Search

This paper takes education data mining as the research theme, mine the existing massive education big data, compares the analysis methods of existing data models, and proposes an improved random forest reference model. The information gain of various features is calculated by introducing the feature weighting system, and the evaluation index is used to improve the existing data analysis. The simulation results show that the improved model is highly efficient as compared to the existing models for classification. In order to resolve the performance bottleneck of a single node in multiple data classification tasks in the era of big data, a classification and prediction model of graduates’ large-scale employment data, based on distributed improved RF algorithm, is proposed. The MapReduce distributed computing framework is used to complete the serial writing and deserialization loading of the training model between the local disk and the distributed file system, and realizing the distributed expansion of the large-scale data classification model based on the improved RF model.

Background In order to end tuberculosis (TB), it is necessary to expand coverage of TB care services, including systematic screening initiatives. However, more evidence is needed for groups among whom systematic screening is only conditionally recommended by the World Health Organization. This study evaluated concurrent screening in multiple target groups using community health workers (CHW). Methods In our two-year intervention study lasting from October 2017 to September 2019, CHWs in six districts of Ho Chi Minh City, Viet Nam verbally screened three urban priority groups: (1) household TB contacts; (2) close TB contacts; and (3) residents of urban priority areas without clear documented exposure to TB including hotspots, boarding homes and urban slums. Eligible persons were referred for further screening with chest radiography and follow-on testing with the Xpert MTB/RIF assay. Symptomatic individuals with normal or without radiography results were tested on smear microscopy. We described the TB care cascade and characteristics for each priority group, and calculated yield and number needed to screen. Subsequently, we fitted a mixed-effect logistic regression to identify the association of these target groups and secondary patient covariates with TB treatment initiation. Results We verbally screened 321 020 people including 24 232 household contacts, 3182 social and close contacts and 293 606 residents of urban priority areas. This resulted in 1138 persons treated for TB, of whom 85 were household contacts, 39 were close contacts and 1014 belonged to urban priority area residents. The yield of active TB in these groups was 351, 1226 and 345 per 100 000, respectively, corresponding to numbers needed to screen of 285, 82 and 290. The fitted model showed that close contacts [adjusted odds ratio (aOR) = 2.07; 95% CI: 1.38–3.11; P OR = 2.18; 95% CI: 1.69–2.79; P Conclusions The study detected a large number of unreached persons with TB, but most of them were not among persons in contact with an index patient. Therefore, while programs should continue to optimize screening in contacts, to close the detection gap in high TB burden settings such as Viet Nam, coverage must be expanded to persons without documented exposure such as residents in hotspots, boarding homes and urban slums.

The objective of this research was to 1) study and develop a smartphone application for food bloggers, and 2) evaluate the result of using the application regarding the benefits of usage. This is mixed-method research and development. The application development processes composed of the study of user requirements, application design, application development, application testing, improvement, and evaluation. In the study of the user requirement process, data were collected from research samples by questionnaires. 400 research samples were selected from Thai people in the Bangkok area by a simple random sampling method. Data were analyzed in terms of frequency, percentage, mean, and standard deviation. Moreover, researchers also conducted interviews in order to collect qualitative data from 2 food bloggers in Bangkok about the requirements and relevant information about food blog application. After that, the smartphone application for food bloggers was designed, developed, tested, improved, and evaluated. Research results were as the followings. In terms of application users, the relevant activities composed of user registration, food contents uploading and presentations, comments, and evaluations. The application provided food details information, such as food menus, ingredients, cooking process, etc. in terms of texts, pictures, and videos. Food information was classified in terms of food types, nature of food culture by countries, and popular food ranking. Comment page show details about user comments with comments about posts. The application focused on colorful, beautiful, pleasant to use, which was appropriate for those who like to present food and let others know about restaurants and food in various stores. The developed smartphone application provided a social media channel. Users can register and log in through the application in order to submit their work by posting a message, pictures, and details about food information that they want to offer. Bloggers can edit posts made by themselves easily. Visitors can follow food bloggers which they are interested in. However, the researcher had suggestions as follows: 1) the food blog owner should provide more foreign languages, such as English, Chinese, Japanese for users in order to promote Thai food tourism, 2) some famous food bloggers should be invited to write some interesting content in order to promote application marketing, 3)interesting cooking lesson should be provided to visitors, 4) food contents must be active regularly, 5) the application sponsors must be arranged as well, and 6) food block owner have to set good rules and regulations for food blockers and visitors clearly.

This study aimed to assess the rate and level of depression, anxiety, and stress among people with infertility and the association between psychological disorders and sexual dysfunction. In this cross-sectional study, we investigated 255 couples with infertility undergoing examination at a university hospital between January and December 2017. The assessment tools included the Depression, Anxiety and Stress Scale-21, Female Sexual Function Index (FSFI), Premature Ejaculation Diagnostic Tool (PEDT), and International Index of Erectile Function-15 (IIEF-15). The rates and mean scores of depression, anxiety, and stress among wives were statistically significantly higher than among husbands. Additionally, there was a slight negative correlation between all these scores and FSFI score among wives. Similarly, these scores among husbands were slightly negatively correlated with IIEF-15 score and positively correlated with PEDT score. Development of a holistic approach to infertility management is necessary and psychological issues should be addressed together with sexual problems.

Many organelles inside and outside a living cell depend on the perfect behavior of Golgi apparatus for smooth and normal functioning. Its poor performance may lead to many inheritable diseases like diabetes and cancer. Therefore, it is highly crucial to detect any strange behavior of Golgi apparatus in advance. Accurate discrimination of cis -Golgi from trans -Golgi proteins surely helps researchers identify the role of Golgi proteins in various diseases and assist pharmacists in drug development. In this work, various hybrid models of Bi-Profile Bayes, Bigram PSSM, Di-Peptide Composition, and Split Amphiphilic Pseudo Amino Acid Composition with SMOTE oversampling technique have been employed to discriminate Golgi protein types. Multiple linear Support Vector Machines have been used to exploit the discrimination power of these models. The proposed prediction system: Golgi-predictor has shown significant performance and achieved promising results compared to other existing state-of-the-art techniques. Through the 10-fold cross-validation, the proposed system achieved an accuracy value of 97.6%, sensitivity value of 98.8%, specificity value of 96.5%, G-mean value of 97.6%, MCC value of 0.95, and F-score value of 0.97. Similarly, through the jackknife cross-validation, the achieved values for accuracy, sensitivity, specificity, G-mean, MCC, and F-score are respectively, 96.5%, 97.8%, 95.2%, 96.4%, 0.93, and 0.96. Moreover, through the independent dataset testing, Golgi-predictor demonstrated significant enhancement in performance over other techniques. The proposed methodology aims at supporting drug designers in pharmaceutical industry and assisting researchers from the fields of bioinformatics and computational biology towards better innovation in predicting the behavior of Golgi proteins.

Background Many tuberculosis (TB) patients incur catastrophic costs. Active case finding (ACF) may have socio-protective properties that could contribute to the WHO End TB Strategy target of zero TB-affected families suffering catastrophic costs, but available evidence remains limited. This study measured catastrophic cost incurrence and socioeconomic impact of an episode of TB and compared those socioeconomic burdens in patients detected by ACF versus passive case finding (PCF). Methods This cross-sectional study fielded a longitudinal adaptation of the WHO TB patient cost survey alongside an ACF intervention from March 2018 to March 2019. The study was conducted in six intervention (ACF) districts and six comparison (PCF) districts of Ho Chi Minh City, Viet Nam. Fifty-two TB patients detected through ACF and 46 TB patients in the PCF cohort were surveyed within two weeks of treatment initiation, at the end of the intensive phase of treatment, and after treatment concluded. The survey measured income, direct and indirect costs, and socioeconomic impact based on which we calculated catastrophic cost as the primary outcome. Local currency was converted into US$ using the average exchange rates reported by OANDA for the study period (VNĐ1 = US$0.0000436, 2018–2019). We fitted logistic regressions for comparisons between the ACF and PCF cohorts as the primary exposures and used generalized estimating equations to adjust for autocorrelation. Results ACF patients were poorer than PCF patients (multidimensional poverty ratio: 16 % vs. 7 %; p = 0.033), but incurred lower median pre-treatment costs (US$18 vs. US$80; p p p = 0.011), especially during the intensive phase (OR = 0.32; 95 % CI: [0.12, 0.90]; p = 0.030). ACF patient experienced less social exclusion (OR = 0.41; 95 % CI: [0.18, 0.91]; p = 0.030), but more often resorted to financial coping mechanisms (OR = 5.12; 95 % CI: [1.73, 15.14]; p = 0.003). Conclusions ACF can be effective in reaching vulnerable populations and mitigating the socioeconomic burden of TB, and can contribute to achieving the WHO End TB Strategy goals. Nevertheless, as TB remains a catastrophic life event, social protection efforts must extend beyond ACF.

BackgroundRespiratory distress syndrome (RDS) is one of the leading causes of early neonatal morbidity and mortality in late preterm infants (LPIs) worldwide. This matched cohort study aimed to assess how the antenatal dexamethasone use affect the respiratory distress (RD) proportion in preterm newborns between 34 0/7 weeks and 36 6/7 weeks of gestation.MethodsThis was a prospective cohort study on 78 women with singleton pregnancy who were in threatened preterm birth and had not received prior dexamethasone, who were admitted between 34 0/7 weeks and 36 6/7 weeks at Hue University of Medicine and Pharmacy Hospital from June 2018 to May 2020. The matched control group without dexamethasone use included 78 pregnant women diagnosed with threatened late preterm births who were at similar gestational ages and estimated fetal weights as the treatment group. The treatment group received 6 mg intramuscular dexamethasone every 12 h for a total of 4 doses or until delivery. Primary outcome was the rate of neonatal RD. Secondary neonatal outcomes included the need for respiratory support, neonatal intensive care unit (NICU) admission, hypoglycemia, necrotizing enterocolitis, intraventricular hemorrhage, and neonatal death. Statistical analyses were performed by using SPSS software, version 26.0.ResultsThe proportion of RD in LPI was significantly lower in the treatment group than in the matched control group (10.3% vs. 23.1%, respectively), adjusted Odds Ratio [aOR] 0.29; 95% confidence interval [CI] 0.10 – 0.83 and p = 0.021. Neonatal hypoglycemia was more common in the dexamethasone group than in the matched group (25.6% vs. 12.8%, respectively; aOR, 2.59; 95% CI, 1.06 – 6.33; p = 0.037). There were no significant between-groups differences in the incidence of respiratory support, NICU admission or length of hospital stay.ConclusionsAdministration of antenatal dexamethasone to women at risk for late preterm birth could help to lower the proportion of respiratory distress in late preterm infants.

Being considered as one of the most prominent and detrimental neurological disorders, diagnosing what category of brain tumor disease as soon as possible is tremendously imperative for patients, which is excessively relied on human factors on determining brain tumor type. In order to address the said issue and enhance the classifying performance in deep learning, the paper proposes a myriad of methods combined with Convolutional Neural Networks, namely transfer learning, data augmentation, the arrangement between Batch Normalization and Dropout. Eventual experimental results prove that the proposed approaches outperform the state-of-the-art papers on the benchmark brain tumor dataset. The proposed architecture for each particular Convolutional Neural Network depicts that the outcomes are more prospective than those original methods and default-set parameters. The highest accuracy conducted experiments is 98.8%.

Objectives: The frequency of, and relationship between, the various manifestations of male sexual dysfunction in infertile couples have been poorly investigated, especially in Vietnam. Our study aims to assess the prevalence of premature ejaculation and erectile dysfunction in infertile couples using validated instruments, and the relationship between these disorders. Method: Cross-sectional descriptive study, using validated questionnaires including the Premature Ejaculation Diagnostic Tool (PEDT) and the International Index of Erectile Function-15 (IIEF-15), to measure the incidence of these problems in 255 male partners of infertile couples who were examined from January through December 2017, at the Center for Reproductive Endocrinology & Infertility, Hue University Hospital. Results: The prevalence of overt premature ejaculation was 4.7%, probable premature ejaculation was 7.1%, and erectile dysfunction was 26.3% (mild: 19.3%, mild-to-moderate: 3.9%, moderate: 2.7%, and severe: 0.4%). The PEDT total score was negatively correlated to IIEF-15-EFD and IIEF-15 total scores (r [Formula: see text]0.322 and r [Formula: see text]0.348, respectively). Conclusions:In light of the identified prevalence of premature ejaculation and erectile dysfunction in the studied population, screening for these conditions should be included in the evaluation of infertile couples. These two disorders could negatively reciprocal effect on each other.

Gender prediction is a common topic in machine learning. It can be seen that many approaches have been proposed for this task and achieved certain successes. However, it is necessary to figure out a method that has high performance and reduce time processing. This paper focuses on applying deep neural networks to solve gender prediction based on facial images. Specifically, Convolutional Neural Networks (CNN) and its enhancements are applied to extract features of facial images. Then, K-means clustering is employed to predict gender on a large-scale GenderFace80K dataset with 80,000 facial images with gender annotation.

Crop cultivation plays an essential role in the agricultural field. Presently, the loss of food is mainly due to infected crops, which reflexively reduces the production rate. Health monitoring and disease detection on plant is very critical for sustainable agriculture. It is very difficult to monitor the plant diseases manually. It requires a tremendous amount of work as well as expertise. Hence, an intelligent crop health assessment system using deep learning based on Convolutional Neural Networks (CNN) was proposed. The steps involved are image acquisition, pre-processing, data augmentation and classification. Image of the plant is captured using a smartphone with a camera. Captured images are pre-processed. Data augmentation has been done on the training data set. Implementation of CNN model yielded an accuracy of 85%. The model has been tested against a set of images collected manually.

BackgroundRespiratory distress (RD) is one of the leading causes of early neonatal morbidity and mortality in late preterm infants (LPIs) worldwide. This study aimed to evaluate the effect of antenatal dexamethasone on the RD rate in preterm newborns between 34 0/7 weeks and 36 6/7 weeks of gestation.MethodsThis was a prospective cohort study of 78 women with singleton pregnancy who were in threatened preterm birth and had not received prior dexamethasone who were admitted between 34 0/7 weeks and 36 6/7 weeks at Hue University of Medicine and Pharmacy Hospital from June 2018 to May 2020. The matched control group (no dexamethasone) included 78 pregnant women with threatened late preterm births who were at similar gestational ages and estimated fetal weights as the treatment group. The treatment group received 6 mg intramuscular dexamethasone every 12 hours for a total of 4 doses or until delivery. Primary outcome was the rate of neonatal RD. Secondary neonatal outcomes included the need for respiratory support, neonatal intensive care unit (NICU) admission, hypoglycemia, necrotizing enterocolitis, intraventricular hemorrhage, and neonatal death. Statistical analyses were performed by using SPSS software.ResultsThe rate of RD in LPI was significantly lower in the treatment group than in the matched control group (10.3% vs. 23.1%, respectively), with relative risk [RR] 0.44; 95% confidence interval [CI] 0.21 – 0.96 and p = 0.03. Neonatal hypoglycemia was more common in the dexamethasone group than in the matched group (25.6% vs. 12.8%, respectively; RR, 2.00; 95% CI, 1.00 – 3.99; p = 0.04). There were no significant between-group differences in the incidence of respiratory support, NICU admission or length of hospital stay.ConclusionsAdministration of antenatal dexamethasone to women at risk for late preterm birth could significantly reduce the rate of respiratory distress in late preterm infants.

Histogram of Oriented Gradient (HOG) is a robust descriptor which is widely used in many real-life applications, including human detection, face recognition, object counting, and video surveillance. In order to extract HOG descriptor from color images whose information is three times more than the grayscale images, researchers currently apply the maximum magnitude selection method. This method makes the information of the resulted image is reduced by selecting the maximum magnitudes. However, after we extract HOG using the unselected magnitudes of the maximum magnitude selection method, we observe that the performance is better than using the maximum magnitudes in several cases. Therefore, in this paper, we propose a novel approach for extracting HOG from color images such as Color Component Selection and Color Component Fusion. We also propose the extended kernels in order to improve the performance of HOG. With our new approaches in the color component analysis, the experimental results of several facial benchmark datasets are enhanced with the increment from 3 to 10% of accuracy. Specifically, a 95.92% of precision is achieved on the Face AR database and 75% on the Georgia Face database. The results are better more than 10 times compared with the original HOG approach.

Hand gesture recognition has various applications in recent years such as robotics, e-commerce, human–machine interaction, e-sport, and assisting people with hearing-impaired humans. The latter is the most useful and interesting application in our daily life. Nowadays, cameras can be installed easily and everywhere. So, gesture recognition faces the most challenging issue under image acquisition by multiple cameras. This paper introduces an approach for hand gesture recognition under multi-views cameras. The proposed approach is evaluated on the HGM-4 benchmark dataset by using local binary patterns.

Unmanned aerial vehicle (UAV) has recently gained significant attention due to their efficient structures, cost-effectiveness, easy availability, and tendency to form an ad hoc wireless mobile network. IoT-enabled UAV is a new research domain that uses location tracking with the advancement of aerial technology. In this context, the importance of 3D aerial networks is attracting a lot of attention recently. It has various applications related to information processing, communication, and location-based services. Location identification of wireless nodes is a challenging job and of extreme importance. In this study, we introduced a novel technique for finding indoor and open-air three-dimensional (3D) areas of nodes by measuring the signal strength. The mathematical formulation is based on a path loss model and decision tree machine learning classifier. We constructed 2D and 3D models to gather more accurate information on the nodes. Simulation findings demonstrate that the proposed machine learning-based model excels in nodes location estimation, the actual and estimated distance of different nodes, and calculation of received signal strength in aerial ad hoc networks. In addition, the decision tree constructs an offline phase control in the flying vehicle’s location to enhance the time complexity along with experimental accuracy.

Rice is one of the primary foods for human, and various rice varieties have been cultivated in many countries. Automatic rice variety inspection system based on computer vision is needed instead of using technical human expert. In this chapter, we have investigated three types of local descriptor such as local binary pattern, histogram of oriented gradients and GIST to characterize rice seed images. In order to enhance the robustness, multi-view approach has been considered to concatenate these features. However, this problem encounters the high-dimensional space feature vector and need to select the relevant features for a compact and better model. We have applied different feature selection methods in a filter way to eliminate the impertinent features. The experimental results on the VNRICE dataset have been shown the efficiency of our proposition.

In this paper, we propose a semisupervised feature selection approach that is based on feature clustering and hypothesis margin maximization. The aim is to improve the classification accuracy by choosing the right feature subset and to allow building more interpretable models. Our approach handles the two core aspects of feature selection, i.e., relevance and redundancy, and is divided into three steps. First, the similarity weights between features are represented by a sparse graph where each feature can be reconstructed from the sparse linear combination of the others. Second, features are then hierarchically clustered identifying groups of the most similar ones. Finally, a semisupervised margin-based objective function is optimized to select the most data discriminative feature from within each cluster, hence maximizing relevance while minimizing redundancy among features. Eventually, we empirically validate our proposed approach on multiple well-known UCI benchmark datasets in terms of classification accuracy and representation entropy, where it proved to outperform four other semisupervised and unsupervised methods and competed with two widely used supervised ones.

Automatic gender determination received many attentions in the recent years due to its potential applications in e-commerce and demographic data collection. Face and voice are the most common factors of human which are used to determine the gender. A comparative study of gender recognition based hand-crated and deep features via ear images is introduced in this paper. The EarVN1.0 dataset is employed to evaluate this study. The experimental results show that deep learning approach clearly outperforms features-based methods for gender determination based on ear images.

Background The World Health Organization recently recommended Video Observed Therapy (VOT) as one option for monitoring tuberculosis (TB) treatment adherence. There is evidence that private sector TB treatment has substandard treatment follow-up, which could be improved using VOT. However, acceptability of VOT in the private sector has not yet been evaluated. Methods We conducted a cross-sectional survey employing a theoretical framework for healthcare intervention acceptability to measure private provider perceptions of VOT across seven constructs in three cities of Viet Nam: Ha Noi, Ho Chi Minh City, and Hai Phong. We investigated the differences in private providers’ attitudes and perceptions of VOT using mixed ordinal models to test for significant differences in responses between groups of providers stratified by their willingness to use VOT. Results A total of 79 private providers completed the survey. Sixty-two providers (75%) indicated they would use VOT if given the opportunity. Between private providers who would and would not use VOT, there were statistically significant differences (p≤0.001) in the providers’ beliefs that VOT would help identify side effects faster and in their confidence to monitor treatment and provide differentiated care with VOT. There were also significant differences in providers’ beliefs that VOT would save them time and money, address problems faced by their patients, benefit their practice and patients, and be relevant for all their patients. Conclusion Private providers who completed the survey have positive views towards using VOT and specific subpopulations acknowledge the value of integrating VOT into their practice. Future VOT implementation in the private sector should focus on emphasizing the benefits and relevance of VOT during recruitment and provide programmatic support for implementing differentiated care with the technology.

This paper deals with the problem of detecting surface cavities on metal castings that must be distinguished from their lookalike non-defective substances embedding on the surface. As general detection methods working on the basis of single 2D appearance and patterns are easily fooled by their similar 2D appearance and less suitable, the 3D geometry, e.g. depth, of the scene must also be considered. However, ground-truth depth maps are formidable and cumbersome to obtain, even with the help of modern depth-sensing devices. We present an automated framework for effectively detecting surface cavities from only monocular vision by leveraging consecutive images from different views. We first employ an object detector to localize regions of interest (ROIs) that are candidates for defect detection, including cavities and cavity lookalikes, intuitively identical to the way human vision operates by focusing on the most important regions. We then predict and extract the pixel-wise depths of the detected ROIs with a depth estimator, which is trained in an unsupervised manner from sequences of 3 consecutive monocular images taken at different camera poses based on their relative depth consistency. The object detector and the depth estimator jointly track the ROIs over a number of images from continuous views for thorough examination. Experimental results with a real-world dataset demonstrate that our method is able to identify and characterize cavity defects into deeper regions for ease of detection.

Plant identification based on deep learning received many attention and effort from the research community with many promising results. It becomes an active trend in the recent years. We apply Convolutional Neural Network (CNN) to recognize Vietnamese medicinal plant images in this paper. Different frameworks are evaluated such as: VGG16, Resnet50, InceptionV3, DenseNet121, Xception and MobileNet. The highest accuracy reached by Xception with 88.26%. We might think this approach will greatly contribute to the discovery and conservation of valuable medicinal plants.

In smart agriculture, rice variety inspection systems based on computer vision need to be used for recognizing rice seeds instead of using technical experts. In this paper, we have investigated three types of local descriptors, such as Local Binary Pattern (LBP), Histogram of Oriented Gradients (HOG) and GIST to characterize rice seed images. However, this approach raises the curse of dimensionality phenomenon and needs to select the relevant features for a compact and better representation model. A new ensemble feature selection is proposed to represent all useful information collected from different single feature selection methods. The experimental results have shown the efficiency of our proposed method in terms of accuracy.

Under-detection and -reporting in the private sector constitute a major barrier in Viet Nam’s fight to end tuberculosis (TB). Effective private-sector engagement requires innovative approaches. We established an intermediary agency that incentivized private providers in two districts of Ho Chi Minh City to refer persons with presumptive TB and share data of unreported TB treatment from July 2017 to March 2019. We subsidized chest x-ray screening and Xpert MTB/RIF testing, and supported test logistics, recording, and reporting. Among 393 participating private providers, 32.1% (126/393) referred at least one symptomatic person, and 3.6% (14/393) reported TB patients treated in their practice. In total, the study identified 1203 people with TB through private provider engagement. Of these, 7.6% (91/1203) were referred for treatment in government facilities. The referrals led to a post-intervention increase of +8.5% in All Forms TB notifications in the intervention districts. The remaining 92.4% (1112/1203) of identified people with TB elected private-sector treatment and were not notified to the NTP. Had this private TB treatment been included in official notifications, the increase in All Forms TB notifications would have been +68.3%. Our evaluation showed that an intermediary agency model can potentially engage private providers in Viet Nam to notify many people with TB who are not being captured by the current system. This could have a substantial impact on transparency into disease burden and contribute significantly to the progress towards ending TB.

Ground-based cloud images recognition is a difficult task because of extreme varieties in cloud appearance under various atmospheric conditions. Local Binary Pattern and its variants are widely applied due to its efficiency for texture classification. Among them, Local Ternary Pattern (LTP) descriptor is the most common to characterize cloud images. In this approach, an additional parameter is used to define a tolerance for measuring the similarity between different gray intensities. This paper proposes a new approach to determine a suitable value of the parameter for LTP descriptor by the supervised learning approach. The proposed method is evaluated on the SWIMCAT database.

Online opinions are used as a data source that contains relevant information about customer sentiments toward a product or service. This can be used to make a specific decision for customers and management. In order to achieve the good models for sentiment analysis, we require a large human-labeled data which is costly to obtain. This paper proposes an approach based on text data augmentation based on product reviews in Vietnamese language. Several basic techniques are applied to generate more comments by random insertions, substitutions. The experimental results demonstrate the efficiency of the proposed approach.

Background To achieve the WHO End TB Strategy targets, it is necessary to detect and treat more people with active TB early. Scale–up of active case finding (ACF) may be one strategy to achieve that goal. Given human resource constraints in the health systems of most high TB burden countries, volunteer community health workers (CHW) have been widely used to economically scale up TB ACF. However, more evidence is needed on the most cost-effective compensation models for these CHWs and their potential impact on case finding to inform optimal scale-up policies. Methods We conducted a two-year, controlled intervention study in 12 districts of Ho Chi Minh City, Viet Nam. We engaged CHWs as salaried employees (3 districts) or incentivized volunteers (3 districts) to conduct ACF among contacts of people with TB and urban priority groups. Eligible persons were asked to attend health services for radiographic screening and rapid molecular diagnosis or smear microscopy. Individuals diagnosed with TB were linked to appropriate care. Six districts providing routine NTP care served as control area. We evaluated additional cases notified and conducted comparative interrupted time series (ITS) analyses to assess the impact of ACF by human resource model on TB case notifications. Results We verbally screened 321,020 persons in the community, of whom 70,439 were eligible for testing and 1138 of them started TB treatment. ACF activities resulted in a + 15.9% [95% CI: + 15.0%, + 16.7%] rise in All Forms TB notifications in the intervention areas compared to control areas. The ITS analyses detected significant positive post-intervention trend differences in All Forms TB notification rates between the intervention and control areas (p = 0.001), as well as between the employee and volunteer human resource models (p = 0.021). Conclusions Both salaried and volunteer CHW human resource models demonstrated additionality in case notifications compared to routine case finding by the government TB program. The salaried employee CHW model achieved a greater impact on notifications and should be prioritized for scale-up, given sufficient resources.

Plant identification is an essential topic in computer vision with various applications such as agronomy, preservation, environmental impact, discovery of natural and pharmaceutical product. However, the standard and available dataset for medicinal plants have not been widely published for research community. This work contributes the first large, public and multi class dataset of medicinal plant images. Our dataset consists of total 20,000 images of 200 different labeled Vietnamese medicinal plant (VNPlant-200). We provide this dataset into two versions of size 256 × 256 and 512 × 512 pixels. The training set consists of 12,000 images and the remainder are used for testing set. We apply the Speed-Up Robust Features (SURF) and Scale Invariant Feature Transform (SIFT) for extracting features and the Random Forest (FR) classifier is associated to recognize plant. The experimental results on the VNPlant-200 have been shown the interesting challenge task for pattern recognition.

The texture classification problem is widely applied for ground-based cloud images recognition due to its efficiency. Local Binary Pattern and its variants are usually investigated to represent cloud images. The appropriate choice of color space might enhance performance of the recognition system for many applications. In this paper, we propose a comparative study to select the best candidate’s space for coding cloud images by fusing multiple texture description methods. The proposed approach is evaluated on the SWIMCAT database.

Face matching is an active research topic in the last decade due to various applications in pattern recognition. Rather than using a single feature type, the fusion of many distinct features might decrease the error rate of facial recognition systems. This also increases the time processing and data storage. In this paper, we first employ feature fusion extracted from HOG and GIST descriptor from facial image and use Canonical Correlation Analysis (CCA) to combine into a single feature. Then, a feature selection approach based on Fisher ranking is considered to remove irrelevant and noisy features. The experiment is evaluated on three common datasets (AR, Georgia Tech and MUCT) which have been shown the improvement of the proposed approach.

Gesture recognition technology is rapidly growing in the recent years due to the demands of many application such as computer game and sport, human robot interaction, assistant systems, sign language interpretation and e-commerce. One of the most important of gesture recognition is hand-gesture recognition. For example, it can be used to control all devices (television, radio, air-condition, and doors) by just hand gestures for smart home application. The HGM-4 dataset is built for hand gesture recognition (the full dataset is available from: https://data.mendeley.com/datasets/jzy8zngkbg/4 ) which contains total 4,160 color images (1280 × 700 pixels) of 26 hand gestures captured by four cameras at different position. The training and testing set are defined to create a benchmark framework for comparing the experimental results.

Forest carbon sequestration offsets are methodologically uncertain, comprise a minor component of carbon markets and do not effectively slow deforestation. The objective of this study is to describe a commercial scale in situ measurement approach for determination of net forest carbon sequestration projects, the Direct Measurement Forest Carbon Protocol™, to address forest carbon market uncertainties. In contrast to protocols that rely on limited forest mensuration, growth simulation and exclusion of CO2 data, the Direct Measurement Forest Carbon Protocol™ is based on standardized methods for direct determination of net ecosystem exchange (NEE) of CO2 employing eddy covariance, a meteorological approach integrating forest carbon fluxes. NEE is used here as the basis for quantifying the first of its kind carbon financial products. The DMFCP differentiates physical, project and financial carbon within a System-of-Systems™ (SoS) network architecture. SoS sensor nodes, the Global Monitoring Platform™ (GMP), housing analyzers for CO2 isotopologues (e.g., 12CO2,13CO2, 14CO2) and greenhouse gases are deployed across the project landscape. The SoS standardizes and automates GMP measurement, uncertainty and reporting functions creating diverse forest carbon portfolios while reducing cost and investment risk in alignment with modern portfolio theory. To illustrate SoS field deployment and operation, published annual NEE data for a tropical (Ankasa Park, Ghana, Africa) and a deciduous forest (Harvard Forest, Petersham, MA, USA) are used to forecast carbon revenue. Carbon pricing scenarios are combined with historical in situ NEE annual time-series to extrapolate pre-tax revenue for each project applied to 100,000 acres (40,469 hectares) of surrounding land. Based on carbon pricing of $5 to $36 per ton CO2 equivalent (tCO2eq) and observed NEE sequestration rates of 0.48 to 15.60 tCO2eq acre−1 yr−1, pre-tax cash flows ranging from $230,000 to $16,380,000 across project time-series are calculated, up to 5× revenue for contemporary voluntary offsets, demonstrating new economic incentives to reverse deforestation. The SoS concept of operation and architecture, with engineering development, can be extended to diverse gas species across terrestrial, aquatic and oceanic ecosystems, harmonizing voluntary and compliance market products worldwide to assist in the management of global warming. The Direct Measurement Forest Carbon Protocol reduces risk of invalidation intrinsic to estimation-based protocols such as the Climate Action Reserve and the Clean Development Mechanism that do not observe molecular CO2 to calibrate financial products. Multinational policy applications such as the Paris Agreement and the United Nations Reducing Emissions from Deforestation and Degradation, constrained by Kyoto Protocol era processes, will benefit from NEE measurement avoiding unsupported claims of emission reduction, fraud, and forest conservation policy failure.

Biomedical image classification often deals with limited training sample due to the cost of labeling data. In this paper, we propose to combine deep learning, transfer learning and generative adversarial network to improve the classification performance. Fine-tuning on VGG16 and VGG19 network are used to extract the good discriminated cancer features from histopathological image before feeding into neuron network for classification. Experimental results show that the proposed approaches outperform the previous works in the state-of-the-art on breast cancer images dataset (BreaKHis).

Aim: Previous studies have suggested that quercetin is effective for treating diverse chronic disorders including organ fibrosis and airway and cardiovascular disorders. To access the pharmacological background for its broad efficacy, we examined the ability of quercetin to modulate the inflammatory and fibrotic responses associated with organ injury that commonly underlie the pathogenesis of those disorders. Methods: A cutaneous wound model on rabbit ear was used for in vivo study. Quercetin was topically applied to the wounds, and the number of macrophages and myofibroblasts and the size of the hypertrophic scar formed were estimated. An in vitro study examined the ability of quercetin to inhibit cell-signaling pathways that activate RAW264.7 macrophages and primary dermal fibroblasts and the tyrosine kinase activity of discoidin domain receptor 2. Results: Quercetin reduced the population of macrophages and myofibroblasts and the scar formation in cutaneous wound healing. Quercetin suppressed the signaling pathways activating RAW264.7 macrophages and dermal fibroblasts, which is associated with its inhibition of multiple tyrosine kinases to regulate the pathways. This pharmacological activity of quercetin to simultaneously inhibit the inflammatory and fibrotic responses upon tissue damage by targeting multi-kinases could be the action mechanism to support its broad efficacy for various chronic disorders.

Objective: To evaluate prognostic values for ovarian response of age, AFC, baseline hormonal profile and Estradiol concentration in ovarian stimulation process for art cycles. Method: Retrospective cohort. Materials: Patients treated in vitro fertilization, is ovarian stimulated with antagonist protocol, at the Center for Reproductive Endocrinology and Infertility - Hue University Hospital, from 9/2013 to 9/ 2016. Results: Women with ≤5 oocytes retrieved had the average value of the parameters: AFC 6.0±3.04, FSH 9.01±2.95 mIU/ml, AMH 1.63±1.45 ng/ml, and E2N7 617.27±440.84 pg/ml. Those with >15 oocytes retrieved had the average value of the parameters: AFC 12.91±4.52, FSH 6.06±2.03 mIU/ml, AMH 7.37±15.79 ng/ml, E2N7 1612.11±612.83 pg/ml, E2NhCG 2889.62±1732.98 pg/ml, and average number of follicles ≥12mm on day 7 was 7.70±6.35. Correlation coefficients between number of oocytes retrieved and AFC, AMH and FSH N2-3, age were 0.62; 0.48; -0.39; -0.22, respectively; and the correlation coefficients between number of oocytes retrieved and E2N7, number of follicles ≥12mm on day 7, number of follicles ≥14mm on day 9 were 0.58; 0.3; 0.7, respectively. Threshold value of E2 on day 7 ≤ 823 pg/ml predicted number of oocytes retrieved ≤5 with sensitivity 78.9%, specificity 65%. Threshold value of E2 on day 7 > 1098 pg/ml predicted number of oocytes retrieved >15 with sensitivity 81.0%, specificity 73.9%. Basal LH, E2 are not related to ovarian response. Conclusions: AFC, FSH on days 2-3 and AMH were prognostic values of number of oocytes retrieved in the ovarian stimulation cycle. Of which, AFC was the best prognostic factor in the same treatment cycle; age had lower level of prognosis than parameters evaluating ovarian reserve. E2 on day 7 was able to prognosis few or too many oocytes retrieved, number of follicles ≥12mm on day 7 was able to prognosis too many oocytes retrieved but not for few or moderate oocytes retrieved. Number of follicle ≥14mm and E2 on day 9 were good for prognosing all three levels of oocytes retrieved. Key words: Number of oocytes retrieved, AFC, AMH, FSH D2-3, estradiol D7, estradiol DhCG

Kinship verification is an interesting topic of face analysis. In this paper, we propose to extract several local image descriptors including Local Binary Pattern, Local Ternary Pattern and Histogram of Gradient for extracting features from facial images. Next, we apply an extension of constraint score for features ranking and then select a subset of feature for training by Support Vector Machine. The fusion at an early stage (feature level) and fusion at a later stage (score level) are compared and evaluated on two benchmark kinship datasets KinFaceW-I and KinFaceW-II. Experimental results show that the proposed approach outperforms other works in the state-of-the-art.

Image classification problem requires a large number of labeled training images to obtain a good generalization. However, it is very time consuming and expensive to label data. The color information has been demonstrated that it contains relevant information for characterizing texture image. This paper presents a novel data augmentation method dedicated to color texture classification. We firstly use color features extracted from different color spaces to augment data in case of one sample training per class. The experimental results is very promising since it significantly improves the classification performance on four benchmark color texture datasets.

Rice is a primary source of people all around the world. Rice quality is mainly depend on the genetic characteristics of a rice variety. One of the most important tasks for quality control of rice production is rice seed classification. We present an approach for rice seed images classification based on Histogram of Oriented Gradient descriptor and feature selection. The experiment is applied on a VNRICE benchmark dataset and show the efficiency of the proposed approach by reducing the number of selected features and increasing the accuracy.

Indoor positioning based on the Wireless Fidelity (WiFi) protocol and the Pedestrian Dead Reckoning (PDR) approach is widely exploited because of the existing WiFi infrastructure in buildings and the advancement of built-in smartphone sensors. In this work, a hybrid algorithm that combines WiFi fingerprinting and PDR to both exploit their advantages as well as limiting the impact of their disadvantages is proposed. Specifically, to build a probability map from noisy Received Signal Strength (RSS), a Gaussian Process (GP) regression is deployed to estimate and construct the RSS fingerprints with incomplete data. Mean and variance of generated points are used to estimate WiFi fingerprinting position by K-nearest weights from the probability of visible RSS measurements of the online phase. In addition, a particle filter is applied to fuse PDR and WiFi fingerprinting by using the information from RSS, inertial sensors and features of indoor maps. To demonstrate the potential of the proposed framework, two case studies are considered. In the first case, a comparison is made between GP regression with K-Nearest Neighbours (KNN) method to show the improvement with a sparse input data set. In the second case, the proposed framework is compared to both the fingerprinting approach as well as the PDR algorithm. The results show significant improvements from our proposed framework. The average positioning accuracy of our proposed system can be lower than 1.2 m, which was reduced by 48% and 70% compared with the WiFi fingerprinting and the PDR method, respectively.

Background Tuberculosis (TB) remains a major cause of avoidable deaths. Economic migrants represent a vulnerable population due to their exposure to medical and social risk factors. These factors expose them to higher risks for TB incidence and poor treatment outcomes. Methods This cross-sectional study evaluated WHO-defined TB treatment outcomes among economic migrants in an urban district of Ho Chi Minh City, Viet Nam. We measured the association of a patient’s government-defined residency status with treatment success and loss to follow-up categories at baseline and performed a comparative interrupted time series (ITS) analysis to assess the impact of community-based adherence support on treatment outcomes. Key measures of interest of the ITS were the differences in step change (β6) and post-intervention trend (β7). Results Short-term, inter-province migrants experienced lower treatment success (aRR = 0.95 [95% CI: 0.92–0.99], p = 0.010) and higher loss to follow-up (aOR = 1.98 [95% CI: 1.44–2.72], p p = 0.041). There was evidence that patients > 55 years of age (aRR = 0.93 [95% CI: 0.89–0.96], p p p 6) = 1.07 [95% CI: 1.00, 1.15], p = 0.041) and reduced loss to follow-up (IRR(β6) = 0.17 [95% CI: 0.04, 0.69], p = 0.013) in the intervention districts. Loss to follow-up continued to decline throughout the post-implementation period (IRR(β7) = 0.90 [95% CI: 0.83, 0.98], p = 0.019). Conclusions Economic migrants, particularly those crossing provincial borders, have higher risk of poor treatment outcomes and should be prioritized for tailored adherence support. In light of accelerating urbanization in many regions of Asia, implementation trials are needed to inform evidence-based design of strategies for this vulnerable population.

Texture analysis has many important applications, including material recognition, face recognition, object detection, image segmentation. Local feature descriptors were the principle approach for texture analysis in the past. Recently, Convolutional Neural Network (CNN) has provided more promising results for texture recognition and other related computer vision tasks. Standard CNN takes labeled RGB images as input. However, other encoded images were used as extra input to CNN, which have been shown to improve the performance. We propose to feed CNN with the new encoded image. The experimental results on four benchmark color texture database show the efficiency of our proposed approach. The source code of our algorithm and all simulations used for this paper are publicly available at: https://sites.google.com/view/vinhsiam/codes

Face recognition is the most active research topics in machine vision because of its highly secured demands. The fusion of multiple features can enhance the accuracy of face recognition systems instead of using only one type of feature. However, this leads to increase the storage and processing time. In this work, we apply feature fusion by using Canonical Correlation Analysis to concatenate two different feature sources for coding a facial image. Three popular descriptors (LBP, HOG, GIST) have been investigated for extracting facial features based on block division.

Kinship verification is one of the interesting and challenging topics which is related to face analysis. Most of existing methods have been trying to find out the relevant information between two persons (edge nose, lip, eyes) to increase the verification rate. In this paper, we propose to apply the Local Binary Pattern for Kinship on different color space. Then the features are computed by $\chi^{2}$ distance and apply feature selection to reduce the noisy and irrelevant features. The Support Vector Machine is used to train and predict the model. We analyze the effectiveness of our method by carrying out experiments on two popular databases such as KinFaceW-I and KinFaceW-II.

In this paper, we propose to extract two types of feature from Neighbor-Center Difference Image (NCDI). NCDI is a variant of Local Binary Pattern (LBP) and originally used as input for Convolutional Neural Network (CNN). NCDI is a high dimensional feature and thus histograms are extracted from these NCDI features to mainly store useful information for statistical analysis. Two types of histograms are extracted from NCDI and then concatenated to further capture useful information. Experimental results on several benchmark color texture datasets show that the proposed approaches outperform the original LBP with a large margin (in accuracy) on several benchmark color texture datasets.

In this paper, we have summarized the well-known multi-class classifiers in the literature and applied them to evaluate on a new benchmark dataset of Vietnamese News (VNNews-01). This database is created from more than thirty Vietnamese online newspaper websites and grouped into twenty-five categories. The proposition and evaluation of this work might promote the related research for Vietnamese text mining.

Across Asia, a large proportion of people with tuberculosis (TB) do not report symptoms, have mild symptoms or only experience symptoms for a short duration. These individuals may not seek care at health facilities or may be missed by symptom screening, resulting in sustained TB transmission in the community. We evaluated the yields of TB from 114 days of community-based, mobile chest X-ray (CXR) screening. The yields at each step of the TB screening cascade were tabulated and we compared cohorts of participants who reported having a prolonged cough and those reporting no cough or one of short duration. We estimated the marginal yields of TB using different diagnostic algorithms and calculated the relative diagnostic costs and cost per case for each algorithm. A total of 34,529 participants were screened by CXR, detecting 256 people with Xpert-positive TB. Only 50% of those diagnosed with TB were detected among participants reporting a prolonged cough. The study’s screening algorithm detected almost 4 times as much TB as the National TB Program’s standard diagnostic algorithm. Community-based, mobile chest X-ray screening can be a high yielding strategy which is able to identify people with TB who would likely otherwise have been missed by existing health services.

Most of vehicle have the similar structures and designs. It is extremely complicated and difficult to identify and classify vehicle brands based on their structure and shape. As we require a quick and reliable response, so vehicle logos are an alternative method of determining the type of a vehicle. In this paper, we propose a method for vehicle logo recognition based on feature selection method in a hybrid way. Vehicle logo images are first characterized by Histograms of Oriented Gradient descriptors and the final features vector are then applied feature selection method to reduce the irrelevant information. Moreover, we release a new benchmark dataset for vehicle logo recognition and retrieval task namely, VLR-40. The experimental results are evaluated on this database which show the efficiency of the proposed approach.

Optimizing the number of ad clicks is a large-scale learning problem that is central to the multi-billion–dollar mobile advertising industry. There are currently several optimization methods being used, including ad mediation and ad positioning. Recently, researchers have recommended using ad refresh interval as a new method for optimizing mobile advertising. This paper applies that new method to optimize content duration for mobile ads. The result achieved from this optimization study could further increase revenue for mobile advertisers and publishers. This research has high applicability for the growing mobile advertising industry. It also lays out a solid background for future research in this promising area.