Entire Dataset Research Articles

Variance-Reduced Stochastic Optimization for Efficient Inference of Hidden Markov Models

Hidden Markov models (HMMs) are popular models to identify a finite number of latent states from sequential data. However, fitting them to large datasets can be computationally demanding because most likelihood maximization techniques require iterating through the entire underlying dataset for every parameter update. We propose a novel optimization algorithm that updates the parameters of an HMM without iterating through the entire dataset. Namely, we combine a partial E step with variance-reduced stochastic optimization within the M step. We prove the algorithm converges under certain regularity conditions. We test our algorithm empirically using a simulation study as well as a case study of kinematic data collected using suction-cup attached biologgers from eight northern resident killer whales (Orcinus orca) off the western coast of Canada. In both, our algorithm converges in fewer epochs, with less computation time, and to regions of higher likelihood compared to standard numerical optimization techniques. Our algorithm allows practitioners to fit complicated HMMs to large time-series datasets more efficiently than existing baselines. Supplemental materials are available online.

Leveraging Descriptor Learning and Functional Map-based Shape Matching for Automatic Landmark Acquisition.

Geometric morphometrics is widely employed across the biological sciences for the quantification of morphological traits. However, the scalability of these methods to large datasets is hampered by the requisite placement of landmarks, which can be laborious and time consuming if done manually. Additionally, the selected landmarks embody a particular hypothesis regarding the critical geometry pertinent to the biological inquiry at hand. Modifying this hypothesis lacks flexibility, necessitating the acquisition of an entirely new set of landmarks on the entire dataset to reflect any theoretical adjustments. In our research, we investigate the precision and accuracy of landmarks derived from the comprehensive set of functional correspondences acquired through the functional map framework of geometry processing. We use a deep functional map network to learn shape descriptors that effectively yield functional map-based and point-to-point correspondences between the specimens in our dataset. We then interrogate these maps to identify corresponding landmarks given manually placed landmarks from the entire dataset. We assess our method by automating the landmarking process on a dataset comprising mandibles from various rodent species, comparing its efficacy against MALPACA, a cutting-edge technique for automatic landmark placement. Compared to MALPACA, our model is notably faster and maintains competitive accuracy. The Root Mean Square Error (RMSE) analysis reveals that while MALPACA generally exhibits the lowest RMSE, our models perform comparably, especially with smaller training datasets, suggesting strong generalizability. Visual evaluations confirm the precision of our landmark placements, with deviations remaining within an acceptable range. These findings underscore the potential of unsupervised learning models in anatomical landmark placement, providing a viable and efficient alternative to traditional methods.

Proposal of an Original Methodology to Evaluate the Performance of Chipper Machines

Wood fuel from the agroforestry sector is one of the main strategies cited by the EU for reducing energetic dependance on foreign markets. Its sustainability, both economic and environmental, can be improved through the optimization of harvesting and chipping operations. This study was focused on the dynamic and energetic balance of the chipping phase carried out by a chipper operated by the power-take-off (PTO) of a medium-power tractor. Both machines were equipped with sensors for real-time monitoring of fuel consumption, PTO torque and speed, trunk diameter and working time during the comminution of 61 poplar trees grown in a medium rotation coppice system. The data analysis was carried out on the entire dataset (about 29,000 records) without considering their belonging to different trees. By means of proper data ordinations, it has been possible to define all the intervals in which the chipping stopped (e.g., between two trees) and to exclude them from the intervals of actual chipping. This has allowed forcomputation of operative and actual working time, as well as of the basic power required to operate the chipper and the power for actual chipping. Subsequently, the parameter values observed during actual chipping were related to the cutting diameters measured at the same instant. Subsequently, the dataset was divided according to seven diameter classes, and, for each class, the descriptive statistical indices of working time, work productivity, CO2 emissions, energy requirement and fuel consumption were calculated. Eventually, the correlation between the variations in trunk diameter and other parameters was verified both on the whole dataset and based on the class average values. The analysis made it possible to identify the conditions of greatest efficiency for the chipper. More generally, the method could help to increase the accuracy of measurements aimed at characterizing the performance of chippers from the point of view of dynamic energy requirements as well as in relation to different wood species.

KADEL: Knowledge-Aware Denoising Learning for Commit Message Generation

Commit messages are natural language descriptions of code changes, which are important for software evolution such as code understanding and maintenance. However, previous methods are trained on the entire dataset without considering the fact that a portion of commit messages adhere to good practice (i.e., good-practice commits), while the rest do not. On the basis of our empirical study, we discover that training on good-practice commits significantly contributes to the commit message generation. Motivated by this finding, we propose a novel knowledge-aware denoising learning method called KADEL. Considering that good-practice commits constitute only a small proportion of the dataset, we align the remaining training samples with these good-practice commits. To achieve this, we propose a model that learns the commit knowledge by training on good-practice commits. This knowledge model enables supplementing more information for training samples that do not conform to good practice. However, since the supplementary information may contain noise or prediction errors, we propose a dynamic denoising training method. This method composes a distribution-aware confidence function and a dynamic distribution list, which enhances the effectiveness of the training process. Experimental results on the whole MCMD dataset demonstrate that our method overall achieves state-of-the-art performance compared with previous methods.

Optimizing motor imagery BCI models with hard trials removal and model refinement

Deep learning models have demonstrated remarkable performance in the classification of motor imagery BCI systems. However, these models exhibit sensitivity to challenging trials, often called hard trials, leading to performance degradation. In this paper, we address this issue by proposing two novel methods for identifying and mitigating the impact of hard trials on model performance. The first method leverages model prediction scores to discern hard trials. The second approach employs a quantitative explainable artificial intelligence (XAI) approach, enabling a more transparent and interpretable means of hard trial identification. The identified hard trials are removed from the entire motor imagery training and validation dataset, and the deep learning model is further re-trained using the dataset without hard trials. To evaluate the efficacy of these proposed methods, experiments were conducted on the Open BMI dataset. The results for hold-out analysis show that the proposed quantitative XAI- based hard trial removal method has statistically improved the average classification accuracy of the baseline deep CNN model from 63.77% to 68.70%, with p-value = 7.66e−11 for the subject-specific MI classification. Additionally, analyzing the scalp map representing the average relevance scores of correctly classified trials compared to a misclassified trial provides a deeper insight into identifying hard trials. The results indicate that the proposed quantitaive-based XAI approach outperformes the prediction-score-based approach in hard trial identification.

A novel score for predicting falls in community-dwelling older people: a derivation and validation study

BackgroundEarly detection of patients at risk of falling is crucial. This study was designed to develop and internally validate a novel risk score to classify patients at risk of falls.MethodsA total of 334 older people from a fall clinic in a medical center were selected. Least absolute shrinkage and selection operator (LASSO) regression was used to minimize the potential concatenation of variables measured from the same patient and the overfitting of variables. A logistic regression model for 1-year fall prediction was developed for the entire dataset using newly identified relevant variables. Model performance was evaluated using the bootstrap method, which included measures of overall predictive performance, discrimination, and calibration. To streamline the assessment process, a scoring system for predicting 1-year fall risk was created.ResultsWe developed a new model for predicting 1-year falls, which included the FRQ-Q1, FRQ-Q3, and single-leg standing time (left foot). After internal validation, the model showed good discrimination (C statistic, 0.803 [95% CI 0.749–0.857]) and overall accuracy (Brier score, 0.146). Compared to another model that used the total FRQ score instead, the new model showed better continuous net reclassification improvement (NRI) [0.468 (0.314–0.622), P < 0.01], categorical NRI [0.507 (0.291–0.724), P < 0.01; cutoff: 0.200–0.800], and integrated discrimination [0.205 (0.147–0.262), P < 0.01]. The variables in the new model were subsequently incorporated into a risk score. The discriminatory ability of the scoring system was similar (C statistic, 0.809; 95% CI, 0.756–0.861; optimism-corrected C statistic, 0.808) to that of the logistic regression model at internal bootstrap validation.ConclusionsThis study resulted in the development and internal verification of a scoring system to classify 334 patients at risk for falls. The newly developed score demonstrated greater accuracy in predicting falls in elderly people than did the Timed Up and Go test and the 30-Second Chair Sit-Stand test. Additionally, the scale demonstrated superior clinical validity for identifying fall risk.

Seismic modeling by combining the finite-difference scheme with the numerical dispersion suppression neural network

Seismic finite-difference (FD) modeling suffers from numerical dispersion including both the temporal and spatial dispersion, which can decrease the accuracy of the numerical modeling. To improve the accuracy and efficiency of the conventional numerical modeling, I develop a new seismic modeling method by combining the FD scheme with the numerical dispersion suppression neural network (NDSNN). This method involves the following steps. First, a training data set composed of a small number of wavefield snapshots is generated. The wavefield snapshots with the low-accuracy wavefield data and the high-accuracy wavefield data are paired, and the low-accuracy wavefield snapshots involve the obvious numerical dispersion including both the temporal and spatial dispersion. Second, the NDSNN is trained until the network converges to simultaneously suppress the temporal and spatial dispersion. Third, the entire set of low-accuracy wavefield data is computed quickly using FD modeling with the large time step and the coarse grid. Fourth, the NDSNN is applied to the entire set of low-accuracy wavefield data to suppress the numerical dispersion including the temporal and spatial dispersion. Numerical modeling examples verify the effectiveness of my proposed method in improving the computational accuracy and efficiency.

Proton polytropic behavior of periodic density structures in the solar wind

Context. In recent years, mesoscales have gained scientific interest because they have been determined to be important in a broad range of phenomena throughout heliophysics. The solar wind mesoscale structures include periodic density structures (PDSs), which are quasi-periodic increases in the density of the solar wind that range from a few minutes to a few hours. These structures have been extensively observed in remote-sensing observations of the solar corona and in in situ observations out to 1 AU, where they manifest as radial length scales greater than or equal to the size of the Earth’s dayside magnetosphere, that is, from tens to hundreds of Earth radii (RE). While the precise mechanisms that form PDSs are still debated, recent studies confirmed that most PDSs are of solar origin and do not form through dynamics during their propagation in the interplanetary space. Aims. We further investigate the origin of PDSs by exploring the thermodynamic signature of these structures. To do this, we estimate the values of the effective polytropic index (Y) and the entropy of protons, which in turn are compared with the corresponding values found for the solar wind. Methods. We used an extensive list of PDS events spanning more than two solar cycles of Wind measurements (the entire Wind dataset from 1995 to 2022) to investigate the thermodynamic signatures of PDSs. With the use of wavelet methods, we classified these PDSs as coherent or incoherent, based on the shared periodic behavior between proton density and alpha-to-proton ratio, and we derive the proton polytropic index. Results. Our results indicate that the coherent PDSs exhibit lower Y values (Ῡ≈1.54) on average and a higher entropy than the values in the entire Wind dataset (Ῡ≈1.79), but also exhibit similarities with the magnetic cloud of an interplanetary coronal mass ejection. In contrast, incoherent PDSs exhibit the same Y values as those of the entire Wind dataset.

Statistical Comparison of Southern and Northern Auroral Electrojet Indices as a Function of Solar Wind and IMF Conditions

AbstractA Southern Auroral Electrojet (SAE) index has been recently constructed using several Antarctica magnetometer stations. It has been compared for case studies with the standard Auroral Electrojet (AE) index, and a near‐conjugate to the southern stations Northern Auroral Electrojet (NAE) index. We compare the three indices statistically as a function of the accompanying solar wind (SW) and Interplanetary Magnetic Field (IMF) conditions to further explore conjugacy issues. We use 274 days of common north/south data presence between December 2005 and August 2010. We calculate the cross‐correlation coefficients and differences between all three pairs. We estimate the effect of the SW/IMF conditions on the index correlations and differences using three groups of data: (a) the entire data set, (b) two separate sets based on the presence or not of Southern Hemisphere stations within the 21‐03 Magnetic Local Time (MLT) sector where substorms occur, and (c) separately for the four different seasons. We find that high north‐south correlation coefficients are more common during strong SW/IMF driving, while the index differences are also higher, suggesting that the SAE index follows better the northern indices' trend, but has even lower values during active times. The UT study shows that the number of high AE/SAE correlations is slightly lower at all clock angles and dynamic pressure levels for the periods within 1454–1941 UT (when no southern station is within 21–03 MLT). Finally, the results show that the number of high correlations is greater during the northern spring than the winter period.

Advancements in breast cancer progression monitoring: Analysis of methylated circulating cell- free DNA.

e12548 Background: Breast cancer is a leading cause of mortality among women, with a significant number developing metastasis. Existing markers like CA-153 and CEA have limited sensitivity. Circulating cell-free DNA (ccfDNA) in plasma can be used for the noninvasive sampling of cancer cells obtained from patients with breast cancer. Aberrant circulating methylated GCM2 and TMEM240 have been reported in patients with breast cancer. The study aims to evaluate the potential of simultaneous detection of methylated GCM2 and TMEM240 in plasma for enhanced prediction of residual disease and tumor progression. Methods: In a prospective study conducted in Taipei, Taiwan, we enrolled a total of 396 patients between 2016 and 2023. The retrospective study involving plasma samples from Western populations included a total of 134 samples. Utilizing the Monte Carlo replicates module within the R software, is employed to automate the random partitioning of 70% of the entire patient dataset into a training set, while the remaining 30% is assigned to the validation set. The model's performance, including sensitivity, specificity, accuracy, false positives, and false negatives, will be thoroughly assessed to ensure its robustness. Results: The test demonstrated an accuracy of 91.5%, a sensitivity of 95.7%, and a specificity of 90.3% for prediction of residual disease and tumor progression. CA15-3 and CEA markers show limited sensitivity (53.3% and 33.3%, respectively). The tests also assess the tumor burden dynamically through monitoring circulating methylated GCM2 and TMEM240 in plasma of patients with breast cancer following their treatment. Conclusions: The simultaneous detection of circulating methylated GCM2 and TMEM240 in plasma presents a novel and accurate noninvasive tool for predicting tumor progression. Clinical trial information: N201708049, N202105091, N202202061.

Integrative analyses of multi-omics data constructing tumor microenvironment and immune-related molecular prognosis model in human colorectal cancer

The increasing prevalence and incidence of colorectal cancer (CRC), particularly in young adults, underscore the imperative to comprehend its fundamental mechanisms, discover novel diagnostic and prognostic markers, and enhance therapeutic strategies. Here, we integrated multi-omics data, including gene expression, somatic mutation data and DNA methylation data, to unravel the intricacies of tumor microenvironment (TME) in CRC and search for novel prognostic markers. By calculating the immune score for each patient from the expression profile, we delineated the differential immune cell fraction, constructed an immune-related multi-omics atlas, and identified molecular characteristics. The entire colorectal dataset (n = 343) was randomly divided into training (n = 249) and testing datasets (n = 94). We screened 144 immune-related genes, 6 mutant genes, and 38 methylation probes associated with overall survival (OS). These makers were then incorporated into a 10-gene prognostic model using Lasso and Cox regression in the training dataset, and the model's performance was evaluated in an independent validation dataset. The model exhibited satisfactory results (average concordance index [C-index] = 0.77), with the average 1-year, 3-year, and 5-year AUCs being 0.79, 0.76, and 0.76 in the training dataset and 0.74, 0.80, and 0.90 in the testing dataset. Furthermore, the prognostic model demonstrated applicability in guiding chemotherapy for CRC patients and exhibited a degree of pan-cancer utility in risk stratification. In conclusion, our integrated analysis of multi-omics data revealed immune-related genetic and epigenetic characteristics of the TME. We propose an integrative prognostic model that can stratify risk and guide chemotherapy for CRC patients. The generalizability of the model in risk stratification across different cancer types was validated in Pan-Cancer cohort.

Comparative Evaluation of Animated Scatter Plot Transitions.

Scatter plots are popular for displaying 2D data, but in practice, many data sets have more than two dimensions. For the analysis of such multivariate data, it is often necessary to switch between scatter plots of different dimension pairs, e.g., in a scatter plot matrix (SPLOM). Alternative approaches include a "grand tour" for an overview of the entire data set or creating artificial axes from dimensionality reduction (DR). A cross-cutting concern in all techniques is the ability of viewers to find correspondence between data points in different views. Previous work proposed animations to preserve the mental map between view changes and to trace points as well as clusters between scatter plots of the same underlying data set. In this article, we evaluate a variety of spline- and rotation-based view transitions in a crowdsourced user study focusing on ecological validity. Using the study results, we assess each animation's suitability for tracing points and clusters across view changes. We evaluate whether the order of horizontal and vertical rotation is relevant for task accuracy. The results show that rotations with an orthographic camera or staged expansion of a depth axis significantly outperform all other animation techniques for the traceability of individual points. Further, we provide a ranking of the animated transition techniques for traceability of individual points. However, we could not find any significant differences for the traceability of clusters. Furthermore, we identified differences by animation direction that could guide further studies to determine potential confounds for these differences. We publish the study data for reuse and provide the animation framework as a D3.js plug-in.

AbMelt: Learning antibody thermostability from molecular dynamics

Antibody thermostability is challenging to predict from sequence and/or structure. This difficulty is likely due to the absence of direct entropic information. Herein, we present AbMelt where we model the inherent flexibility of homologous antibody structures using molecular dynamics simulations at three temperatures and learn the relevant descriptors to predict the temperatures of aggregation (Tagg), melt onset (Tm,on), and melt (Tm). We observed that the radius of gyration deviation of the complementarity determining regions at 400 K is the highest Pearson correlated descriptor with aggregation temperature (rp = −0.68 ± 0.23) and the deviation of internal molecular contacts at 350 K is the highest correlated descriptor with both Tm,on (rp = −0.74 ± 0.04) as well as Tm (rp = −0.69 ± 0.03). Moreover, after descriptor selection and machine learning regression, we predict on a held-out test set containing both internal and public data and achieve robust performance for all endpoints compared with baseline models (Tagg R2 = 0.57 ± 0.11, Tm,on R2 = 0.56 ± 0.01, and Tm R2 = 0.60 ± 0.06). In addition, the robustness of the AbMelt molecular dynamics methodology is demonstrated by only training on <5% of the data and outperforming more traditional machine learning models trained on the entire data set of more than 500 internal antibodies. Users can predict thermostability measurements for antibody variable fragments by collecting descriptors and using AbMelt, which has been made available.

An IoMT-based Federated Learning Survey in Smart Transportation

Abstract: Internet of Medical Things (IoMT) is a technology that encompasses medical devices, wearable sensors, and applications connected to the Internet. In road accidents, it plays a crucial role in enhancing emergency response and reducing the impact of accidents on victims. Smart Transportation uses this technology to improve the efficiency and safety of transportation systems. The current Artificial Intelligence applications lack transparency and interpretability which is of utmost importance in critical transportation scenarios, such as autonomous vehicles, air traffic control systems, and traffic management systems. Explainable Artificial Intelligence (XAI) provides a clear, transparent explanation and actions. Traditional Machine Learning techniques have enabled Intelligent Transportation systems by performing centralized vehicular data training at the server where data sharing is needed, thus introducing privacy issues. To reduce transmission overhead and achieve privacy, a collaborative and distributed machine learning approach called Federated Learning (FL) is used. Here only model updates are transmitted instead of the entire dataset. This paper provides a comprehensive survey on the prediction of traffic using Machine Learning, Deep Learning, and FL. Among these, FL can predict traffic accurately without compromising privacy. We first present the overview of XAI and FL in the introduction. Then, we discuss the basic concepts of FL and its related work, the FL-IoMT framework, and motivations for using FL in transportation. Subsequently, we discuss the applications of using FL in transportation and open-source projects. Finally, we highlight several research challenges and their possible directions in FL

Investigating the Physical and Operational Characteristics of Manufacturing Processes for MFI-Type Zeolite Membranes for Ethanol/Water Separation via Principal Component Analysis

In this study, Principal Component Analysis (PCA) was applied to discern the underlying trends for 31 distinct MFI (Mobil No. 5)-zeolite membranes of 11 textural, chemical, and operational factors related to manufacturing processes. Initially, a comprehensive PCA approach was employed for the entire dataset, revealing a moderate influence of the first two principal components (PCs), which collectively accounted for around 38% of the variance. Membrane samples exhibited close proximity, which prevented the formation of any clusters. To address this limitation, a subset acquisition strategy was followed, based on the findings of the PCA for the entire dataset. This resulted in an enhanced overall contribution and the revelation of diverse patterns among the membranes and the considered manufacturing factors (total variance between 55% and 77%). The segmentation of the data unveiled a robust correlation between silica (SiO2) concentration and pervaporation conditions. Additionally, a notable clustering of the chemical compositions of the preparation solutions underscored their significant influence on the operational efficacy of MFI zeolite membranes. On the other hand, an exclusive chemical composition of the preparation solution was noticed. This highlighted the high influence of the chemical composition on the operational efficiency of MFI zeolite membranes. The coupling of PCA with experimental results can provide a data-driven enhancement strategy for the manufacturing of MFI-type zeolite membranes used for ethanol/water separation.

Development of Android App on Scheduling Payments and Financial Management of Workers in Construction Industry

Abstract: A innovation centralized platform designed specifically for labour payment to efficiently manage project management, labour payments and generate detailed financial reports. Mobile computing offers helps to construction workers improve efficiency and minimize the paper works. However, the uptake of the technology is quite low in the industry, with little research on key constraint factors and the priority needs of the construction workforce. In recent year mobile applications have become increasingly popular years and is expected to continue like that as all the tech is moving towards mobile. However, unfortunately, in the face of accelerating user demand, businesses are not validating that their apps are safe, secure and they are more dragged by the speed-to-market pressure as well as the need to maintain a high level of user experience. As the application contains all the payment information tedious writing work can be avoided on site. Just set the payment in the application and are free to go. Extensive time and effort are spent on inspecting and documenting of payments. Managing large number of projects with various subcontractors will be difficult. If the hard copy is lost, entire data set cannot be taken back. In financial updates client also face same problem, to get clients according to their requirements. To bridge this gap and to develop better connection between client, engineers and subcontractors, we need a platform. Project cost based on labour payments and payments pending can be accessed easily. Maintain a history of project payments, weekly payments balance payments tracker

Identification and validation of the TRHDE-AS1/hsa-miR-449a/ADAMTS5 axis as a novel prognostic biomarker in prostate cancer.

Despite advancements in cancer research, the prognostic implications of competing endogenous RNA (ceRNA) networks in prostate cancer (PCa) remain incompletely understood. This study aimed to elucidate the prognostic relevance of ceRNA networks in PCa, utilizing a comprehensive bioinformatics approach alongside experimental validation. After searching The Cancer Genome Atlas (TCGA) database, RNA sequencing (RNA-Seq) data were extracted to identify differentially expressed RNAs (DERs) between 491 PCa samples and 51 normal prostate tissues, following which a comprehensive bioinformatics strategy was implemented to construct a ceRNA network. An optimal prognostic signature comprising these DERs was then established and validated using TCGA data. In addition, functional validation was performed through RNA pull-down, dual-luciferase reporter assays, quantitative real-time PCR, and western blot analysis conducted in PC-3 and DU145 cell lines, thereby complementing the bioinformatics analysis. A total of 613 DERs, comprising 103 long noncoding RNAs (lncRNAs), 60 microRNAs (miRNAs), and 450 messenger RNAs (mRNAs), were identified and utilized in constructing a ceRNA network, which encompassed 23 lncRNAs, 9 miRNAs, and 52 mRNAs. An optimal prognostic signature was established, including VPS9D1 antisense RNA 1 (VPS9D1-AS1), miR-449a, cyclin-dependent kinase 5 regulatory subunit 1 (CDK5R1), targeting protein for Xklp2 (TPX2), solute carrier family 7 member 11 (SLC7A11), copine7 (CPNE7), and maternal embryonic leucine zipper kinase (MELK), yielding area under the curve (AUC) values exceeding 0.8 across training, validation, and entire datasets. Our experiments results revealed an interaction between lncRNA TRHDE antisense RNA 1 (TRHDE-AS1) and miR-449a and that miR-449a could target the ADAM metallopeptidase with thrombospondin type 1 motif 5 (ADAMTS5) mRNA. Knockdown of miR-449a significantly impeded cell proliferation, G1/S transition, migration and invasion, and promoted apoptosis in PC-3 and DU145 cells. Furthermore, knockdown of miR-449a notably downregulated protein expression of CDK4, cyclin D1, N-cadherin and vimentin, while upregulating protein expression of cleaved caspase-3 and E-cadherin. This study contributes to a deeper understanding of the prognostic-linked ceRNA network in PCa, providing fundamental insights that could improve diagnostic and therapeutic approaches for PCa management.

Shedding more light on shedders

We report on testing 100 individuals for their shedder status with the aim of demonstrating whether the process of cell staining is reproducible when testing a large number of people. A previous report using the same method was based on 11 donors and indicated that there may be a continuum of shedder types within this small sample set. In this report we also expand the time points post-handwashing to 0, 15, 30, 60, and 180 min. Triplicate samples were collected from both the right and left thumbs. Samples were collected by donors placing a thumb on a clean glass slide and then adding a DNA binding dye. The number of cells were recorded within three separate square millimetre areas (cells/mm2) at 220x magnification. The experiments were conducted in triplicate on three different days, giving a total of 72 thumbprints per individual. Finally, there were 3438 observed frames in the entire dataset. Of the 100 donors, 98 gave consistent and reproducible cell number deposition. There was no difference between the cells deposited by the left and right thumbs in 13 of 15 tested. Males tended to deposit more cells than females. If applying arbitrary boundary to a cell count to definitively determine shedder status, then many of the donors fell within two categories. This study based on 100 individuals strongly suggests that shedder status is a continuum phenomenon.

Automatic breast ultrasound (ABUS) tumor segmentation based on global and local feature fusion

Accurate segmentation of tumor regions in automated breast ultrasound (ABUS) images is of paramount importance in computer-aided diagnosis system. However, the inherent diversity of tumors and the imaging interference pose great challenges to ABUS tumor segmentation. In this paper, we propose a global and local feature interaction model combined with graph fusion (GLGM), for 3D ABUS tumor segmentation. In GLGM, we construct a dual branch encoder-decoder, where both local and global features can be extracted. Besides, a global and local feature fusion module is designed, which employs the deepest semantic interaction to facilitate information exchange between local and global features. Additionally, to improve the segmentation performance for small tumors, a graph convolution-based shallow feature fusion module is designed. It exploits the shallow feature to enhance the feature expression of small tumors in both local and global domains. The proposed method is evaluated on a private ABUS dataset and a public ABUS dataset. For the private ABUS dataset, the small tumors (volume smaller than 1 cm3) account for over 50% of the entire dataset. Experimental results show that the proposed GLGM model outperforms several state-of-the-art segmentation models in 3D ABUS tumor segmentation, particularly in segmenting small tumors.

Can Learned Indexes be Built Efficiently? A Deep Dive into Sampling Trade-offs

By embedding the distribution of keys in indexing structure, learned indexes can minimize the index size and maximize the lookup performance. Yet, one of the problems in the present learned index is the long index-building time. The conventional learned index requires a complete traversal of the entire dataset, which makes it less practical than traditional index. This paper challenges the efficiency of build time to make the learned index practical. Our approach for a build time-efficient learned index is to employ sampled learning. In this paper, we present two error-bounded sampling schemes: Sample EB-PLA, and Sample EB-Histogram. Although sampling is a simple idea, there are several considerations to make it practical. For example, sampling interval, error-boundness, and index hyper-parameters are inter-related each other, presenting complicated trade-offs between build-time, index size, accuracy and lookup latency. Throughout the extensive experiments over six real-world datasets, we show that the index-building time can be efficiently reduced over an order of magnitude by our sampling schemes. The results reveal that the sampling expands the design space of learned indexes, including the build-time as well as lookup performance and index size. Our Pareto analysis shows that a learned index can be built more efficiently than a traditional index through sampling.