Scholarly Indexed Publications

Permanent URI for this collectionhttps://hdl.handle.net/10361/28605

Browse

Now showing 1 - 20 of 558

Metadata Only
Rating generation of video games using sentiment analysis and contextual polarity from microblog
(Institute of Electrical and Electronics Engineers Inc., 2018-12-01) Chakraborty, Shandro; Mobin, Iftekharul; Roy, Abhijeet; Khan, Mobtasim Hasan; Department of Computer Science and Engineering
In these days people tend to check reviews and ratings of video games before spending money and time for a game. This paper proposes a new model of sentiment analysis for video game's reviews. In the proposed model, video game's rating will be generated by doing sentiment analysis on public opinion data from Micro-blog site, Twitter. To segregate users' sentiment Naïve Bayes, Support Vector Machine, Logistic Regression and Stochastic Gradient Descent machine learning algorithms were used. This algorithms themselves were trained and tested on the Amazon game review dataset before doing sentiment analysis on Twitter Data. Furthermore, customized classifiers model was implemented which acted as voting classifiers to determine contextual polarity. This voting classifier takes results of other algorithms into account and select the best one which can obtain the most number of votes. Before implementing the classifiers, data pre pro-processing had been performed for accurate sentiment analysis. This process ensures higher accuracy for generating review ratings and distinguishing users' sentiment. Finally, algorithm's accuracy results are analyzed in vivid details. Analysis showed that our proposed technique outperforms others.
Metadata Only
Design of a solar powered LED street light: Effect of panel's mounting angle and traffic sensing
(Institute of Electrical and Electronics Engineers Inc., 2013-01-01) Ahmed, Sanjana; Zenan, Ahmed Hosne; Tasneem, Nisat; Rahman, Mosaddequr; Department of Electrical and Electronic Engineering
In this work, the effect of mounting angle of solar panels and traffic sensing on the requirement of system size, i.e., panel size and battery capacity, for a solar powered street light system is investigated. Street lights need more energy in winter than in summer due to longer winter nights. However, for a solar powered system, less energy is available in winter with less intense sunlight and shorter days. The mounting angle of panels has a great impact on the cumulative energy output of the panels. Our investigation shows that while a mounting angle of 23.1° yields maximum energy collection in summer and minimum energy collection in winter, a mounting angle of 46.5° increases the energy collection in winter and yields more or less uniform energy collection throughout the year. Assuming a panel efficiency of 16%, an average cumulative energy output of 950 Wh/m2/day with about 5-8% variation throughout the year has been found with a mounting angle of 46.5°. This results in about 9% reduction in the panel size than that required with a mounting angle of 231°. Moreover, installing a traffic sensor in the lighting system will allow detection of traffic density, thereby operating the lamps at different intensity levels as per requirements. This will save energy wastage in time of low or no traffic and will further relax the requirement on system size.
Metadata Only
A case study of internet banking security of banks operated in Bangladesh
(Institute of Electrical and Electronics Engineers Inc., 2023-01-01) Rahman S.M.M.; Alam, Md. Golam Rabiul; Department of Computer Science and Engineering
Now a day, Internet Banking is a popular service for the customer of the Banks. As a convenient way of doing banking more and more customers are registering for the internet banking. The banks also getting benefits of providing services to the customer round the clock without any manual involvement of the banker. As all the services done through an automated process, the security features should be implemented properly to protect the customers for any fraudulent transactions. The system should be available round the clock and transactions should be monitors as well as the systems should also monitored for any abnormal behavior of transactions and the system. The hacker group continuously try to penetrate the system and if become successful, the bank and customer both will bear loss. for banks, if the hackers cannot be protected, the bank may go out of business. This study intends to find out the issues of different internet banking site of Bangladeshi banks and recommend the best practices for the banks to be followed to do banking business securely. This will also secure the economy of the Country as a whole, as the banking system is the key to the Financial system of a country.
Open Access
Changes in prevalence and risk factors of hypertension among adults in Bangladesh: An analysis of two waves of nationally representative surveys
(Public Library of Science, 2021-12-01) Chowdhury, Muhammad Abdul Baker; Islam, Mirajul; Rahman, Jakia; Uddin, Mohammed Taj; Haque, Md Rabiul; Uddin, Md Jamal; Department of Mathematics and Natural Sciences
Introduction: Bangladesh is one of the countries where the prevalence of non-communicable diseases (NCDs) such as hypertension is rising due to rising living standards, sedentary lifestyles, and epidemiological transition. Among the NCDs, hypertension is a major risk factor for CVD, accounting for half of all coronary heart disease worldwide. However, detailed research in this area has been limited in Bangladesh. The objective of the study was to estimate changes in the prevalence and risk factors of hypertension among Bangladeshi adult population. The study also sought to identify socioeconomic status-related inequality of hypertension prevalence in Bangladesh. Methods: Cross-sectional analysis was conducted using nationally representative two waves of the Bangladesh Demographic and Health Survey (BDHS) in 2011 and 2017-18. Survey participants were adults 18 years or older- which included detailed biomarker and anthropometric measurements of 23539 participants. The change in prevalence of hypertension was estimated, and adjusted odds ratios were obtained using multivariable survey logistic regression models. Further, Wagstaff decomposition method was also used to analyze the relative contributions of factors to hypertension. Results: From 2011 to 2018, the hypertension prevalence among adults aged ≥35 years increased from 25.84% to 39.40% (p<0.001), with the largest relative increase (97%) among obese individuals. The prevalence among women remained higher than men whereas the relative increase among men and women were 75% and 39%, respectively. Regression analysis identified age and BMI as the independent risk factors of hypertension. Other risk factors of hypertension were sex, marital status, education, geographic region, wealth index, and diabetes status in both survey years. Female adults had significantly higher hypertension risk in both survey years in the overall analysis in, however, in the subgroup analysis, the gender difference in hypertension risk was not significant in rural 2011 and urban 2018 samples. Decomposition analysis revealed that the contributions of socio-economic status related inequality of hypertension in 2011 were46.58% and 20.85% for wealth index and BMI, respectively. However, the contributions of wealth index and BMI have shifted to 12.60% and 55.29%, respectively in 2018. Conclusion: The prevalence of hypertension among Bangladeshi adults has increased significantly, and there is no subgroup where it is decreasing. Population-level approaches directed at high risk groups (overweight, obese) should be implemented thoroughly. We underscore prevention strategies by following strong collaboration with stakeholders in the health system of the country to adopt healthy lifestyle choices. © 2021 Chowdhury et al.
Metadata Only
Improving chronic kidney disease detection efficiency: fine tuned catboost and nature-inspired algorithms with explainable AI
(Institute of Electrical and Electronics Engineers Inc., 2025-01-01) Haque, Md Ehsanul; Jahidul Islam S.M.; Maliha, Jeba; Hossan Sumon, Md Shakhauat; Sharmin, Rumana; Rokoni, Sakib; Department of Computer Science and Engineering
Chronic Kidney Disease (CKD) is a major global health issue which is affecting million people around the world and with increasing rate of mortality. Mitigation of progression of CKD and better patient outcomes requires early detection. Nevertheless, limitations lie in traditional diagnostic methods, especially in resource constrained settings. This study proposes an advanced machine learning approach to enhance CKD detection by evaluating four models: Random Forest (RF), Multi-Layer Perceptron (MLP), Logistic Regression (LR), and a fine-tuned CatBoost algorithm. Specifically, among these, the fine-tuned CatBoost model demonstrated the best overall performance having an accuracy of 98.75%, an AUC of 0.9993 and a Kappa score of 97.35% of the studies. The proposed CatBoost model has used a nature inspired algorithm such as Simulated Annealing to select the most important features, Cuckoo Search to adjust outliers and grid search to fine tune its settings in such a way to achieve improved prediction accuracy. Features significance is explained by SHAP-a well-known XAI technique-for gaining transparency in the decision-making process of proposed model and bring up trust in diagnostic systems. Using SHAP, the significant clinical features were identified as specific gravity, serum creatinine, albumin, hemoglobin, and diabetes mellitus. The potential of advanced machine learning techniques in CKD detection is shown in this research, particularly for low income and middle-income healthcare settings where prompt and correct diagnoses are vital. This study seeks to provide a highly accurate, interpretable, and efficient diagnostic tool to add to efforts for early intervention and improved healthcare outcomes for all CKD patients.
Metadata Only
Renewable energy-driven microgrid design for a desalination plant in Noakhali, Bangladesh
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Faiyaz, Md. Abrar Hossen; Saha, Gourab; Suvo, Shabbir Hoshen; Rafid, Sk Tahmed Salim; Department of Electrical and Electronic Engineering
The scarcity of drinking water is a threat to the survival of any living being. The sea is the largest source of water, yet this source remains unutilized due to salinity. Production of drinking water from sea water using reverse osmosis (RO) desalination is viable using modern technology. This paper proposes a solar powered microgrid system for desalination plants with a capacity of 100 cubic meter per day drinking water located in Noakhali. The optimal solution for microgrid design is based on the criteria of energy costs, meeting demand requirements, and limiting excess electricity, utilizing HOMER. The optimal design includes 170 kW of solar panels, meeting the target energy demand for the RO plant. The predicted reduction of around 100 tons of carbon dioxide emissions by the proposed microgrid has a favorable effect on the environment.
Metadata Only
Smart soil monitoring system with crop and fertilizer recommendation features
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Saha, Gourab; Suvo, Shabbir Hoshen; Tonmoy, Farhan Tanjim; Asad, Jiad Bin; Imran, Mohammed Thushar; Azad, Akm Abdul Malek; Department of Electrical and Electronic Engineering
This paper represents the design of a portable and IoT based device for monitoring nitrogen, phosphorus, potassium, pH, and moisture of soil, along with surrounding temperature and humidity, for crops and fertilizer recommendation. The idea and execution of a wireless, portable system for assessing soil quality, recommending the optimal amount of fertilizer, and suggesting appropriate crops depending on the soil conditions are the main goals of this research project. The system is designed to have a user-friendly interface. It uses a cloud-based platform for analyzing data and implementation of machine learning models. Multiple classification-based machine learning algorithms are compared to gain better accuracy of the system. The results of the system are validated using different samples of both indoor and outdoor environments. This cost-effective design, which includes monitoring of seven parameters necessary for crop cultivation along with crop recommendation feature and fertilizer suggestion, makes it desirable for agroindustry.
Open Access
Older adults with non-communicable chronic conditions and their health care access amid COVID-19 pandemic in Bangladesh: Findings from a cross-sectional study
(Public Library of Science, 2021-07-01) Mistry, Sabuj Kanti; Ali, A.R.M. Mehrab; Yadav, Uday Narayan; Ghimire, Saruna; Hossain, Md Belal; Shuvo, Suvasish Das; Saha, Manika; Sarwar, Sneha; Nirob, Md Mohibur Hossain; Sekaran, Varalakshmi Chandra; Harris, Mark F.; BRAC James P Grant School of Public Health
Background Burgeoning burden of non-communicable disease among older adults is one of the emerging public health problems. In the COVID-19 pandemic, health services in low- and middleincome countries, including Bangladesh, have been disrupted. This may have posed challenges for older adults with non-communicable chronic conditions in accessing essential health care services in the current pandemic. The present study aimed at exploring the challenges experienced by older Bangladeshi adults with non-communicable chronic conditions in receiving regular health care services during the COVID-19 pandemic. Materials and methods The study followed a cross-sectional design and was conducted among 1032 Bangladeshi older adults aged 60 years and above during October 2020 through telephone interviews. Self-reported information on nine non-communicable chronic conditions (osteoarthritis, hypertension, heart disease, stroke, hypercholesterolemia, diabetes, chronic respiratory diseases, chronic kidney disease, cancer) was collected. Participants were asked if they faced any difficulties in accessing medicine and receiving routine medical care for their medical conditions during the COVID-19 pandemic. The association between non- communicable chronic conditions and accessing medication and health care was analysed using binary logic regression model. Results Most of the participants aged 60-69 years (77.8%), male (65.5%), married (81.4%), had no formal schooling (58.3%) and resided in rural areas (73.9%). Although more than half of the participants (58.9%) reported having a single condition, nearly one-quarter (22.9%) had multimorbidity. About a quarter of the participants reported difficulties accessing medicine (23%) and receiving routine medical care (27%) during the pandemic, and this was significantly higher among those suffering from multimorbidity. In the adjusted analyses, participants with at least one condition (AOR: 1.95, 95% CI: 1.33-2.85) and with multimorbidity (AOR: 4.75, 95% CI: 3.17-7.10) had a higher likelihood of experiencing difficulties accessing medicine. Similarly, participants with at least one condition (AOR: 3.08, 95% CI: 2.11- 4.89) and with multimorbidity (AOR: 6.34, 95% CI: 4.03-9.05) were significantly more likely to face difficulties receiving routine medical care during the COVID-19 pandemic. Conclusions Our study found that a sizeable proportion of the older adults had difficulties in accessing medicine and receiving routine medical care during the pandemic. The study findings highlight the need to develop an appropriate health care delivery pathway and strategies to maintain essential health services during any emergencies and beyond. We also argue the need to prioritise the health of older adults with non-communicable chronic conditions in the centre of any emergency response plan and policies of Bangladesh. © 2021 Mistry et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Open Access
Growth dynamics among adolescent girls in Bangladesh: Evidence from nationally representative data spanning 2011–2014
(Public Library of Science, 2021-07-01) Adams A.M.; Khan A.; Roy A.S.; Hassan M.T.; Mridha M.K.; Ahmed N.U.; Mustaphi P.; Chowdhury I.; Khondker R.; Hyder Z.; BRAC James P Grant School of Public Health
Background Adolescence is the last opportunity to reverse any growth faltering accumulated from fetal life through childhood and it is considered a crucial period to optimize human development. In Bangladesh, a growing double burden of underweight and obesity in adolescents is recognized, yet limited data exists on how, when, and where to intervene. This study assesses the dynamics of growth among adolescent girls in Bangladesh, providing insight about critical junctures where faltering occurs and where immediate interventions are warranted. Methods We pooled data from Bangladesh’s Food Security and Nutrition Surveillance Project collected between 2011 and 2014 to document the age dynamics of weight and linear growth. 20,572 adolescent girls were measured for height and 19,345 for weight. We constructed growth curves for height, weight, stunting, and underweight. We also stratified growth dynamics by wealth quintile to assess socioeconomic inequities in adolescent trajectories. Results Height-for-age z-score (HAZ) in Bangladeshi girls deteriorates throughout adolescence and especially during the early years. Mean HAZ decreases by 0.20 standard deviations (sd) per year in early adolescence (10–14 years) vs 0.06 sd/year during late adolescence (15–19 years), while stunting increases by 16 percentage points (pp) vs 6.7 pp, respectively. Conversely, BMI-for-age z-score (BAZ) increases by 0.13 sd/year in early adolescence vs 0.02 sd/year in late adolescence, and underweight decreases by 12.8 pp vs 3.2 pp. Adolescent girls in all socioeconomic groups show a similar pattern of HAZ and BAZ dynamics, but the curve for the richest quintile stays above that of the poorest across all ages. Conclusions Trends and levels of stunting and underweight among adolescent girls in Bangladesh are worrisome, suggesting substantial linear growth faltering in early adolescence, with improving weight-for-age occurring only as linear growth slows and stops. Given the rising burden of non-communicable diseases (NCDs) in Bangladesh and emerging evidence of the link between stunting and later chronic diseases, greater attention to adolescent growth and development is needed. Our findings suggest that, to address stunting, interventions in early adolescence would have the greatest benefits. School-based interventions could be a way to target this population. © 2021 Adams et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Open Access
Perceived fear of COVID-19 and its associated factors among Nepalese older adults in eastern Nepal: A cross-sectional study
(Public Library of Science, 2021-07-01) Yadav, Uday Narayan; Yadav, Om Prakash; Singh, Devendra Raj; Ghimire, Saruna; Rayamajhee, Binod; Mistry, Sabuj Kanti; Rawal, Lal Bahadur; Ali, Arm Mehrab; Tamang, Man Kumar; Mehta, Suresh
Coronavirus disease 2019 (COVID-19) has affected all age groups worldwide, but older adults have been affected greatly with an increased risk of severe illness and mortality. Nepal is struggling with the COVID-19 pandemic. The normal life of older adults, one of the vulnerable populations to COVID-19 infection, has been primarily impacted. The current evidence shows that the COVID-19 virus strains are deadly, and non-compliance to standard protocols can have serious consequences, increasing fear among older adults. This study assessed the perceived fear of COVID-19 and associated factors among older adults in eastern Nepal. Methods A cross-sectional study was conducted between July and September 2020 among 847 older adults (?60 years) residing in three districts of eastern Nepal. Perceived fear of COVID-19 was measured using the seven-item Fear of COVID-19 Scale (FCV-19S). Multivariate logistic regression identified the factors associated with COVID-19 fear. Results The mean score of the FCV-19S was 18.1 (SD = 5.2), and a sizeable proportion of older adults, ranging between 12%-34%, agreed with the seven items of the fear scale. Increasing age, Dalit ethnicity, remoteness to the health facility, and being concerned or overwhelmed with the COVID-19 were associated with greater fear of COVID-19. In contrast, preexisting health conditions were inversely associated with fear. Conclusion Greater fear of the COVID-19 among the older adults in eastern Nepal suggests that during unprecedented times such as the current pandemic, the psychological needs of older adults should be prioritized. Establishing and integrating community-level mental health support as a part of the COVID-19 preparedness and response plan might help to combat COVID-19 fear among them. Copyright: ©2021 Yadav et al.
Open Access
Environmental prevalence of toxigenic Vibrio cholerae O1 in Bangladesh coincides with V. cholerae non-O1 non-O139 genetic variants which overproduce autoinducer-2
(Public Library of Science, 2021-07-01) Naser, Iftekhar Bin; Shishir, Tushar Ahmed; Faruque, Shah Nayeem; Mozammel Hoque M.; Hasan, Anamul; Faruque, Shah M.; Department of Mathematics and Natural Sciences
Prevalence of toxigenic Vibrio cholerae O1 in aquatic reservoirs in Bangladesh apparently increases coinciding with the occurrence of seasonal cholera epidemics. In between epidemics, these bacteria persist in water mostly as dormant cells, known as viable but nonculturable cells (VBNC), or conditionally viable environmental cells (CVEC), that fail to grow in routine culture. CVEC resuscitate to active cells when enriched in culture medium supplemented with quorum sensing autoinducers CAI-1 or AI-2 which are signal molecules that regulate gene expression dependent on cell density. V. cholerae O1 mutant strains with inactivated cqsS gene encoding the CAI-1 receptor has been shown to overproduce AI-2 that enhance CVEC resuscitation in water samples. Since V. cholerae non-O1 non-O139 (non-cholera-vibrios) are abundant in aquatic ecosystems, we identified and characterized naturally occurring variant strains of V. cholerae non-O1 non-O139 which overproduce AI-2, and monitored their co-occurrence with V. cholerae O1 in water samples. The nucleotide sequence and predicted protein products of the cqsS gene carried by AI-2 overproducing variant strains showed divergence from that of typical V. cholerae O1 or non-O1 strains, and their culture supernatants enhanced resuscitation of CVEC in water samples. Furthermore, prevalence of V. cholerae O1 in the aquatic environment was found to coincide with an increase in AI-2 overproducing non-O1 non-O139 strains. These results suggest a possible role of non-cholera vibrios in the environmental biology of the cholera pathogen, in which non-O1 non-O139 variant strains overproducing AI-2 presumably contribute in resuscitation of the latent pathogen, leading to seasonal cholera epidemics. Importance. Toxigenic Vibrio cholerae which causes seasonal epidemics of cholera persists in aquatic reservoirs in endemic areas. The bacteria mostly exist in a dormant state during inter-epidemic periods, but periodically resuscitate to the active form. The resuscitation is enhanced by signal molecules called autoinducers (AIs). Toxigenic V. cholerae can be recovered from water samples that normally test negative for the organism in conventional culture, by supplementing the culture medium with exogenous AIs. V. cholerae belonging to the non-O1 non-O139 serogroups which do not cause cholera are also abundant in natural waters, and they are capable of producing AIs. In this study we characterized V. cholerae non-O1 non-O139 variant strains which overproduce an autoinducer called AI-2, and found that the abundance of the cholera pathogen in aquatic reservoirs correlates with an increase in the AI-2 overproducing strains. Our results suggest a probable role of these variant strains in the environmental biology and epidemiology of toxigenic V. cholerae, and may lead to novel means for surveillance, prevention and control of cholera. © 2021 Naser et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Open Access
Are older adults of Rohingya community (Forcibly Displaced Myanmar Nationals or FDMNs) in Bangladesh fearful of COVID-19? Findings from a cross-sectional study
(Public Library of Science, 2021-06-01) Mistry, Sabuj Kanti; Ali, A.R.M. Mehrab; Akther, Farhana; Peprah, Prince; Reza, Sompa; Prova, Shaidatonnisha; Yadav, Uday Narayan; BRAC James P Grant School of Public Health
Aim: This study aimed to assess the fear of COVID-19 and its associates among older Rohingya (Forcibly Displaced Myanmar Nationals or FDMNs) in Bangladesh. Method: We conducted a cross-sectional survey among 416 older FDMNs aged 60 years and above living in camps of Cox's Bazar, Bangladesh. A semi-structured questionnaire was used to collect information on participants' socio-demographic and lifestyle characteristics, preexisting non-communicable chronic conditions, and COVID-19 related information. Level of fear was measured using the seven-item Fear of COVID-19 Scale (FCV-19S) with the cumulative score ranged from 7 to 35. A multiple linear regression examined the factors associated with fear. Results: Among 416 participants aged 60 years or above, the mean fear score was 14.8 (range 8-28) and 88.9% of the participants had low fear score. Participants who were concerned about COVID-19 (β: 0.63, 95% CI: -0.26 to 1.53) and overwhelmed by COVID-19 (β: 3.54, 95% CI: 2.54 to 4.55) were significantly more likely to be fearful of COVID-19. Other factors significantly associated with higher level of fear were lesser frequency of communication during COVID-19, difficulty in obtaining food during COVID-19, perception that older adults are at highest risk of COVID-19 and receiving COVID-19 related information from Radio/ television and friends/family/neighbours. Conclusions: Our study highlighted that currently there little fear of COVID-19 among the older Rohingya FDMNs. This is probably due to lack of awareness of the severity of the disease in. Dissemination of public health information relevant to COVID-19 and provision of mental health services should be intensified particularly focusing on the individual who were concerned, overwhelmed or fearful of COVID-19. However, further qualitative research is advised to find out the reasons behind this. © 2021 Mistry et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Metadata Only
Understanding facial expression of children with autism using learning theory
(Institute of Electrical and Electronics Engineers Inc., 2023-01-01) Zaman, Anushka; Khan, Evea Zerin; Rabiul Alam, Md. Golam; Chakrabarty, Amitabha; Department of Computer Science and Engineering
It is challenging for autistic individuals to gather socially and empathically, as we understand, so this subject matter differs from the identification of a common human facial expression. In our research, we have taken pictures of children and teenagers from school of special children based in Bangladesh, aged between 5 to 20 with Autism Spectrum Disorder (ASD) and identified their emotions from their images. Our objective is to find a technique that will allow their emotions to be accurately perceived and thus, make it easy for them including other individuals around them to interact socially without any barriers. In our paper, we implemented VGG16 and six Machine Learning Algorithms along with one Feature Extraction technique to detect Facial Expression of children with autism. The proposed system has showcased an accuracy of 75% for VGG16, 68% for Random Forest Classifier, 61% for Random Forest Classifier along with PCA, 61% for Support Vector Machine (SVM) and 67% for SVM with PCA, 56% for Logistic Regression and 53% for Logistic Regression along with PCA, 52% for Linear Discriminant Analysis (LDA) and 54% for LDA with PCA, 46% for Decision Tree and 50% for Decision Tree along with PCA and 46% for Gaussian Naïve Bayes and 38% for Gaussian Naïve Bayes along with PCA.
Open Access
Pandemic catch-22: The role of mobility restrictions and institutional inequalities in halting the spread of COVID-19
(Public Library of Science, 2021-06-01) Fakir, Adnan M.S.; Bharati, Tushar; Department of Economics and Social Sciences
Countries across the world responded to the COVID-19 pandemic with what might well be the set of biggest state-led mobility and activity restrictions in the history of humankind. But how effective were these measures across countries? Compared to multiple recent studies that document an association between such restrictions and the control of the contagion, we use an instrumental variable approach to estimate the causal effect of these restrictions on mobility, and the growth rate of confirmed cases and deaths during the first wave of the pandemic. Using the level of stringency in the rest of the world to predict the level of stringency of the restriction measures in a country, we show while stricter contemporaneous measures affected mobility, stringency in seven to fourteen days mattered most for containing the contagion. Heterogeneity analysis, by various institutional inequalities, reveals that even though the restrictions reduced mobility more in relatively less-developed countries, the causal effect of a reduction in mobility was higher in more developed countries. We propose several explanations. Our results highlight the need to complement mobility and activity restrictions with other health and information measures, especially in less-developed countries, to combat the COVID-19 pandemic effectively. Copyright: © 2021 Fakir, Bharati. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Metadata Only
Gender classification in Bangla language using deep learning-based voice analysis
(Institute of Electrical and Electronics Engineers Inc., 2023-01-01) Labib, Md. Zarif; Monsur, Sayema Binte; Shuvo, Abtahi Maskawath ; Azrine, Tasmia; Hakim, Talukder Juhaer; Ali, Syed Muaz; Alam, Md. Ashraful; Department of Computer Science and Engineering
Gender classification based on voice analysis is one of the essential tasks in speech and audio processing, with various applications such as speech recognition systems, voice assistants, call center analytics, Etc. For speech synthesis, human-computer interaction, and speaker identification - gender classification plays a vital role. Although extensive research on this topic has been done in various languages, studies can hardly be found regarding gender classification in the Bangla language. Our research aims to recognize gender in the Bangla language using deep learning approaches and voice analysis. The proposed strategy in this study consists of three stages: i) Pre-processing of the data; ii) Feature extraction utilizing the Short-Time Fourier Transforms (STFT) and Mel-Frequency Cepstral Coefficients (MFCC); iii) Classification using Convolutional Neural Network (CNN) models such as ResNet50, EfficientNetB0, InceptionV3, and DenseNet-121. Notably, 12 distinct feature combinations are used for model training and testing, using both the MFCC and STFT features singly or in combination. After thorough training and testing, InceptionV3 and EfficientNetB0 CNN models with MFCC features as input resulted in the highest accuracy of 92%, which demonstrates the system's excellent accuracy rate and its potential for use in practical settings.
Open Access
Ten-years cardiovascular risk among Bangladeshi population using non-laboratorybased risk chart of the World Health Organization: Findings from a nationally representative survey
(Public Library of Science, 2021-05-01) Hanif, Abu Abdullah Mohammad; Hasan, Mehedi; Khan, Showkat Ali; Hossain, Mokbul; Shamim, Abu Ahmed; Hossaine, Moyazzam; Ullah, Mohammad Aman; Sarker, Samir Kanti; Rahman, S. M Mustafizur; Bulbul, Md Mofijul Islam; Mitra, Dipak Kumar; Mridha, Malay Kanti; BRAC James P Grant School of Public Health
The World Health Organization (WHO) has recently developed a non-laboratory based cardiovascular disease (CVD) risk chart considering the parameters age, sex, current smoking status, systolic blood pressure, and body mass index. Using the chart, we estimated the 10-years CVD risk among the Bangladeshi population aged 40-74 years. We analyzed data from a nationally representative survey conducted in 2018-19. The survey enrolled participants from 82 clusters (57 rural, 15 non-slum urban, and 10 slums) selected by multistage cluster sampling. Using the non-laboratory-based CVD risk chart of the World Health Organization (WHO), we categorized the participants into 5 risk groups: Very low (<5%), low (5% to <10%), moderate (10% to <20%), high (20% to <30%) and very high (> = 30%) risk. We performed descriptive analyses to report the distribution of CVD risk and carried out univariable and multivariable logistic regression to identify factors associated with elevated CVD risk (> = 10% CVD risk). Of the 7,381 participants, 46.0% were female. The median age (IQR) was 59.0 (48.0-64.7) years. Overall, the prevalence of very low, low, moderate, high, and very high CVD risk was 34.7%, 37.8%, 25.9%, 1.6%, and 0.1%, respectively. Elevated CVD risk (> = 10%) was associated with poor education, currently unmarried, insufficient physical inactivity, smokeless tobacco use, and self-reported diabetes in both sexes, higher household income, and higher sedentary time among males, and slum-dwelling and non-Muslim religions among females. One in every four Bangladeshi adults had elevated levels of CVD risk, and males are at higher risk of occurring CVD events. Non-laboratory-based risk prediction charts can be effectively used in low resource settings. The government of Bangladesh and other developing countries should train the primary health care workers on the use of WHO non-laboratory-based CVD risk charts, especially in settings where laboratory tests are not available. © 2021 Hanif et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Metadata Only
Automated sentiment analysis for web-based stock and cryptocurrency news summarization with transformer-based models
(Institute of Electrical and Electronics Engineers Inc., 2023-01-01) Hasan, Mehedi; Rahman M.T.; Alavee, Kazi Ahnaf; Zillanee, Abu Hasnayen; Uddin J.; Alam M.G.R.; Department of Computer Science and Engineering
In the fast-paced realm of global financial markets, characterized by rapid trading of both stocks and cryptocurren-cies, it has become essential to grasp the influence of sentiment on market dynamics. With more than 630,000 publicly traded companies worldwide and major stock exchanges like the NYSE handling a substantial portion of global equity transactions, the inherent volatility of the stock market is well-established. Over the past decade, various factors have contributed to the consistent fluctuations in stock prices. One key factor is the influence of investor reviews sourced from diverse news outlets and social media platforms such as Twitter. Understanding how these reviews can be collected and effectively summarized is crucial. This paper centers on the intricate field of market sentiment analysis and its profound impact on user sentiment, subsequently affecting price fluctuations in both stocks and cryptocurrencies. In this study, we present a comprehensive exploration of the development and evaluation of an automated sentiment analysis system tailored for summarizing web-based news related to stocks and cryptocurrencies.We have implemented BERT (Bidirectional Encoder Representations from Transformers) in combination with NLTK for text summarization, a highly accurate model with a performance level of 95.84%, as part of our proposed approach.
Metadata Only
Detection of deepfake videos using computer vision and deep learning
(Institute of Electrical and Electronics Engineers Inc., 2023-01-01) Rahman, Anisur; Rahman, Faria; Labib, Tahmidul; Uschash, Ehteshamul Islam; Chowdhury Adiba, Shihaba Jamal; Karim, Dewan Ziaul; Department of Computer Science and Engineering
DeepFakes are one of the most alarming concepts in this era of Metaverse and technological advancement. DeepFakes are artificially-generated manipulated photos or videos using Deep learning, Generated Adversarial Network (GAN), autoencoder-decoder pairing structure etc. There are several other Deepfaking tools such as; FaceSwap, DeepFace-Lab, DFaker, DeepFake-tensorflow etc. DeepFakes can become concerning if it is used for political purpose, committing fraud, spreading misinformation, pornography, defamation on social media etc. As a result, it is obvious that DeepFakes can be very distressing on the wrong hand if not detected properly. To address this issue, our research aims to develop effective methods for DeepFake video detection, focusing on deep learning approaches, and computer vision techniques. We deployed a dataset consisting of both real and fake videos, obtained from DeepFake Detection Challenge (DFDC) and FaceForensics++. To detect the fake videos, we followed the method of employing temporal feature and exploring visual artifacts within frames. Employing temporal feature uses LSTM and CNN whereas visual artifacts within frames mostly employs deep learning method to detect DeepFakes. We ensembled LSTM and CNN to detect DeepFakes successfully. ResNeXt101-32x8d have been used to extract features and a custom CNN model is added with LSTM for better accuracy for detecting DeepFake. Our ensemble model, which combines LSTM and CNN, successfully detects Deepfakes with an accuracy of 94.05%. Through further improvements and the implementation of learning rate schedulers, such as CosineAnnealingLR, CyclicLR, MultiStepLR, and ReduceLRonPlateau, we achieved even higher accuracy. Among these schedulers, MultiStepLR demonstrated the highest accuracy of 95.33%.
Metadata Only
An efficient deep learning approach to detect neurodegenerative diseases using retinal images
(Institute of Electrical and Electronics Engineers Inc., 2023-01-01) Irfanuddin, Chowdhury Mohammad; Shafin, Wasique Islam; Ahmed, Koushik; Khan, Md. Hasib; Ashraful Alam M.; Rahman, Rafeed; Dipto, Shakib Mahmud; Department of Computer Science and Engineering
The leading cause of unseasonable death worldwide is heart complaints. Day by day the causes of heart disease are increasing at a rapid-fire rate and it's veritably important and concerning to prognosticate any such disease beforehand. Predicting how illness will affect a person is a delicate challenge. Machine learning is being applied in different fields around the world. In the healthcare sector, there is no exception. Data classification models and machine learning algorithms stoutly introduce diagnostic guidelines and enable experts to increase the effectiveness of the diagnostic process. The body's remaining organs are given advanced precedence over the nucleus. It provides the body with oxygen. The distribution of heart illnesses among medical practitioners can be estimated via data exploration. Medical facilities can examine various disorders and evaluate emerging diseases thanks to data collecting. Predicting the condition based on recent medical research has the biggest impact. To learn statistics more effectively, a variety of methods are being investigated in the scientific community. The renovation has had a big impact on the metropolitan community's way of life in addition to improving it. In this situation, it's crucial to offer a comprehensive tool that will enable medical professionals to foresee the sickness. Different machine learning applications suggest varying prediction precision. It should be analyzed with Logistic Regression, KNN, Decision tree, Random Forest, SVM, Gaussian NB, Ada Boost Classifier Gradient Boosting Classifier, Quadratic Discriminant Analysis, and MLP Classifier with comparative mean of three data sets. It will be better to investigate the accuracy of prediction from the most concerning algorithms of Machine learning with recall and f-score of the heart data and present it in the table with visual representation. This underpinning research has shown that multilayer perceptron with cross-validation has surpassed all other algorithms in terms of accuracy, which is the conclusion that can be derived from it. The best accuracy is attained with 97%.
Open Access
Development and performance evaluation of a rapid in-house ELISA for retrospective serosurveillance of SARS-CoV-2
(Public Library of Science, 2021-02-01) Sil, Bijon Kumar; Jahan, Nowshin; Haq, Md Ahsanul; Oishee, Mumtarin Jannat; Ali, Tamanna; Khandker, Shahad Saif; Kobatake, Eiry; Mie, Masayasu; Khondoker, Mohib Ullah; Jamiruddin, Mohd Raeed; Adnan, Nihad; School of Pharmacy
Background In the ongoing pandemic situation of COVID-19, serological tests can complement the molecular diagnostic methods, and can be one of the important tools of sero-surveillance and vaccine evaluation. Aim To develop and evaluate a rapid SARS-CoV-2 specific ELISA for detection of anti-SARSCoV2 IgG from patients’ biological samples. Methods In order to develop this ELISA, three panels of samples (n = 184) have been used: panel 1 (n = 19) and panel 2 (n = 60) were collected from RT-PCR positive patients within 14 and after 14 days of onset of clinical symptoms, respectively; whereas panel 3 consisted of negative samples (n = 105) collected either from healthy donors or pre-pandemic dengue patients. As a capturing agent full-length SARS-CoV2 specific recombinant nucleocapsid was immobilized. Commercial SARS-CoV2 IgG kit based on chemiluminescent assay was used for the selection of samples and optimization of the assay. The threshold cut-off point, inter-assay and intra-assay variations were determined. Results The incubation/reaction time was set at a total of 30 minutes with the sensitivity of 84% (95% confidence interval, CI, 60.4%, 96.6%) and 98% (95% CI, 91.1%, 100.0%), for panel 1 and 2, respectively; with overall 94.9% sensitivity (95% CI 87.5%, 98.6%). Moreover, the clinical specificity was 97.1% (95% CI, 91.9%, 99.4%) with no cross reaction with dengue samples. The overall positive and negative predictive values are 96.2% (95% CI 89.2%, 99.2%) and 96.2% (95% CI, 90.6% 99.0%), respectively. In-house ELISA demonstrated 100% positive and negative percent agreement with Elecsys Anti-SARS-CoV-2, with Cohen’s kappa value of 1.00 (very strong agreement), while comparing 13 positive and 17 negative confirmed cases. Conclusion The assay is rapid and can be applied as one of the early and retrospective sero-monitoring tools in all over the affected areas. Copyright: © 2021 Sil et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Browse

Recent Submissions