future challenges in the application of DL techniques to distributed remote sensing data interpretation problems. We emphasize the role of the cloud in providing a powerful architecture that is now able to manage vast amounts of remotely sensed data due to its implementation simplicity, low cost, and high efficiency compared to other parallel and distributed architectures, such as grid computing or dedicated clusters.

HNO : Hals-Nasen-Ohren-Heilkunde, Kopf- und Halschirurgie (2021). doi:10.1007/s00106-021-01070-9, Published by Springer, New York

BACKGROUND AND OBJECTIVES Health services research often relies on readily available data, originally collected for administrative purposes and used for public reporting and pay-for-performance initiatives. We examined the prevalence of underreporting of diagnostic procedures for acute myocardial infarction (AMI), deep venous thrombosis (DVT), and pulmonary embolism (PE), used for public reporting and pay-for-performance initiatives. METHOD We retrospectively identified procedures for AMI, DVT, and PE in the National Inpatient Sample (NIS) database between 2012 and 2016. From January 1, 2012, through September 30, 2015, the NIS used the International Classification of Diseases, Ninth Revision (ICD-9) coding scheme. From October 1, 2015, through December 31, 2016, the NIS used the International Classification of Diseases, Tenth Revision (ICD-10) coding scheme. We grouped the data by ICD code definitions (ICD-9 or ICD-10) to reflect these code changes and to prevent any confounding or misclassification. In addition, we used survey weighting to examine the utilization of venous duplex ultrasound scan for DVT, electrocardiogram (ECG) for AMI, and chest computed tomography (CT) scan, pulmonary angiography, echocardiography, and nuclear medicine ventilation/perfusion () scan for PE. RESULTS In the ICD-9 period, by primary diagnosis, only 0.26% (n = 5930) of patients with reported AMI had an ECG. Just 2.13% (n = 7455) of patients with reported DVT had a peripheral vascular ultrasound scan. For patients with PE diagnosis, 1.92% (n = 12 885) had pulmonary angiography, 3.92% (n = 26 325) had CT scan, 5.31% (n = 35 645) had cardiac ultrasound scan, and 0.45% (n = 3025) had scan. In the ICD-10 period, by primary diagnosis, 0.04% (n = 345) of reported AMI events had an ECG and 0.91% (n = 920) of DVT events had a peripheral vascular ultrasound scan. For patients with PE diagnosis, 2.08% (n = 4805) had pulmonary angiography, 0.63% (n = 1460) had CT scan, 1.68% (n = 3890) had cardiac ultrasound scan, and 0.06% (n = 140) had scan. Small proportions of diagnostic procedures were observed for any diagnoses of AMI, DVT, or PE. CONCLUSIONS Our findings question the validity of using NIS and other administrative databases for health services and outcomes research that rely on certain diagnostic procedures. Unfortunately, the NIS does not provide granular data that can control for differences in diagnostic procedure use, which can lead to surveillance bias. Researchers and policy makers must understand and acknowledge the limitations inherent in these databases, when used for pay-for-performance initiatives and hospital benchmarking.

Ultrasound guidance is an effective technique for obtaining short peripheral catheter (SPC) access but requires training and practice for proficiency. The aim of this quality improvement initiative was to develop and assess a formal training program to increase the confidence and competency of intravenous (IV) therapy nurses in the placement of ultrasound-guided SPCs. IV therapy nurses completed a didactic and hands-on training course where they practiced ultrasound-guided SPC placement techniques on a poultry phantom during simulation, followed by performing ultrasound-guided SPC insertion on patients proctored by an interventional radiology physician. Data collection included preintervention and postintervention confidence self-assessment, frequency tracker, Difficult Intravenous Access (DIVA) scale scores, and total number of ultrasound-guided SPCs placed by the nurses. Ultrasound-guided SPC placement increased significantly after the training program. The IV therapy nurses placed 29 ultrasound-guided SPCs in 2017, 391 ultrasound-guided SPCs in 2018, and 711 ultrasound-guided SPCs in 2019. Mean DIVA scores rose from 4.54 in May 2018 to 5.17 in July 2018, indicating success in placing SPCs in more difficult patients. Implementation of an ultrasound-guided SPC placement program using poultry phantom simulation is a recommended nursing resource for increasing competency in ultrasound-guided SPC placement in pediatric patients.

An editorial is presented on the mitigating the July effect. Topics include the arrival of a crop of newly graduated medical students beginning their internships, the influx of so many freshly trained physicians arriving at the same time always triggers concern, and the belief in a July effect with presumed adverse consequences for patient safety and quality of care.

Convolutional neural networks (CNNs) have demonstrated excellent performance in hyperspectral image (HSI) classification. However, tuning some critical hyperparameters of a CNN—such as the receptive field (RF) size—presents a major challenge due to the presence of features with different scales in HSIs. Contrary to the conventional design of CNNs, which fixes the RF size, it has been proven that the RF size is modulated by the stimulus and hence, depends on the scene being considered. Such a dilemma has been rarely considered in CNN design. In this letter, a new multibranch selective kernel network (MSKNet) is introduced, in which the input image is convolved using different RF sizes to create multiple branches so that the effect of each branch is adjusted by an attention mechanism according to the input contrast. As a result, our newly developed MSKNet is capable of modeling different scales. Our experimental results, conducted on three widely used HSIs, reveal that the MSKNet can outperform state-of-the-art CNNs in the context of HSI classification problems. The source code of our newly developed MSKNet is available from: https://github.com/mhaut/MSKNet-HSI

The role of pediatric hospitals in the COVID-19 pandemic changed quickly. The team of clinical nurse specialists and clinical nurse educators in a large pediatric hospital were instrumental in the institutional response through simulations, serving as change agents, collaboration, and implementing systems thinking. Leveraging the expertise of this team during this historical and unprecedented time optimized patient and associate safety as part of a pediatric hospital's COVID-19 response.

Given high relapse rates and the prevalence of overdose deaths, novel treatments for substance use disorder (SUD) are desperately needed for those who are treatment refractory. The objective of this study was to evaluate the safety of deep brain stimulation (DBS) for SUD and the effects of DBS on substance use, substance craving, emotional symptoms, and frontal/executive functions. DBS electrodes were implanted bilaterally within the Nucleus Accumbens/Ventral anterior internal capsule (NAc/VC) of a man in his early 30s with >10-year history of severe treatment refractory opioid and benzodiazepine use disorders. DBS of the NAc/VC was found to be safe with no serious adverse events noted and the participant remained abstinent and engaged in comprehensive treatment at the 12-week endpoint (and 12-month extended follow-up). Using a 0-100 visual analog scale, substance cravings decreased post-DBS implantation; most substantially in benzodiazepine craving following the final DBS titration (1.0 ± 2.2) compared to baseline (53.4 ± 29.5; p < .001). A trend toward improvement in frontal/executive function was observed on the balloon analog risk task performance following the final titration (217.7 ± 76.2) compared to baseline (131.3 ± 28.1, p = .066). FDG PET demonstrated an increase in glucose metabolism in the dorsolateral prefrontal and medial premotor cortices at the 12-week endpoint compared to post-surgery/pre-DBS titration. Heart Rate Variability (HRV) improved following the final titration (rMSSD = 56.0 ± 11.7) compared to baseline (19.2 ± 8.2; p < .001). In a participant with severe, treatment refractory opioid and benzodiazepine use disorder, DBS of the NAc/VC was safe, reduced substance use and craving, and improved frontal and executive functions. Confirmation of these findings with future studies is needed. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Previous studies have shown that reconstructive surgery alone following injury to the anterior cruciate ligament (ACL) does not prevent the development of post-traumatic osteoarthritis (PTOA). Poloxamer 188 (P188) has been shown to prevent cell death following trauma in both articular cartilage and meniscal tissue. This study aims to test the efficacy of single or multiple administrations of P188 in conjunction with reconstructive surgery to help prevent or delay the onset of the disease. Thirty skeletally mature rabbits underwent closed-joint trauma that resulted in ACL rupture and meniscal damage and were randomly assigned to one of four treatment groups with varying doses of P188. ACL reconstruction was then performed using an autograft from the semitendinosus tendon. Animals were euthanized 1-month following trauma, meniscal tissue was assessed for changes in morphology, mechanical properties, and proteoglycan content. Femurs and tibias were scanned using microcomputed tomography to determine changes in bone quality, architecture, and osteophyte formation. The medial meniscus experienced more damage and a decrease in the instantaneous modulus regardless of treatment group, while P188 treatment tended to limit degenerative changes in the lateral meniscus. Both lateral and medial menisci had documented decreases in the equilibrium modulus and inconsistent changes in proteoglycan content. Minimal changes were documented in the tibias and femurs, with the only significant change being the formation of osteophytes in both bones regardless of treatment group. The data suggest that P188 was able to limit some degenerative changes in the meniscus associated with PTOA and may warrant future studies.

Hyperspectral images (HSIs) record scenes at different wavelength channels, providing detailed spatial and spectral information. How to storage and process this high-dimensional data plays a vital role in many practical applications, where classification technologies have emerged as excellent processing tools. However, their high computational complexity and energy requirements bring some challenges. Adopting low-power consumption architectures and deep learning (DL) approaches has to provide acceptable computing capabilities without reducing accuracy demand. However, most DL architectures employ single-precision (FP32) to train models, and some big DL architectures will have a limitation on memory and computation resources. This can negatively affect the network learning process. This letter leads these challenges by using mixed precision into DL architectures for HSI classification to speed up the training process and reduce the memory consumption/access. Proposed models are evaluated on four widely used data sets. Also, low and high-power consumption devices are compared, considering NVIDIA Jetson Xavier and Titan RTX GPUs, to evaluate the proposal viability in on-board processing devices. Obtained results demonstrate the efficiency and effectiveness of these models within HSI classification task for both devices. Source codes: https://github.com/mhaut/CNN-MP-HSI .

Seizure clusters must be treated quickly and effectively to prevent progression to prolonged seizures and status epilepticus. Rescue therapy for seizure clusters has focused on the use of benzodiazepines. Although intravenous benzodiazepine administration is the primary route in hospitals and emergency departments, seizure clusters typically occur in out‐of‐hospital settings, where a more portable product that can be easily administered by nonmedical caregivers is needed. Thus, other methods of administration have been examined, including rectal, intranasal, intramuscular, and buccal routes. Following US Food and Drug Administration (FDA) approval in 1997, rectal diazepam became the mainstay of out‐of‐hospital treatment for seizure clusters in the United States. However, social acceptability and consistent bioavailability present limitations. Intranasal formulations have potential advantages for rescue therapies, including ease of administration and faster onset of action. A midazolam nasal spray was approved by the FDA in 2019 for patients aged 12 years or older. In early 2020, the FDA approved a diazepam nasal spray for patients aged 6 years or older, which has a different formulation than the midazolam nasal product and enhances aspects of bioavailability. Benzodiazepines, including diazepam, present significant challenges in developing a suitable intranasal formulation. Diazepam nasal spray contains dodecyl maltoside (DDM) as an absorption enhancer and vitamin E to increase solubility in an easy‐to‐use portable device. In a Phase 1 study, absolute bioavailability of the diazepam nasal spray was 97% compared with intravenous diazepam. Subsequently, the nasal spray demonstrated less variability in bioavailability than rectal gel (percentage of geometric coefficient of variation of area under the curve = 42%–66% for diazepam nasal spray compared with 87%–172% for rectal gel). The diazepam nasal spray safety profile is consistent with that expected for rectal diazepam, with low rates of nasal discomfort (≤6%). To further improve the efficacy of rescue therapy, investigation of novel intranasal benzodiazepine formulations is underway.

Nowadays, deep convolutional neural networks (CNNs) for face recognition exhibit a performance comparable to human ability in the presence of the appropriate amount of labelled training data. However, training CNNs remains as an arduous task due to the lack of training samples. To overcome this drawback, applications demand one-shot learning to improve the obtained performances over traditional machine learning approaches by learning representative information about data categories from few training samples. In this context, Siamese convolutional network ( SiConvNet ) provides an interesting deep architecture to tackle the data limitation. In this regard, applying the convolution operation on real world images by using the trainable correlative Gaussian kernel adds correlations to the output images, which hinder the recognition process due to the blurring effects introduced by the convolution kernel application. As a result the pixel-wise and channel-wise correlations or redundancies could appear in both single and multiple feature maps obtained by a hidden layer. In this sense, convolution-based models fail to generalize the feature representation because of both the strong correlations presence in neighboring pixels and the channel-wise high redundancies between different channels of the feature maps, which hamper the effective training. Deconvolution operation helps to overcome the shortcomings that limit the conventional SiConvNet performance, learning successfully correlation-free features representation. In this paper, a simple but efficient Siamese convolution deconvolution feature fusion network ( SiCoDeF 2 Net ) is proposed to learn the invariant and discriminative complementary features generated from both the (i) sub-convolution (SCoNet) and (ii) sub deconvolutional (SDeNet) networks using a concatenation operation which significantly improves the one-shot unconstrained facial recognition task. Extensive experiments performed on several widely used benchmarks, provide promising results, where the proposed SiCoDeF 2 Net model significantly outperforms the current state-of-art in terms of classification accuracy, F1, precision and recall. The code will be available on: https://github.com/purbayankar/SiCoDeF2Net .

The instructional design of meaningful experimental activities completely integrated into on- line Engineering courses is still a significant challenge, by following the sustainable development goal of the United Nations for quality education, the SDG number 4. Remote and virtual laboratories can be a solution to improve this design in distance methodology. This fact has been more noticeable during the COVID-19 pandemic and its consequences with respect to education. Its transparent integration into the learning process and the inclusion of learning analytic features are essential features of these kinds of technologies. Without them, lecturers cannot track the performance and the satisfaction of their students. Our LoT@UNED (Laboratory-of-Things@UNED) platform has been implemented to include all these educational features for IoT environments, ranging from edge to cloud computing. Taking the advantages of the IoT paradigm, this platform involves acquiring practical competences within IoT and edge-cloud programming topics, among others. Furthermore, its flexibility allows us to host other areas, such as cybersecurity. Specifically, our platform incorporates a set of sustainable capabilities (scalability, availability, and security), so contributing to the SDG 4 challenge with a low-complex and efficient solution for quality distance education. To analyze the impact of our proposed solution, this work also focuses on studying the influence among several acceptance UTAUT/TAM factors during the students’ learning and teaching process. To achieve this goal, both exploratory and confirmatory analyses have been conducted. The reliability and validity of data are also analyzed. An improved structural equation model has also been studied, satisfying several statistical indicators according to the recommended thresholds in the current literature. The usefulness of our platform, easing the users’ attitude and having a high availability of resources affect positively the intention of use our IoT cloud solution and, as a consequence, the great quality of the education process.

During the past decades, the volume of big dataavailable in remote sensing (RS) applications has grown significantly.In addition, a number of applications related tomonitoring human activity are being developed based on thiskind of data. This has considerably increased the demand forRS processing methods. In this sense, the scientific community isfacing the challenge of how to maximize the potential of the datathat are produced in a fast and efficient way. In particular, theprovision of processing algorithms that can be developed in aneasy way is a fundamental problem for the RS community, due tothe large volume of data offered by different portals and agencies,and the need for algorithms developed on different platforms andusing a variety of programming frameworks. To address thesechallenges, this paper takes advantage of social media tools tobring images and algorithms closer to users. In particular, a newsystem based on the Telegram messaging application Bot (called@ThuleRSbot) has been developed to provide a wide variety ofRS image processing methods to users under the same interface.The system has been developed using the Python language andthe Telegram Bot application program interface (API). The mostremarkable characteristics of the system are: 1) a completelyopen architecture that facilitates the incorporation of new algorithmswithout effort, and 2) an easy way to automatically gathersatellite images from the Sentinel Hub platform.

Convolutional neural networks (CNNs) have becomequite popular for solving many different tasks in remote sensingdata processing. The convolution is a linear operation whichextracts features from the input data. However, nonlinear operationsare able to better characterize the internal relationshipsand hidden patterns within complex remote sensing data, suchas hyperspectral images (HSIs). Morphological operations arepowerful nonlinear transformations for feature extraction thatpreserve the essential characteristics of the image, such asborders, shape and structural information. In this paper, anew end-to-end morphological deep learning framework (calledMorphConvHyperNet) is introduced. The proposed approachefficiently models nonlinear information during the trainingprocess of HSI classification. Specifically our method includesspectral and spatial morphological blocks to extract relevantfeatures from the HSI input data. These morphological blocksconsist of two basic 2D morphological operators (erosion anddilation) in the respective layers, followed by a weighted combinationof the feature maps. Both layers can successfully encodethe nonlinear information related to shape and size, playing animportant role in classification performance. Our experimentalresults, obtained on five widely used HSIs, reveal that our newlyproposed MorphConvHyperNet offers comparable (and evensuperior) performance when compared to traditional 2D and 3DCNNs for HSI classification.

Background Cytomegalovirus infection is the most frequent viral congenital infection, with possible consequences such as deafness, or psychomotor retardation. In 2016, the French High Council of Public Health was mandated to update recommendations regarding prevention of cytomegalovirus infection in pregnant women. We summarize a critical appraisal of knowledge and deterministic decision analysis comparing the current no-screening situation to serological screening during pregnancy, and to hygiene promotion. Methods Screening was defined as systematic serological testing, during the first trimester, with repeated tests as needed, to all pregnant women. Outcomes were: 1) severe sequela: intellectual deficiency with IQ ≤ 50 or hearing impairment Results Relevant data were scarce or imprecise. We estimated that 4352 maternal primary infections would result in 1741 foetal infections, and an unknown number of maternal reinfections would result in 1699 foetal infections. There would be 788 cytomegalovirus-related consequences, including 316 foetal deaths or terminations of pregnancy, and 424 moderate and 48 severe sequelae. Screening would result in a 1.66-fold increase of poor outcomes, mostly related to a 2.93-fold increase in deaths and terminations of pregnancy, not compensated by the decrease in severe symptomatic newborns. The promotion of hygiene would result in a 0.75-fold decrease of poor outcomes, related to both a decrease in severe sequelae among symptomatic newborns (RR = 0.75; min-max: 1.00–0.68), and in deaths and terminations of pregnancy (RR = 0.75; min-max: 0.97–0.68). Conclusions Prevention of cytomegalovirus infection during pregnancy should promote hygiene; serological screening should not be recommended.

ObjectiveTo determine whether there is a disparity in access to telemedical care that may be a function of socioeconomic status, language, or other demographic factors during the peak of the coronavirus disease 2019 (COVID-19) pandemic at a highly affected urban center (Montefiore Medical Center) in Bronx, NY.MethodsWe retrospectively investigated potential patient characteristics that might be associated with an increased likelihood of receiving a telephone visit as opposed to a televideo visit for patients followed in the pediatric neurology, adult epilepsy, and general neurology practices at Montefiore Medical Center during the 30-day period starting April 2, 2020, at the peak of the COVID-19 pandemic in New York.ResultsWe found that patients who had telephone encounters, as opposed to televideo encounters, were overall older, less likely to have commercial insurance, and more likely to have Medicaid. Among pediatric patients, a preferred language other than English was also associated with a higher proportion of telephone encounters. New patients in both the adult and pediatric groups were more likely to have televideo visits.ConclusionsOur findings identify demographic factors, including age, insurance type, and language preference, which may play a role in access to televideo encounters among neurology patients in an urban center during the COVID-19 pandemic. We suggest several potential practice, institution, and community-based interventions, which might further expand access to televideo care for neurology patients.

Background Venous thromboembolism (VTE) is a leading cause of preventable harm in hospitalized patients. However, many doses of prescribed pharmacologic VTE prophylaxis are frequently missed. We investigated the effect of a patient-centered education bundle on missed doses of VTE prophylaxis in a community hospital. Methods We performed a pre-post analysis examining missed doses of VTE prophylaxis in a community hospital. A real-time alert from the electronic health record system facilitated the delivery of a patient education bundle intervention. We included all patient visits on a single floor where at least 1 dose of VTE prophylaxis was prescribed during pre- (January 1, 2018, - November 31, 2018) and post- (January 1 - June 31, 2019) intervention periods. Outcomes included any missed dose (primary) and reasons for missed doses (refusal, other [secondary]) and were compared between both periods. Results 1,614 patient visits were included. The proportion of any missed dose significantly decreased (13.8% vs. 8.2% [OR, 0.56; 95% CI, 0.48, 0.64]) between the pre-post intervention periods. Patient refusal was the most frequent reason for missed doses. In the post-intervention period, patient refusal significantly decreased from 8.8% to 5.0% (OR, 0.54; 95% CI, 0.46, 0.64). Similarly, other reasons for missed doses significantly decreased from 5.0% to 3.2% (OR, 0.62; 95% CI, 0.51, 0.77). Conclusions A real-time alert-triggered patient-centered education bundle developed and tested in an academic hospital, significantly reduced missed doses of prescribed pharmacologic VTE prophylaxis when disseminated to a community hospital.

Maple syrup production (“maple sugaring”) is an important cultural and economic activity in eastern Canada and the northeastern United States (U.S.A.). Climate change is a concern for maple producers because maple sugaring is dependent upon climate for both sap sugar content as well as for the specific patterns of freezing and thawing in springtime needed for sap flow. Meanwhile, the maple industry itself is in transition: historically, maple sugaring has been a small-scale seasonal activity, but today large producers tap thousands of trees, contributing to a widening gap in production volumes and production practices between small- and large-scale producers. Drawing on original survey data from 354 maple producers in Canada and the United States we ask how attitudes towards climate change risks and adaptation strategies differ between smaller-scale and larger-scale producers. Findings suggest that small-scale, medium-scale and large-scale maple producers have different perceptions of climate change risks, report different impacts of climate change on maple sap yields, and have different levels of willingness to use adaptation strategies ranging from tapping more trees to adopting new production technologies. In multivariate ordered logistic regression models controlling for producer age, education, and political affiliation, we consistently find the strongest correlates of attitudes towards climate change risks—and the willingness or ability of producers to adopt different adaptation strategies—are country of residence (with preferred adaptation strategies varying across Canada versus the U.S.A.) and producer scale (with larger producers adopting more adaptation strategies). We conclude that more detailed and up-to-date information on the different types of maple operations—including effects of climate change on maple sugaring and available adaptation strategies across geographies and across scales of production—is needed to help maple producers, maple producer associations, and policymakers understand how the maple industry is evolving and how best to prepare for the future.

BACKGROUND Disparities in THA use may lead to inequitable care. Prior research has focused on disparities based on individual-level and isolated socioeconomic and demographic variables. To our knowledge, the role of composite, community-level geographic socioeconomic disadvantage has not been studied in the United States. As disparities persist, exploring the potential underlying drivers of these inequities may help in developing more targeted recommendations on how to achieve equitable THA use. QUESTIONS/PURPOSES (1) Is geographic socioeconomic disadvantage associated with decreased THA rates in Medicare-aged patients? (2) Do these associations persist after adjusting for differences in gender, race, ethnicity, and proximity to hospitals performing THA? METHODS In a study with a cross-sectional design, using population-based data from five-digit ZIP codes in Maryland, USA, from July 1, 2012 to March 31, 2019, we included all inpatient and outpatient primary THAs performed in individuals 65 years of age or older at acute-care hospitals in Maryland, as reported in the Health Services Cost Review Commission database. This database was selected because it provided the five-digit ZIP code data necessary to answer our study question. We excluded THAs performed for nonelective indications. We examined the annual rate of THA in our study population for each Maryland ZIP code, adjusted for differences across areas in distributions of gender, race, ethnicity, and distance to the nearest hospital performing THAs. Four hundred fourteen ZIP codes were included, with an overall mean ± SD THA rate of 371 ± 243 per 100,000 persons 65 years or older, a rate similar to that previously reported in individuals aged 65 to 84 in the United States. Statistical significance was assessed at α = 0.05. RESULTS THA rates were higher in more affluent areas, with the following mean rates per 100,000 persons 65 years or older: 422 ± 259 in the least socioeconomically disadvantaged quartile, 339 ± 223 in the second-least disadvantaged, 277 ± 179 in the second-most disadvantaged, and 214 ± 179 in the most-disadvantaged quartile (p < 0.001). After adjustment for distributions in gender, race, ethnicity, and hospital proximity, we found that geographic socioeconomic disadvantage was still associated with THA rate. Compared with the least-disadvantaged quartile, the second-least disadvantaged quartile had 63 fewer THAs per 100,000 people (95% confidence interval 12 to 114), the second-most disadvantaged quartile had 136 fewer THAs (95% CI 62 to 211), and the most-disadvantaged quartile had 183 fewer THAs (95% CI 41 to 325). CONCLUSION Geographic socioeconomic disadvantage may be the underlying driver of disparities in THA use. Although our study does not determine the "correct" rate of THA, our findings support increasing access to elective orthopaedic surgery in disadvantaged geographic communities, compared with prior research and efforts that have studied and intervened on the basis of isolated factors such as race and gender. Increasing access to orthopaedic surgeons in disadvantaged neighborhoods, educating physicians about when surgical referral is appropriate, and educating patients from these geographic communities about the risks and benefits of THA may improve equitable orthopaedic care across neighborhoods. Future studies should explore disparities in rates of appropriate THA and the role of density of orthopaedic surgeons in an area. LEVEL OF EVIDENCE Level III, therapeutic study.

Background Limited work has been done in predicting discharge disposition in trauma patients; most studies use single institutional data and have limited generalizability. This study develops and validates a model to predict, at admission, trauma patients’ discharge disposition using NTDB, transforms the model into an easy-to-use score, and subsequently evaluates its generalizability on institutional data. Methods NTDB data were used to build and validate a binary logistic regression model using derivation-validation (ie, train-test) approach to predict patient disposition location (home vs non-home) upon admission. The model was then converted into a trauma disposition score (TDS) using an optimization-based approach. The generalizability of TDS was evaluated on institutional data from a single Level I trauma center in the U.S. Results A total of 614 625 patients in the NTDB were included in the study; 212 684 (34.6%) went to a non-home location. Patients with a non-home disposition compared to home had significantly higher age (69 ± 19.7 vs 48.3 ± 20.3) and ISS (11.2 ± 8.2 vs 8.2 ± 6.3); P < .001. Older age, female sex, higher ISS, comorbidities (cancer, cardiovascular, coagulopathy, diabetes, hepatic, neurological, psychiatric, renal, substance abuse), and Medicare insurance were independent predictors of non-home discharge. The logistic regression model’s AUC was 0.8; TDS achieved a correlation of 0.99 and performed similarly well on institutional data (n = 3161); AUC = 0.8. Conclusion We developed a score based on a large national trauma database that has acceptable performance on local institutions to predict patient discharge disposition at the time of admission. TDS can aid in early discharge preparation for likely-to-be non-home patients and may improve hospital efficiency.

Maxillofacial trauma costs emergency departments approximately one billion dollars annually. Facial trauma has increased since 2000 and has been attributed to both increased use of computed tomography and interpersonal violence. Alcohol, male sex, and age 18-35 years are significant risk factors for interpersonal violence. This study is novel in using a large database to look at the effect of alcohol on clinical outcomes in facial trauma. A data analysis was performed in Stata/MP 14.2 using variables coded from the National Trauma Data Bank (NTDB); logistic regression was applied. A total 580,313 patient records were analyzed. Operations for facial fractures were performed in 20.19% of cases (n = 117,139). A positive alcohol test reduced the odds of requiring operative fixation in both the unadjusted (odds ratio (OR) 0.8, 95% confidence interval (CI) 0.79-0.82, P 0.001) and adjusted (OR 0.67, 95% CI 0.66-0.68, P 0.001) models. Age and being struck (adjusted OR 1.99, 95% CI 1.91-2.07, P 0.001) or shot (adjusted OR 1.95, 95% CI 1.84-2.06, P 0.001) had a significant effect on operative fixation. Injury mechanisms related to interpersonal violence appeared to have higher operative fixation rates. This study did not find a correlation between acute intoxication and the need for an operative intervention. This further demonstrates the multifactorial nature of facial trauma and stresses the importance of injury and violence prevention on clinical outcomes.

In response to the COVID-19 pandemic epicenter in Bronx, NY, the Montefiore Neuroscience Center required rapid and drastic changes when considering the delivery of neurologic care, health and safety of staff, and continued education and safety for house staff. Health care leaders rely on principles that can be in conflict during a disaster response such as this pandemic, with equal commitments to ensure the best care for those stricken with COVID-19, provide high-quality care and advocacy for patients and families coping with neurologic disease, and advocate for the health and safety of health care teams, particularly house staff and colleagues who are most vulnerable. In our attempt to balance these principles, over 3 weeks, we reformatted our inpatient neuroscience services by reducing from 4 wards to just 1, in the following weeks delivering care to over 600 hospitalized patients with neuro-COVID and over 1,742 total neuroscience hospital bed days. This description from members of our leadership team provides an on-the-ground account of our effort to respond nimbly to a complex and evolving surge of patients with COVID in a large urban hospital network. Our efforts were based on (1) strategies to mitigate exposure and transmission, (2) protection of the health and safety of staff, (3) alleviation of logistical delays and strains in the system, and (4) facilitating coordinated communication. Each center's experience will add to knowledge of best practices, and emerging research will help us gain insights into an evidence-based approach to neurologic care during and after the COVID-19 pandemic.

Objective A 2007 study performed at Montefiore Medical Center (Bronx, NY) identified high prevalence of reduced bone density in an urban population of patients with epilepsy and suggested that bone mineralization screenings should be regularly performed for these patients. We conducted a long‐term follow‐up study to determine whether bone mineral density (BMD) loss, osteoporosis, and fractures have been successfully treated or prevented. Methods In the current study, patients from the 2007 study who had two dual‐energy absorptiometry (DXA) scans performed at least 5 years apart were analyzed. The World Health Organization (WHO) criteria to diagnose patients with osteopenia or osteoporosis were used, and each patient's probability of developing fractures was calculated with the Fracture Risk Assessment Tool (FRAX). Results The median time between the first and second DXA scans for the 81 patients analyzed was 9.4 years (range 5‐14.7). The median age at the first DXA scan was 41 years (range 22‐77). Based on WHO criteria, 79.0% of patients did not have worsening of bone density, while 21.0% had new osteopenia or osteoporosis; many patients were prescribed treatment for bone loss. Older age, increased duration of anti‐epileptic drug (AED) usage, and low body mass index (BMI) were risk factors for abnormal BMDs. Based on the first DXA scan, the FRAX calculator estimated that none of the patients in this study had a 10‐year risk of more than 20% for developing major osteoporotic fracture (hip, spine, wrist, or humeral fracture). However, in this population, 11 patients (13.6%) sustained a major osteoporotic fracture after their first DXA scan. Significance Despite being routinely screened and frequently treated for bone mineral density loss and fracture prevention, many patients with epilepsy suffered new major osteoporotic fractures. This observation is especially important as persons with epilepsy are at high risk for falls and traumas.

Advances in deep learning (DL) have allowed for the development of more complex and powerful neural architectures. The adoption of deep convolutional-based architectures with residual learning [residual networks (ResNets)] has reached the state-of-the-art performance in hyperspectral image (HSI) classification. Traditionally, ResNets have been considered as stacks of discrete layers, where each one obtains a hidden state of the input data. This formulation must deal with very deep networks, which suffer from an important data degradation as they become deeper. Moreover, these complex models exhibit significant requirements in terms of memory due to the amount of parameters that need to be fine tuned. This leads to inadequate generalization and loss of accuracy. In order to address these issues, this article redesigns the ResNet as a continuous-time evolving model, where hidden representations (or states) are obtained with respect to time (understood as the depth of the network) through the evaluation of an ordinary differential equation (ODE), which is combined with a deep neural architecture. Our experimental results, conducted with four well-known HSI data sets, indicate that redefining deep networks as continuous systems through ODEs offers flexibility when processing and classifying these kinds of remotely sensed data, achieving significant performance even when a very few training samples are available.

Hyperspectral imaging (HSI) collects hundreds of images over large spatial observation areas on the Earth’s surface, recording scenes at different wavelength channels and providing a vast amount of information. Recurrent neural networks (RNNs) have been widely used for the classification of HSI datasets, understood as a single sequence of pixel vectors with high dimensionality. However, the RNN model scales poorly when dealing with HSI scenes with large dimensionality. In order to mitigate this problem, this paper presents a new RNN classifier based on simple recurrent units that performs HSI classification in a highly scalable and efficient way. Our experimental results (conducted on four real HSI datasets) reveal very good performance, not only in terms of classification accuracy (in line with existing methods), but also in terms of computational performance when dealing with large datasets.

Background Venous thromboembolism (VTE) is a cause of considerable morbidity and mortality in hospitalized patients. An evidence-based algorithm was developed and implemented at our institution to guide perioperative VTE prophylaxis management. Objective We evaluated compliance with prescription of risk-appropriate VTE prophylaxis and administration of prescribed VTE prophylaxis in neurosurgery patients. Methods This was a retrospective analysis of postoperative neurosurgery patients at a single institution with subsequent diagnosis of acute VTE during their inpatient stay. Descriptive statistics were used to characterize pharmacologic VTE prophylaxis and prescribing patterns. Results The incidence of VTE in our neurosurgery population was 248/13,913 (1.8%). Of the 123 patients, the median time to VTE diagnosis was 96 hours after surgery (interquartile range [IQR], 58–188 hours). A total of 108 patients (87.8%) were prescribed risk-appropriate VTE prophylaxis, among whom 61 (56.5%) received all doses as prescribed. Fifty-three patients (43.1%) missed ≥1 dose of prescribed prophylaxis and the median missed doses was 3 (IQR, 0–3). The median time to first dose of pharmacologic VTE prophylaxis was 42 hours (IQR, 28–51). More than half (n = 63, 51.2%) of the VTE risk assessments contained ≥1 error, of which 15 (23.8%) would have resulted in a change in recommendation. Conclusions Our evidence-based VTE prophylaxis algorithm was not accurately completed in more than half of patients. Many patients who developed VTE had a defect in their VTE prophylaxis management during their inpatient stay. Research to improve optimal VTE prevention practice in neurosurgery patients is needed.

Nationally, it was estimated that 11.4 million people misused opioids in 2017 with more than 47,000 opioid-related deaths. Although medication-assisted treatment (MAT) has been effective in enhancing treatment retention and decreasing frequency of opioid use, relapse rates for opioids and other substances remain high, emphasizing the importance of investigating novel interventions to augment MAT. One potential treatment approach is repetitive transcranial magnetic stimulation (rTMS)-a noninvasive, electrophysiological method of neuromodulation. Recently published studies of rTMS in individuals with alcohol, nicotine, and cocaine use disorder have suggested that this treatment shows promise in reducing cravings and substance use. The literature specific to rTMS and opioid use disorder (OUD) is limited to a single published study in heroin users, which showed that a single session of rTMS to the left dorsolateral prefrontal cortex (DLPFC) reduced cue-induced craving, with a further reduction following 5 consecutive days of rTMS. The following case report involved a 25-year-old Caucasian male diagnosed with OUD and cocaine use disorder. This subject continued to demonstrate ongoing substance use despite participating in comprehensive MAT with buprenorphine/naloxone in combination with psychosocial interventions. He was administered 7 separate sessions of rTMS targeting the left DLPFC. Substance-related cues were presented prior to, during, and following these rTMS administration sessions and the subject rated his substance cravings via a 100-point Visual Analog Scale. When compared with his cue-induced craving ratings, there was a mean reduction in craving for heroin and cocaine by ∼60% to 82% following the 7 administration sessions. Although this is a single case, further investigation of rTMS as an augmentation strategy for OUD and polysubstance use is warranted. (PsycINFO Database Record (c) 2020 APA, all rights reserved).

OBJECTIVE: The purpose of this study was to evaluate the impact of optimization of preoperative comorbidities by nonsurgical clinicians on short-term postoperative outcomes. SUMMARY BACKGROUND DATA: Preoperative comorbidities can have substantial effects on operative risk and outcomes. The modifiability of these comorbidity-associated surgical risks remains poorly understood. METHODS: We identified patients with a major comorbidity (eg, diabetes, heart failure) undergoing an elective colectomy in a multipayer national administrative database (2010–2014). Patients were included if they could be matched to a preoperative surgical clinic visit within 90 days of an operative intervention by the same surgeon. The explanatory variable of interest (“preoperative optimization”) was defined by whether the patient was seen by an appropriate nonsurgical clinician between surgical consultation and subsequent surgery. We assessed the impact of an optimization visit on postoperative complications with use of propensity score matching and multilevel, multivariable logistic regression. RESULTS: We identified 4531 colectomy patients with a major potentially modifiable comorbidity (propensity weighted and matched effective sample size: 6037). After matching, the group without an optimization visit had a higher rate of complications (34.6% versus 29.7%, P = 0.001). An optimization visit conferred a 31% reduction in the odds of a complication (P < 0.001) in an adjusted analysis. Median preoperative costs increased by $684 (P < 0.001) in the optimized group, and a complication increased total costs of care by $14,724 (P < 0.001). CONCLUSIONS AND RELEVANCE: We demonstrated an association between use of nonsurgical clinician visits by comorbid patients prior to surgery and a significantly lower rate of complications. These findings support the prospective study of preoperative optimization as a potential mechanism for improving postoperative outcomes.

West Virginia (WV) is situated at the epicenter of the opioid epidemic with the highest rates of overdose deaths and some of the lowest rates of access to life saving evidence-based medication assisted treatment (MAT) for patients with opioid use disorder (OUD). WV used a modified hub-and-spoke model to build organizational capacity for facilities to use buprenorphine to treat patients with OUD and to provide ongoing case consultation. The purpose of this study is to 1) describe the group-base model of buprenorphine treatment and the model used to build organizational capacity, 2) to describe the preliminary results of buprenorphine expansion in WV and 3) to report preliminary data describing and comparing the characteristics of the patients served across five hubs. A single Coordinating Center uses video conferencing to train hubs and provide ongoing case consultation, as well as clinical support. Hubs were trained to deliver a buprenorphine treatment model that is multi-disciplinary and includes group-based medication management and psychosocial therapy. Five regional hubs independently treat patients and are leading MAT expansion in their local areas by training and mentoring spokes (n = 13). As a result of the WV STR funding, 14 health care facilities have started to use buprenorphine, 56 health professionals were trained and 196 patients with OUD have been treated. There were few sociodemographic characteristic differences across patients treated at the five hubs, while there were differences in self-reported alcohol and drug use in the 30 days prior to intake. Additional research is needed to determine whether the WV modified hub-and-spoke model resulted in statistically significant improvements in buprenorphine treatment capacity; there is a need to address MAT stigma and regulatory barriers in order to ensure the long-term sustainability of the buprenorphine expansion.

Detection of surface material based on hyperspectral imaging (HSI) analysis is an important and challenging task in remote sensing. It is widely known that spectral-spatial data exploitation performs better than traditional spectral pixel-wise procedures. Nowadays, convolutional neural networks (CNNs) have shown to be a powerful deep learning (DL) technique due their strong feature extraction ability. CNNs not only combine spectral-spatial information in a natural way, but have also shown to be able to learn translation-equivariant representations, i.e. a translation of input features into an equivalent internal CNN feature map. This provides great robustness to spatial feature locations. However, as far as we know, CNNs do not exhibit a natural way to exploit rotation equivariance, i.e. make use of the fact that data patches in a HSI data cube are observed in different orientations due to their orientation or on the varying paths/orbits of the airborne/spaceborne spectrometers. This article presents a rotation-equivariant CNN2D model for HSI analysis, where traditional convolution kernels have been replaced by circular harmonic filters (CHFs). The obtained results over three well-known HSI datasets showcase the potential of the approach.

Denoising is a common preprocessing step prior to the analysis and interpretation of hyperspectral images (HSIs). However, the vast majority of methods typically adopted for HSI denoising exploit architectures originally developed for grayscale or RGB images, exhibiting limitations when processing high-dimensional HSI data cubes. In particular, traditional methods do not take into account the high spectral correlation between adjacent bands in HSIs, which leads to unsatisfactory denoising performance as the rich spectral information present in HSIs is not fully exploited. To overcome this limitation, this article considers deep learning models-such as convolutional neural networks (CNNs)-to perform spectral-spatial HSI denoising. The proposed model, called HSI single denoising CNN (HSI-SDeCNN), efficiently takes into consideration both the spatial and spectral information contained in HSIs. Experimental results on both synthetic and real data demonstrate that the proposed HSI-SDeCNN outperforms other state-of-the-art HSI denoising methods. Source code: https://github.com/mhaut/HSI-SDeCNN.