Separate Set Research Articles

Artificial intelligence-estimated biological heart age using 12 lead electrocardiogram predicts mortality and cardiovascular outcomes

Abstract Funding Acknowledgements Type of funding sources: None. Main funding source(s): none Background There is a paucity of data on the artificial intelligence-estimated biological ECG heart age (AI ECG-heart age) for predicting cardiovascular outcomes as distinct from the chronological age (CA). Purpose We sought to investigate whether a deep learning-based algorithm to estimate the AI ECG-heart age using standard 12-lead ECGs predicted mortality and cardiovascular outcomes. Methods We trained and validated a deep neural network using raw ECG digital data from 425,051 12-lead ECGs acquired between January 2006 and December 2021. The network performed a holdout test using a separate set of 97,058 ECGs. Additionally, randomly age-sex matched patients with reduced and preserved ejection fraction (EF) were compared. Results The deep neural network was trained to estimate the AI ECG-heart age (the mean absolute error 5.8 ± 3.9 years and R squared of 0.7 (r=0.84, p<0.0001) (Fig 1A). In the Cox proportional-hazards models after adjusting for relevant co-morbidity factors, the subjects with an AI ECG-heart age of seven years older than the chronological age had higher all-cause mortality (HR 1.62 [1.43-1.84]) and major adverse cardiovascular events (MACEs) (HR 1.92 [1.65-2.23]), while those under seven years had an inverse relationship (HR 0.86 [0.77-0.95] for all-cause mortality; HR 0.73 [0.63-0.84] for MACEs) (Fig 2). Subjects with a reduced EF had a substantially higher mean AI ECG heart-age, QRS duration, and corrected QT intervals than those with a preserved EF (all p<0.001) (Fig 1B). Conclusion The biological heart age estimated by AI had a significant impact on mortality and MACE. Those data suggested that the AI ECG-heart age might facilitate the primary prevention and health care for cardiovascular outcomes.

Performance of a Volatile Corrosion Inhibitor for Mitigating Corrosion Under Insulation

This study evaluated the efficacy of a commercial volatile corrosion inhibitor (VCI) in mitigating the corrosion of carbon steel insulated with mineral wool insulation, especially as a retrofitted solution to corrosion under insulation (CUI) management. When VCI was added in dry insulation before the test solution was introduced, the severity of CUI was reduced after 14 d of exposure. However, when the exposure time was extended to 90 d, no significant mitigation effect was observed. Furthermore, VCI was ineffective when added to precorroded insulated systems, indicating that it could not be used as a retrofitted solution to alleviate CUI problems. A separate set of tests with bare steel samples exposed to 10 mL VCI in closed jars showed that VCI was adsorbed on the corrosion products but could not effectively impede the propagation of localized corrosion. The possible influences of insulation jacketing, insulation properties, VCI dosages, and dosing methods were also discussed, suggesting the need for further investigations.

Mergers of neutron stars and black holes with cores of giant stars: a population synthesis study

ABSTRACT We perform population synthesis of massive binaries to study the mergers of neutron stars (NSs) and black holes (BHs) with the cores of their giant secondaries during common envelope evolution (CEE). We use different values of the efficiency parameter αCE in the framework of the energy formalism for traditional CEE (αCE ≤ 1) and including additional energy sources to unbind the envelope (αCE > 1). We constrain the possible values of αCE by comparing the results of our simulations with local rate densities of binary compact object mergers as inferred from gravitational-wave observations. We find two main evolutionary pathways of binary systems that result in NS–core mergers, while only one of them can also lead to the merger of a BH with the core of the giant star. We explore the zero-age main-sequence statistical properties of systems that result in NS/BH–core mergers and find that the two evolutionary channels correspond to a bimodal distribution of orbital separations. We estimate the percentage of the mergers’ event rates relative to core-collapse supernovae. We include the effect of mass accreted by the NS/BH during CEE in a separate set of simulations and find that it does not affect the mergers’ event rates.

Control oriented modeling of twin-screw granulation in the ConsiGmaTM-25 production plant

ConsiGmaTM-25 is a continuous production plant integrating a twin-screw granulation, fluid bed drying, granule conditioning, and a tableting unit. The particle size distribution (PSD), active pharmaceutical ingredient (API) content, and liquid content of wet granules after twin-screw granulation affect the quality of intermediate and final products. This paper proposes methods for real-time monitoring of these quantities and control-oriented modeling of the granulator.The PSD of wet granules is monitored via an in-line process analytical technology (PAT) probe based on the spatial velocimetry principle. The algorithm for signal processing and evaluation of PSD characteristics is developed and applied to the acquired PSD data. A dynamic process model predicting PSD characteristics from granulation parameters is trained via the local linear model tree (LoLiMoT) approach. The experimental data required for the model training are collected via systematically designed excitation runs. Finally, the performance of the identified model is examined and verified by means of a new set of validation runs. Furthermore, an in-line PAT probe based on Raman spectroscopy is developed and integrated after the granulator. The API- and liquid content of produced wet granules are evaluated from the spectral data by means of chemometric modeling, and chemometric models are validated on a separate set of experimental data. The solutions proposed in this research can be used as a reliable (and necessary) basis for the development of advanced quality-by-design control concepts (e.g., PSD process control). Such concepts would ultimately improve the ConsiGmaTM-25 process performance in terms of robustness against disturbances and quality of intermediate and final products.

Astrogliosis and associated CSPG upregulation adversely affect dendritogenesis, spinogenesis and synaptic activity in the cerebellum of a double-hit rat model of protein malnutrition (PMN) and lipopolysaccharide (LPS) induced bacterial infection

The extracellular matrix (ECM) plays a vital role in growth, guidance and survival of neurons in the central nervous system (CNS). The chondroitin sulphate proteoglycans (CSPGs) are a type of ECM proteins that are crucial for CNS homeostasis. The major goal of this study was to uncover the effects of astroglial activation and associated intensified expression of CSPGs on dendritogenesis, spinogenesis as well as on synaptic activity in cerebellum following protein malnutrition (PMN) and lipopolysaccharide (LPS) induced bacterial infection. Female Wistar albino rats (3 months old) were switched to control (20% protein) or low protein (LP, 8% protein) diet for 15 days followed by breeding. A set of pups born to control/LP mothers and maintained on respective diets throughout the experimental period constituted the control and LP groups, while a separate set of both control and LP group pups exposed to bacterial infection by a single intraperitoneal injection of LPS (0.3 mg/ kg body weight) on postnatal day-9 (P-9) constituted control+LPS and LP+LPS groups respectively. The consequences of astrogliosis induced CSPG upregulation on cerebellar cytoarchitecture and synaptic activity were studied using standard immunohistochemical and histological tools on P-21 and 6 months of age. The results revealed reactive astrogliosis and associated CSPG upregulation in a double-hit model of PMN and LPS induced bacterial infection resulted in disrupted dendritogenesis, reduced postsynaptic density protein (PSD-95) levels and a deleterious impact on normal spine growth. Such alterations frequently have the potential to cause synaptic dysregulation and inhibition of plasticity both during development as well as adulthood. At the light of our results, we can envision that upregulation of CSPGs in PMN and LPS co-challenged individuals might emerge as an important modulator of brain circuitry and a major causative factor for many neurological disorders.

The effects of corporate governance mechanisms on voluntary corporate carbon disclosures: evidence from the emerging economy

PurposeThis study examines the effects of corporate governance mechanisms on voluntary corporate carbon disclosure in Bangladeshi firms.Design/methodology/approachTo investigate the association between corporate governance mechanisms and corporate carbon disclosures, this study employs ordinary least square (OLS) methods. To mitigate the potential endogeneity concerns, the authors also introduce firm fixed effect (FE) and random effect (RE). Primarily, the study sample includes 250 firm-year observations over the period 2015–2019 for listed companies on the Dhaka Stock Exchange (DSE) in Bangladesh. Subsequently, corporate governance mechanisms that influence voluntary carbon disclosure were examined using both univariate and OLS models.FindingsThe findings of this study suggest that firms with a larger board size and more independent directors have a positive impact on the firm's intensity to disclose carbon-related information. However, no evidence has been found of the existence of an environmental committee, and the presence of female directors on the board tends to be associated with a higher level of voluntary corporate carbon disclosure.Originality/valueThe study offers necessary evidence of the determinants of corporate carbon disclosures, which will be useful for managers, senior executives, policymakers and regulatory bodies. To improve corporate governance practices and formulate separate sets of regulations and reporting criteria, disclosing extensive and holistic carbon-related information obligatory. Further, the outcomes of this study based on Bangladeshi firms can be comprehensive for other developing countries to take precautions to tackle the effect of global climate change.

Sarcopenia Prediction for Elderly People Using Machine Learning: A Case Study on Physical Activity.

Sarcopenia is a well-known age-related disease that can lead to musculoskeletal disorders and chronic metabolic syndromes, such as sarcopenic obesity. Numerous studies have researched the relationship between sarcopenia and various risk factors, leading to the development of predictive models based on these factors. In this study, we explored the impact of physical activity (PA) in daily life and obesity on sarcopenia prediction. PA is easier to measure using personal devices, such as smartphones and watches, or lifelogs, than using other factors that require medical equipment and examination. To demonstrate the feasibility of sarcopenia prediction using PA, we trained various machine learning models, including gradient boosting machine (GBM), xgboost (XGB), lightgbm (LGB), catboost (CAT), logistic regression, support vector classifier, k-nearest neighbors, random forest (RF), multi-layer perceptron, and deep neural network (DNN), using data samples from the Korea National Health and Nutrition Examination Survey. Among the models, the DNN achieved the most precise accuracy on average, 81%, with PA features across all data combinations, and the accuracy increased up to 90% with the addition of obesity information, such as total fat mass and fat percentage. Considering the difficulty of measuring the obesity feature, when adding waist circumference to the PA features, the DNN recorded the highest accuracy of 84%. This model accuracy could be improved by using separate training sets according to gender. As a result of measurement with various metrics for accurate evaluation of models, GBM, XGB, LGB, CAT, RF, and DNN demonstrated significant predictive performance using only PA features including waist circumference, with AUC values at least around 0.85 and often approaching or exceeding 0.9. We also found the key features for a highly performing model such as the quantified PA value and metabolic equivalent score in addition to a simple obesity measure such as body mass index (BMI) and waist circumference using SHAP analysis.

Reduced immunogenicity of a live Salmonella enterica serovar Typhimurium vaccine in aged mice.

Non-typhoidal Salmonella (NTS) is responsible for a high burden of foodborne infections and deaths worldwide. In the United States, NTS infections are the leading cause of hospitalizations and deaths due to foodborne illnesses, and older adults (≥65 years) are disproportionately affected by Salmonella infections. Due to this public health concern, we have developed a live attenuated vaccine, CVD 1926 (I77 ΔguaBA ΔclpP ΔpipA ΔhtrA), against Salmonella enterica serovar Typhimurium, a common serovar of NTS. Little is known about the effect of age on oral vaccine responses, and due to the decline in immune function with age, it is critical to evaluate vaccine candidates in older age groups during early product development. In this study, adult (six-to-eight-week-old) and aged (18-month-old) C57BL/6 mice received two doses of CVD 1926 (109 CFU/dose) or PBS perorally, and animals were evaluated for antibody and cell-mediated immune responses. A separate set of mice were immunized and then pre-treated with streptomycin and challenged orally with 108 CFU of wild-type S. Typhimurium SL1344 at 4 weeks postimmunization. Compared to PBS-immunized mice, adult mice immunized with CVD 1926 had significantly lower S. Typhimurium counts in the spleen, liver, and small intestine upon challenge. In contrast, there were no differences in bacterial loads in the tissues of vaccinated versus PBS aged mice. Aged mice exhibited reduced Salmonella-specific antibody titers in the serum and feces following immunization with CVD 1926 compared to adult mice. In terms of T cell responses (T-CMI), immunized adult mice showed an increase in the frequency of IFN-γ- and IL-2-producing splenic CD4 T cells, IFN-γ- and TNF-α-producing Peyer's Patch (PP)-derived CD4 T cells, and IFN-γ- and TNF-α-producing splenic CD8 T cells compared to adult mice administered PBS. In contrast, in aged mice, T-CMI responses were similar in vaccinated versus PBS mice. CVD 1926 elicited significantly more PP-derived multifunctional T cells in adult compared to aged mice. These data suggest that our candidate live attenuated S. Typhimurium vaccine, CVD 1926, may not be sufficiently protective or immunogenic in older humans and that mucosal responses to live-attenuated vaccines decrease with increasing age.

Automated Detection of Anatomical Landmarks During Colonoscopy Using a Deep Learning Model.

Identification and photo-documentation of the ileocecal valve (ICV) and appendiceal orifice (AO) confirm completeness of colonoscopy examinations. We aimed to develop and test a deep convolutional neural network (DCNN) model that can automatically identify ICV and AO, and differentiate these landmarks from normal mucosa and colorectal polyps. We prospectively collected annotated full-length colonoscopy videos of 318 patients undergoing outpatient colonoscopies. We created three nonoverlapping training, validation, and test data sets with 25,444 unaltered frames extracted from the colonoscopy videos showing four landmarks/image classes (AO, ICV, normal mucosa, and polyps). A DCNN classification model was developed, validated, and tested in separate data sets of images containing the four different landmarks. After training and validation, the DCNN model could identify both AO and ICV in 18 out of 21 patients (85.7%). The accuracy of the model for differentiating AO from normal mucosa, and ICV from normal mucosa were 86.4% (95% CI 84.1% to 88.5%), and 86.4% (95% CI 84.1% to 88.6%), respectively. Furthermore, the accuracy of the model for differentiating polyps from normal mucosa was 88.6% (95% CI 86.6% to 90.3%). This model offers a novel tool to assist endoscopists with automated identification of AO and ICV during colonoscopy. The model can reliably distinguish these anatomical landmarks from normal mucosa and colorectal polyps. It can be implemented into automated colonoscopy report generation, photo-documentation, and quality auditing solutions to improve colonoscopy reporting quality.

Predicting motor and cognitive outcomes from MRIs of brain structure in children with acquired brain injury: A pilot study.

Acquired Brain Injury (ABI) describes a range of brain injuries occurring after birth, including tumor, traumatic brain injury or stroke. Although MRIs are routinely used for diagnosis, prediction of outcome following brain injury is challenging. Quantitative structural information from brain images may provide an opportunity to predict patient outcomes; however, due to the high prevalence of severe pathology in children with ABI, quantitative approaches must be robust to injury severity. In this pilot cross-sectional study, automated quantitative measures were extracted from the MRIs of a cohort of children with ABI (n=30, 8-16 years, follow up MRI taken 1.8-13.4 years after time of injury) as well as 36 typically developing controls with no brain injury (7-17 years) using a pathology-robust technique. Measures of brain volume, lesion volume and cortical morphology were associated with concurrent motor, behavioral, visual and communicative function using Least Absolute Shrinkage and Selection Operator (LASSO) regression. These regression models were validated on a separate test set (n=8 of the ABI cohort), which revealed significant correlations between measures of brain structure with motor, cognitive, visual and communicative function (r=0.65-0.85, all p<0.01). Furthermore, comparisons of the structural measures to the typically developing cohort revealed overall reductions in global grey matter volume among the ABI cohort, as well as cortical thinning in several cortical areas. These preliminary associations reveal that motor and behavioral function can be estimated from MRI alone, highlighting the potential utility of the proposed pathology-robust MRI quantification tools to provide estimates of long-term clinical prognosis of children with ABI following injury.

PO-04-210 NEW-ONSET DIABETES SCREENING USING ARTIFICIAL INTELLIGENCE-ENHANCED ELECTROCARDIOGRAM

Diabetes has a long asymptomatic period which can remain undiagnosed for years. Glycated hemoglobin (HbA1c) enables the discovery of asymptomatic diabetes. Yet, undiagnosed diabetes is present in 23.0% of all US adults with Diabetes and 3.4% of all US adults. American Diabetes Association (ADA) guidlines promote the use of risk tests that help identify candidates for diagnostic testing. To develop and evaluate an artificial intelligence (AI)-enhanced electrocardiogram algorithm (ECGAI-DM) for screening of Diabetes. We trained a neural network to estimate HbA1c using a 12-lead ECG and readily available demographics (ECG model), and evaluated its ability to screen for Diabetes in a complete outpatient population. The dataset comprised of patients (aged ≥18 years) with paired 12-lead, 10-second ECG and HbA1c test acquired during an outpatient encounter. Our primary outcome was the performance of the AI-enhanced ECG at classifying new-onset diabetes (HbA1c ≥ 6.5 in patients without a prior history of diabetes), which was assessed using the area under the receiver operator curve (AUC) and positive predictive value (PPV) at the sensitivity defined by the ADA risk test, with 95% bootstrap CIs. Population bias was addressed by reconstruction of a pseudopopulation with adjusted individual patient weights. We trained on 198,857 ECGs from 160,788 patients, and our test set comprised 34,106 ECGs from 30,593 patients, recorded between January 1, 2013 and September 17, 2021. The prevalence of Diabetes (HbA1c ≥ 6.5) was 4.9%. The ECGAI-DM significantly outperformed the ADA Risk test (AUC, 0.80 [95% CI, 0.79-0.81] vs. AUC, 0.68 [95% CI, 0.68-0.69]; p-value <0.001). The ECGAI-DM model restricted to using a single lead (Lead-I) outperformed the ADA risk test as well (AUC, 0.78 [95% CI, 0.78-0.79] vs. 0.68 [95% CI, 0.68-0.69]; p-value <0.001). Patients with false positive prediction by ECGAI-DM had a 3-fold risk for Diabetes over a 1 year follow up as compared to the true negative population, suggesting unmasking of latent high risk. In a separate test set, ECGAI-DM significantly outperformed 2 human Cardiologists (AUC 0.82 vs. 0.61, p<0.001) assessed on a class balanced 100 randomly sampled ECGs. An AI-enhanced ECG can automate the screening of patients for Diabetes with fewer false positives than questionnaire-based assessment. This also applies to single-lead ECGs commonly found on wearable devices, which suggests an avenue for community-wide opportunistic Diabetes screening.

Revealing Individual Neuroanatomical Heterogeneity in Alzheimer Disease Using Neuroanatomical Normative Modeling.

Alzheimer disease (AD) is highly heterogeneous, with marked individual differences in clinical presentation and neurobiology. To explore this, we used neuroanatomical normative modeling to index regional patterns of variability in cortical thickness. We aimed to characterize individual differences and outliers in cortical thickness in patients with AD, people with mild cognitive impairment (MCI), and controls. Furthermore, we assessed the relationships between cortical thickness heterogeneity and cognitive function, β-amyloid, phosphorylated-tau, and ApoE genotype. Finally, we examined whether cortical thickness heterogeneity was predictive of conversion from MCI to AD. Cortical thickness measurements across 148 brain regions were obtained from T1-weighted MRI scans from 62 sites of the Alzheimer's Disease Neuroimaging Initiative. AD was determined by clinical and neuropsychological examination with no comorbidities present. Participants with MCI had reported memory complaints, and controls were cognitively normal. A neuroanatomical normative model indexed cortical thickness distributions using a separate healthy reference data set (n = 33,072), which used hierarchical Bayesian regression to predict cortical thickness per region using age and sex, while adjusting for site noise. Z-scores per region were calculated, resulting in a Z-score brain map per participant. Regions with Z-scores <-1.96 were classified as outliers. Patients with AD (n = 206) had a median of 12 outlier regions (out of a possible 148), with the highest proportion of outliers (47%) in the parahippocampal gyrus. For 62 regions, over 90% of these patients had cortical thicknesses within the normal range. Patients with AD had more outlier regions than people with MCI (n = 662) or controls (n = 159) (F(2, 1,022) = 95.39, p = 2.0 × 10-16). They were also more dissimilar to each other than people with MCI or controls (F(2, 1,024) = 209.42, p = 2.2 × 10-16). A greater number of outlier regions were associated with worse cognitive function, CSF protein concentrations, and an increased risk of converting from MCI to AD within 3 years (hazard ratio 1.028, 95% CI 1.016-1.039, p = 1.8 × 10-16). Individualized normative maps of cortical thickness highlight the heterogeneous effect of AD on the brain. Regional outlier estimates have the potential to be a marker of disease and could be used to track an individual's disease progression or treatment response in clinical trials.

PO-04-184 ARTIFICIAL INTELLIGENCE-ESTIMATED BIOLOGICAL HEART AGE USING 12 LEAD ELECTROCARDIOGRAM PREDICTS MORTALITY AND CARDIOVASCULAR OUTCOMES

There is a paucity of data on the artificial intelligence-estimated biological ECG heart age (AI ECG-heart age) for predicting cardiovascular outcomes as distinct from the chronological age (CA). We sought to investigate whether a deep learning-based algorithm to estimate the AI ECG-heart age using standard 12-lead ECGs predicted mortality and cardiovascular outcomes. We trained and validated a deep neural network using raw ECG digital data from 425,051 12-lead ECGs acquired between January 2006 and December 2021. The network performed a holdout test using a separate set of 97,058 ECGs. Additionally, randomly age-sex matched patients with reduced and preserved ejection fraction (EF) were compared. The deep neural network was trained to estimate the AI ECG-heart age (the mean absolute error 5.8 ± 3.9 years and R squared of 0.7 (r=0.84, p<0.0001) (Fig 1A). In the Cox proportional-hazards models after adjusting for relevant co-morbidity factors, the subjects with an AI ECG-heart age of seven years older than the chronological age had higher all-cause mortality (HR 1.62 [1.43-1.84]) and major adverse cardiovascular events (MACEs) (HR 1.92 [1.65-2.23]), while those under seven years had an inverse relationship (HR 0.86 [0.77-0.95] for all-cause mortality; HR 0.73 [0.63-0.84] for MACEs). Subjects with a reduced EF had a substantially higher mean AI ECG heart-age, QRS duration, and corrected QT intervals than those with a preserved EF (all p<0.001) (Fig 1B). The biological heart age estimated by AI had a significant impact on mortality and MACE. Those data suggested that the AI ECG-heart age might facilitate the primary prevention and health care for cardiovascular outcomes.

Fully Aerosol Jet-Printed Passive Components for 140–220-GHz Operation

This letter focuses on the unique capabilities that aerosol jet printing (AJP) provides for manufacturing electronics at near-THz frequencies. For the first time, we demonstrate fully aerosol jet-printed passive components above the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$W$</tex-math> </inline-formula> -band. Dielectric substrates and conductors are printed on metal carriers via AJP. These components include microstrip (MS) transmission lines (TLs) and radial stubs, and utilize vialess grounded coplanar waveguide (GCPW)-to-MS TL launches. Two separate sets of components and calibration structures are realized, for two different dielectric thicknesses. For the TLs, measured from 140 to 220 GHz, the average loss varies from 0.86 to 2.6 dB/mm. The radial stub return loss, measured from 200 to 220 GHz, is closely aligned with simulations.

Comparing oxidation of aluminum by oxygen and ozone using reactive force field molecular dynamics simulations

Ozone has attracted more scholars’ attention due to its less environmental injury during oxidation. Oxidation with ozone has been more regarded because ozone evaluates oxygen and can form a terrific oxide layer even at lower temperatures compared to oxygen. This work investigated the Al(100) surface oxidation simulations by O2 and O3 molecules at 400, 600, and 800 K temperatures with the reactive force field (ReaxFF) method. In separate simulation boxes, one hundred ozone molecules and 150 oxygen molecules have been placed in a space above the Al(100) surfaces to examine this metal’s behavior with them (gases incipient density of 0.47 g/cm3). We found further growth of the oxide layer in the case of ozone-aluminum. Subsequently, we surveyed the correlation between temperatures and the development of alumina layers in the case of oxygen and ozone in separate sets through ReaxFF simulation to untangle the mechanism of oxidation kinetics of the Al(100) surface by ozone. Even though there are demonstrated constraints on growing an oxide layer with ozone and oxygen, it is possible to produce a thicker oxide layer at lower temperatures using ozone. Here, we used LAMMPS as our program package for all simulations and used reactive molecular dynamics simulation, which is a relatively inexpensive tool for studying complex conditions that are impossible to reproduce with other techniques.

Towards Automated Ethogramming: Cognitively-Inspired Event Segmentation for Streaming Wildlife Video Monitoring

Advances in visual perceptual tasks have been mainly driven by the amount, and types, of annotations of large-scale datasets. Researchers have focused on fully-supervised settings to train models using offline epoch-based schemes. Despite the evident advancements, limitations and cost of manually annotated datasets have hindered further development for event perceptual tasks, such as detection and localization of objects and events in videos. The problem is more apparent in zoological applications due to the scarcity of annotations and length of videos-most videos are at most ten minutes long. Inspired by cognitive theories, we present a self-supervised perceptual prediction framework to tackle the problem of temporal event segmentation by building a stable representation of event-related objects. The approach is simple but effective. We rely on LSTM predictions of high-level features computed by a standard deep learning backbone. For spatial segmentation, the stable representation of the object is used by an attention mechanism to filter the input features before the prediction step. The self-learned attention maps effectively localize the object as a side effect of perceptual prediction. We demonstrate our approach on long videos from continuous wildlife video monitoring, spanning multiple days at 25 FPS. We aim to facilitate automated ethogramming by detecting and localizing events without the need for labels. Our approach is trained in an online manner on streaming input and requires only a single pass through the video, with no separate training set. Given the lack of long and realistic (includes real-world challenges) datasets, we introduce a new wildlife video dataset–nest monitoring of the Kagu (a flightless bird from New Caledonia)–to benchmark our approach. Our dataset features a video from 10 days (over 23 million frames) of continuous monitoring of the Kagu in its natural habitat. We annotate every frame with bounding boxes and event labels. Additionally, each frame is annotated with time-of-day and illumination conditions. We will make the dataset, which is the first of its kind, and the code available to the research community. We find that the approach significantly outperforms other self-supervised, traditional (e.g., Optical Flow, Background Subtraction) and NN-based (e.g., PA-DPC, DINO, iBOT), baselines and performs on par with supervised boundary detection approaches (i.e., PC). At a recall rate of 80%, our best performing model detects one false positive activity every 50 min of training. On average, we at least double the performance of self-supervised approaches for spatial segmentation. Additionally, we show that our approach is robust to various environmental conditions (e.g., moving shadows). We also benchmark the framework on other datasets (i.e., Kinetics-GEBD, TAPOS) from different domains to demonstrate its generalizability. The data and code are available on our project page: https://aix.eng.usf.edu/research_automated_ethogramming.html

A deep learning-based technique for the diagnosis of epidural spinal cord compression on thoracolumbar CT.

To develop a deep learning (DL) model for epidural spinal cord compression (ESCC) on CT, which will aid earlier ESCC diagnosis for less experienced clinicians. We retrospectively collected CT and MRI data from adult patients with suspected ESCC at a tertiary referral institute from 2007 till 2020. A total of 183 patients were used for training/validation of the DL model. A separate test set of 40 patients was used for DL model evaluation and comprised 60 staging CT and matched MRI scans performed with an interval of up to 2months. DL model performance was compared to eight readers: one musculoskeletal radiologist, two body radiologists, one spine surgeon, and four trainee spine surgeons. Diagnostic performance was evaluated using inter-rater agreement, sensitivity, specificity and AUC. Overall, 3115 axial CT slices were assessed. The DL model showed high kappa of 0.872 for normal, low and high-grade ESCC (trichotomous), which was superior compared to a body radiologist (R4, κ = 0.667) and all four trainee spine surgeons (κ range = 0.625-0.838)(all p < 0.001). In addition, for dichotomous normal versus any grade of ESCC detection, the DL model showed high kappa (κ = 0.879), sensitivity (91.82), specificity (92.01) and AUC (0.919), with the latter AUC superior to all readers (AUC range = 0.732-0.859, all p < 0.001). A deep learning model for the objective assessment of ESCC on CT had comparable or superior performance to radiologists and spine surgeons.Earlier diagnosis of ESCC on CT could reduce treatment delays, which areassociated with poor outcomes, increased costs, and reduced survival.

Multistep Automated Data Labelling Procedure (MADLaP) for thyroid nodules on ultrasound: An artificial intelligence approach for automating image annotation

Machine learning (ML) for diagnosis of thyroid nodules on ultrasound is an active area of research. However, ML tools require large, well-labeled datasets, the curation of which is time-consuming and labor-intensive. The purpose of our study was to develop and test a deep-learning-based tool to facilitate and automate the data annotation process for thyroid nodules; we named our tool Multistep Automated Data Labelling Procedure (MADLaP). MADLaP was designed to take multiple inputs including pathology reports, ultrasound images, and radiology reports. Using multiple step-wise ‘modules’ including rule-based natural language processing, deep-learning-based imaging segmentation, and optical character recognition, MADLaP automatically identified images of a specific thyroid nodule and correctly assigned a pathology label. The model was developed using a training set of 378 patients across our health system and tested on a separate set of 93 patients. Ground truths for both sets were selected by an experienced radiologist. Performance metrics including yield (how many labeled images the model produced) and accuracy (percentage correct) were measured using the test set. MADLaP achieved a yield of 63 % and an accuracy of 83 %. The yield progressively increased as the input data moved through each module, while accuracy peaked part way through. Error analysis showed that inputs from certain examination sites had lower accuracy (40 %) than the other sites (90 %, 100 %). MADLaP successfully created curated datasets of labeled ultrasound images of thyroid nodules. While accurate, the relatively suboptimal yield of MADLaP exposed some challenges when trying to automatically label radiology images from heterogeneous sources. The complex task of image curation and annotation could be automated, allowing for enrichment of larger datasets for use in machine learning development.

Effectiveness of MOOCs on learning efficiency of students: a perception study

PurposeThe study is aimed at analyzing the perceptions of students and teachers regarding the effectiveness of massive open online courses (MOOCs) on learning efficiency of students and also evaluating MOOCs as an ideal tool for designing a blended model for education.Design/methodology/approachThe analysis was carried out by using the data gathered from the students as well as teachers of University of Mysore, Karnataka, India. Two separate sets of questionnaires were developed for both the categories of respondents. Also, the respondents were required to have prior experience in MOOCs. Further, the collected data was analyzed using statistical package for social sciences (SPSS).FindingsThe study showed that MOOCs have a more positive influence on learning efficiency, as opined by both teachers and students. Negative views such as cheating during the assessment, lack of individual attention to students and low teacher-student ratio were also observed.Practical implicationsMany educational institutions view that the MOOCs do not influence learning efficiency and also do not support in achieving their vision. However, this study provides evidence that MOOCs are positively influencing the learning efficiency and also can be employed in a blended model of education so as to promote collaborative learning.Originality/valueTechnology is playing a pivotal role in all fields of life and the education sector is not an exception. It can be rightly said that the technology-based education models such as MOOCs are the need of the hour. This study may help higher education institutions to adopt MOOCs as part of their blended model of education, and, if already adopted, the outcome of the present study will help them to improve the effectiveness of the MOOCs they are offering.

Filler in the Tear Trough and Lid-Cheek Junction: A Cadaveric Evaluation of Filler Type and Manipulation Technique

Abstract Goals/Purpose The tear trough and lid-cheek junction demonstrate among the earliest and most noticeable signs of aging as a result of thin periorbital skin and loss of volume causing shadowing and darkness of the infraorbital region. Thus, rejuvenation of this area is commonly requested by patients. Soft tissue filler injections are the second most common non-surgical cosmetic procedure performed in the U.S, with over 2.6 million procedures reported in the 2020 Plastic Surgery Statistics Report. As such, there is a high demand for filler injection in the infraorbital area. However, there is currently only one FDA-approved filler for the tear trough, Juvéderm Volbella XC, although many practitioners also use Restylane-L. Interestingly, these two products differ significantly in their rheology (G’, cohesivity, and flexibility), and the Juvéderm Vycross is smooth gel while the Restylane NASHA is particulate. Despite the high demand of filler in the infraorbital area, there remains debate on injection practices and precise anatomical placement leading to several prior cadaveric examinations of the region. We herein aimed to contribute to the anatomic understanding of infraorbital filler injection via a cadaveric study using Juvéderm and Restylane products with different constitutions and manipulation techniques. Methods/Technique Institutional approval for cadaveric study was obtained. Six cadaveric cephalus specimens were used for filler injection and dissection. All fillers were dyed with food coloring. The Juvéderm products Volbella XC (tear trough, red) and Voluma XC (malar, blue) were used together in hemi-faces (Juvéderm pair), and Restylane products Restylane-L (tear trough, purple) and Restylane-LYFT (malar, green) were used together in separate sets of hemi-faces (Restylane pair). All filler injections were performed by the senior author as follows: 0.5 cc of filler was injected into the malar region in the deep fat compartment above the maxilla both medial and lateral to the infraorbital foramen. In the tear trough region, the orbit was protected with the nondominant hand and 0.3 cc of filler injected just superficial to the infraorbital rim. One hemi-face was injected with the original constitution of filler product and the other was injected with filler blended with 0.2 cc of saline. In two cadavers, after injection the hemifaces were massaged and compared to those which remained unmanipulated. All hemi-faces were then dissected from medial to lateral in the subcutaneous plane and between the superficial and deep facial fat compartments by a single author. All photographs and videography were taken by a single author. All hemi-faces were subjectively analyzed by all authors. Results/Complications All cadavers were in excellent condition, and the subcutaneous tissue, superficial fat, and deep fat compartments were dissected in clear planes. Volbella was found on the periosteum of the infraorbital rim and along the orbicularis retaining ligament (ORL) and crossed into the sub-orbicularis oculi fat (SOOF) and the subcutaneous fat while the Restylane-L did not leech into the fat and remained as a distinct mass on the rim and ORL. Voluma was found in the deep medial cheek compartment and crossed into the superficial medial cheek and superficial nasolabial compartments and subcutaneous fat. In contrast, Restylane-LYFT remained mostly within the deep medial cheek compartment without spreading into the superficial fat pads (Figure). Dilutional manipulation by blending with saline and physical manipulation by massaging both increased spread of all products within the anatomical planes. However, Restylane products appeared to remain distinctly within the anatomical planes they were injected in despite the manipulations. Conclusion This cadaveric study was successful in demonstrating the behavior of different hyaluronic acid fillers injected in the infraorbital region. Tear trough and malar region filler injections just above the periosteum resulted in precise anatomical placement along the infraorbital rim/ORL and in the deep medial cheek fat compartment, respectively. The Juvéderm products appeared to spread across anatomical planes more freely than Restylane products. This may be explained by the lower elastic modulus (G’) of the Juvéderm fillers. Manipulation of the filler product including blending and post-injection massage appeared to spread the fillers medially and laterally within the injected anatomical planes compared to unmanipulated filler. Of note, there are limitations to this study including the cadaver model as this does not allow for hydrophilic swelling of the filler nor spread and integration of the products as in live patients. The results of this study provide information about filler products commonly used in the infraorbital region and the effects of manipulation using blending or massage. Confirmation of precise anatomical placement of the fillers and improved knowledge of their behavior in situ through this cadaver study may assist clinical practitioners’ selection of rheological properties of fillers, method of constitution, and manipulation technique depending on patient-specific anatomy.