Tag: pubmed

IEEE Trans Vis Comput Graph. 2022 Oct 24;PP. doi: 10.1109/TVCG.2022.3216712. Online ahead of print.

ABSTRACT

Image-warping, a per-pixel deformation of one image into another, is an essential component in immersive visual experiences such as virtual reality or augmented reality. The primary issue with image warping is disocclusions, where occluded (and hence unknown) parts of the input image would be required to compose the output image. We introduce a new image warping method, Metameric image inpainting – an approach for hole-filling in real-time with foundations in human visual perception. Our method estimates image feature statistics of disoccluded regions from their neighbours. These statistics are inpainted and used to synthesise visuals in real-time that are less noticeable to study participants, particularly in peripheral vision. Our method offers speed improvements over the standard structured image inpainting methods while improving realism over colour-based inpainting such as push-pull. Hence, our work paves the way towards future applications such as depth image-based rendering, 6-DoF 360 rendering, and remote render-streaming.

PMID:36279345 | DOI:10.1109/TVCG.2022.3216712

IEEE Trans Neural Netw Learn Syst. 2022 Oct 24;PP. doi: 10.1109/TNNLS.2022.3213522. Online ahead of print.

ABSTRACT

Real-world data usually present long-tailed distributions. Training on imbalanced data tends to render neural networks perform well on head classes while much worse on tail classes. The severe sparseness of training instances for the tail classes is the main challenge, which results in biased distribution estimation during training. Plenty of efforts have been devoted to ameliorating the challenge, including data resampling and synthesizing new training instances for tail classes. However, no prior research has exploited the transferable knowledge from head classes to tail classes for calibrating the distribution of tail classes. In this article, we suppose that tail classes can be enriched by similar head classes and propose a novel distribution calibration (DC) approach named as label-aware DC (). transfers the statistics from relevant head classes to infer the distribution of tail classes. Sampling from calibrated distribution further facilitates rebalancing the classifier. Experiments on both image and text long-tailed datasets demonstrate that significantly outperforms existing methods. The visualization also shows that provides a more accurate distribution estimation.

PMID:36279339 | DOI:10.1109/TNNLS.2022.3213522

IEEE Trans Neural Netw Learn Syst. 2022 Oct 24;PP. doi: 10.1109/TNNLS.2022.3213518. Online ahead of print.

ABSTRACT

The finite inverted beta mixture model (IBMM) has been proven to be efficient in modeling positive vectors. Under the traditional variational inference framework, the critical challenge in Bayesian estimation of the IBMM is that the computational cost of performing inference with large datasets is prohibitively expensive, which often limits the use of Bayesian approaches to small datasets. An efficient alternative provided by the recently proposed stochastic variational inference (SVI) framework allows for efficient inference on large datasets. Nevertheless, when using the SVI framework to address the non-Gaussian statistical models, the evidence lower bound (ELBO) cannot be explicitly calculated due to the intractable moment computation. Therefore, the algorithm under the SVI framework cannot directly use stochastic optimization to optimize the ELBO, and an analytically tractable solution cannot be derived. To address this problem, we propose an extended version of the SVI framework with more flexibility, namely, the extended SVI (ESVI) framework. This framework can be used in many non-Gaussian statistical models. First, some approximation strategies are applied to further lower the ELBO to avoid intractable moment calculations. Then, stochastic optimization with noisy natural gradients is used to optimize the lower bound. The excellent performance and effectiveness of the proposed method are verified in real data evaluation.

PMID:36279334 | DOI:10.1109/TNNLS.2022.3213518

IEEE Trans Neural Netw Learn Syst. 2022 Oct 24;PP. doi: 10.1109/TNNLS.2022.3213589. Online ahead of print.

ABSTRACT

Multimodal data provide complementary information of a natural phenomenon by integrating data from various domains with very different statistical properties. Capturing the intramodality and cross-modality information of multimodal data is the essential capability of multimodal learning methods. The geometry-aware data analysis approaches provide these capabilities by implicitly representing data in various modalities based on their geometric underlying structures. Also, in many applications, data are explicitly defined on an intrinsic geometric structure. Generalizing deep learning methods to the non-Euclidean domains is an emerging research field, which has recently been investigated in many studies. Most of those popular methods are developed for unimodal data. In this article, a multimodal graph wavelet convolutional network (M-GWCN) is proposed as an end-to-end network. M-GWCN simultaneously finds intramodality representation by applying the multiscale graph wavelet transform to provide helpful localization properties in the graph domain of each modality and cross-modality representation by learning permutations that encode correlations among various modalities. M-GWCN is not limited to either the homogeneous modalities with the same number of data or any prior knowledge indicating correspondences between modalities. Several semisupervised node classification experiments have been conducted on three popular unimodal explicit graph-based datasets and five multimodal implicit ones. The experimental results indicate the superiority and effectiveness of the proposed methods compared with both spectral graph domain convolutional neural networks and state-of-the-art multimodal methods.

PMID:36279325 | DOI:10.1109/TNNLS.2022.3213589

Oper Dent. 2022 Oct 21. doi: 10.2341/21-148-L. Online ahead of print.

ABSTRACT

STATEMENT OF PROBLEM: Advertisements of glass-ionomer-containing restorative materials recommend suitability as load-bearing permanent or semi-permanent restorations. Historically, unacceptably high wear rates limit clinical indications of glass-ionomer-containing restorations in this regard.

OBJECTIVE: To compare the in vitro wear of contemporary glass-ionomer-containing dental materials commercially advertised for use in permanent dentition as load-bearing restorations in a chewing simulator. Resin composite was tested as a control.

METHODS AND MATERIALS: A resin-modified glass ionomer (Ionolux, VOCO gmbH), a high viscosity glass-ionomer hybrid system (Equia Forte HT with Equia Coat, GC America), and a bioactive ionic resin with reactive glass filler (Activa Bioactive Restorative, Pulpdent) were evaluated. Filtek Supreme Ultra (3M ESPE) is a visible light-activated resin composite that served as a control. Standardized flat disk-shaped specimens (n=12/group) were submitted to 500,000 cycles with continuous thermal cycling against steatite antagonists. Volumetric wear was measured at 1000, 10,000, 200,000, and 500,000 cycles.

RESULTS: There was a statistically significant difference in mean volumetric wear for Activa Bioactive Restorative (p=0.0081, 95% CI: 0.3973, 0.4982) and Equia Forte HT (p<0.001, 95% CI: 1.2495, 1.8493), but no statistically significant difference in mean volumetric wear for Ionolux (p=0.6653) compared to control. Activa Bioactive Restorative wore approximately 60% less than, and Equia Forte HT twice more than Filtek Supreme Ultra on average, respectively.

CONCLUSIONS: Compared to a resin composite, contemporary glass-ionomer-containing restorative materials advertised for use as load-bearing restorations display measurably variable in vitro wear rates.

PMID:36279318 | DOI:10.2341/21-148-L

Dermatol Ther. 2022 Oct 24:e15958. doi: 10.1111/dth.15958. Online ahead of print.

ABSTRACT

BACKGROUND: Secukinumab demonstrated high efficacy and favorable safety profile in patients with moderate-to-severe plaque psoriasis (PsO) in clinical trials. However, understanding of patient characteristics and clinical outcomes in real world in Thailand is still limited.

AIMS: To describe patient characteristics, effectiveness and safety of secukinumab in Thai PsO patients.

METHODS: This retrospective study analyzed data from medical records of adult PsO patients who initiated secukinumab at 7 dermatology centers from September 2017 to April 2021. Study outcomes included patient characteristics and changes in Psoriasis Area and Severity Index (PASI) score from baseline at weeks 4 and 16 after secukinumab initiation. Adverse events were recorded. Subgroup analyses by adherence rate and completeness of loading dose were performed.

RESULTS: Of 163 patients, the mean (SD) age was 44.0 (14.0) years. Most patients (84.7%) were previously treated with topical therapy while 62.0% and 21.5% of patients had received systemic and biologic therapy, respectively. The mean baseline PASI score was 15.4 (9.3). Overall, the mean PASI score improved by 58.0% at week 4 and 78.4% at week 16. Statistically significant differences in PASI approvement were revealed among subgroups of patients with different loading dose and adherence rate. Adverse effects were reported in 8.0% of patients.

CONCLUSION: The characteristics of patients in this study were slightly different from clinical trials in terms of demographic and clinical characteristics, as well as PsO treatment. Secukinumab was effective and safe in Thai patients with PsO, especially among those with complete loading dose and a higher adherence rate. This article is protected by copyright. All rights reserved.

PMID:36279306 | DOI:10.1111/dth.15958

PLoS One. 2022 Oct 24;17(10):e0276514. doi: 10.1371/journal.pone.0276514. eCollection 2022.

ABSTRACT

Ranked set sampling (RSS) has created a broad interest among researchers and it is still a unique research topic. It has at long last begun to find its way into practical applications beyond its initial horticultural based birth in the fundamental paper by McIntyre in the nineteenth century. One of the extensions of RSS is median ranked set sampling (MRSS). MRSS is a sampling procedure normally utilized when measuring the variable of interest is troublesome or expensive, whereas it might be easy to rank the units using an inexpensive sorting criterion. Several researchers introduced ratio, regression, exponential, and difference type estimators for mean estimation under the MRSS design. In this paper, we propose three new mean estimators under the MRSS scheme. Our idea is based on three-fold utilization of supplementary information. Specifically, we utilize the ranks and second raw moments of the supplementary information and the original values of the supplementary variable. The appropriateness of the proposed group of estimators is demonstrated in light of both real and artificial data sets based on the Monte-Carlo simulation. Additionally, the performance comparison is also conducted regarding the reviewed families of estimators. The results are empowered and the predominant execution of the proposed group of estimators is seen throughout the paper.

PMID:36279286 | DOI:10.1371/journal.pone.0276514

PLoS One. 2022 Oct 24;17(10):e0276609. doi: 10.1371/journal.pone.0276609. eCollection 2022.

ABSTRACT

Drug-target protein interaction (DTI) identification is fundamental for drug discovery and drug repositioning, because therapeutic drugs act on disease-causing proteins. However, the DTI identification process often requires expensive and time-consuming tasks, including biological experiments involving large numbers of candidate compounds. Thus, a variety of computation approaches have been developed. Of the many approaches available, chemo-genomics feature-based methods have attracted considerable attention. These methods compute the feature descriptors of drugs and proteins as the input data to train machine and deep learning models to enable accurate prediction of unknown DTIs. In addition, attention-based learning methods have been proposed to identify and interpret DTI mechanisms. However, improvements are needed for enhancing prediction performance and DTI mechanism elucidation. To address these problems, we developed an attention-based method designated the interpretable cross-attention network (ICAN), which predicts DTIs using the Simplified Molecular Input Line Entry System of drugs and amino acid sequences of target proteins. We optimized the attention mechanism architecture by exploring the cross-attention or self-attention, attention layer depth, and selection of the context matrixes from the attention mechanism. We found that a plain attention mechanism that decodes drug-related protein context features without any protein-related drug context features effectively achieved high performance. The ICAN outperformed state-of-the-art methods in several metrics on the DAVIS dataset and first revealed with statistical significance that some weighted sites in the cross-attention weight matrix represent experimental binding sites, thus demonstrating the high interpretability of the results. The program is freely available at https://github.com/kuratahiroyuki/ICAN.

PMID:36279284 | DOI:10.1371/journal.pone.0276609

JMIR Mhealth Uhealth. 2022 Oct 24;10(10):e35722. doi: 10.2196/35722.

ABSTRACT

BACKGROUND: Sensors and digital devices have revolutionized the measurement, collection, and storage of behavioral and physiological data, leading to the new term digital biomarkers.

OBJECTIVE: This study aimed to investigate the scope of clinical evidence covered by systematic reviews (SRs) of randomized controlled trials involving digital biomarkers.

METHODS: This scoping review was organized using the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. With the search limited to English publications, full-text SRs of digital biomarkers included randomized controlled trials that involved a human population and reported changes in participants’ health status. PubMed and the Cochrane Library were searched with time frames limited to 2019 and 2020. The World Health Organization’s classification systems for diseases (International Classification of Diseases, Eleventh Revision), health interventions (International Classification of Health Interventions), and bodily functions (International Classification of Functioning, Disability, and Health [ICF]) were used to classify populations, interventions, and outcomes, respectively.

RESULTS: A total of 31 SRs met the inclusion criteria. The majority of SRs studied patients with circulatory system diseases (19/31, 61%) and respiratory system diseases (9/31, 29%). Most of the prevalent interventions focused on physical activity behavior (16/31, 52%) and conversion of cardiac rhythm (4/31, 13%). Looking after one’s health (physical activity; 15/31, 48%), walking (12/31, 39%), heart rhythm functions (8/31, 26%), and mortality (7/31, 23%) were the most commonly reported outcomes. In total, 16 physiological and behavioral data groups were identified using the ICF tool, such as looking after one’s health (physical activity; 14/31, 45%), walking (11/31, 36%), heart rhythm (7/31, 23%), and weight maintenance functions (7/31, 23%). Various digital devices were also studied to collect these data in the included reviews, such as smart glasses, smartwatches, smart bracelets, smart shoes, and smart socks for measuring heart functions, gait pattern functions, and temperature. A substantial number (24/31, 77%) of digital biomarkers were used as interventions. Moreover, wearables (22/31, 71%) were the most common types of digital devices. Position sensors (21/31, 68%) and heart rate sensors and pulse rate sensors (12/31, 39%) were the most prevalent types of sensors used to acquire behavioral and physiological data in the SRs.

CONCLUSIONS: In recent years, the clinical evidence concerning digital biomarkers has been systematically reviewed in a wide range of study populations, interventions, digital devices, and sensor technologies, with the dominance of physical activity and cardiac monitors. We used the World Health Organization’s ICF tool for classifying behavioral and physiological data, which seemed to be an applicable tool to categorize the broad scope of digital biomarkers identified in this review. To understand the clinical value of digital biomarkers, the strength and quality of the evidence on their health consequences need to be systematically evaluated.

PMID:36279171 | DOI:10.2196/35722

Clin Chem Lab Med. 2022 Oct 24. doi: 10.1515/cclm-2022-0642. Online ahead of print.

ABSTRACT

OBJECTIVES: Vitamin D-binding protein (VDBP), a serum transport protein for 25-hydroxyvitamin D [25(OH)D], has three common proteoforms which have co-localized amino acid variations and glycosylation. A monoclonal immunoassay was found to differentially detect VDBP proteoforms and methods using liquid chromatography-tandem mass spectrometry (LC-MS/MS) might be able to overcome this limitation. Previously developed multiple reaction monitoring LC-MS/MS methods for total VDBP quantification represent an opportunity to probe the potential effects of proteoforms on proteolysis, instrument response and quantification accuracy.

METHODS: VDBP was purified from homozygous human donors and quantified using proteolysis or acid hydrolysis and LC-MS/MS. An interlaboratory comparison was performed using pooled human plasma [Standard Reference Material^® 1950 (SRM 1950) Metabolites in Frozen Human Plasma] and analyses with different LC-MS/MS methods in two laboratories.

RESULTS: Several shared peptides from purified proteoforms were found to give reproducible concentrations [≤2.7% coefficient of variation (CV)] and linear instrument responses (R²≥0.9971) when added to human serum. Total VDBP concentrations from proteolysis or amino acid analysis (AAA) of purified proteoforms had ≤1.92% CV. SRM 1950, containing multiple proteoforms, quantified in two laboratories resulted in total VDBP concentrations with 7.05% CV.

CONCLUSIONS: VDBP proteoforms were not found to cause bias during quantification by LC-MS/MS, thus demonstrating that a family of proteins can be accurately quantified using shared peptides. A reference value was assigned for total VDBP in SRM 1950, which may be used to standardize methods and improve the accuracy of VDBP quantification in research and clinical samples.

PMID:36279170 | DOI:10.1515/cclm-2022-0642