Browsing by Author "Islam, Md Saiful"

Now showing 1 - 18 of 18

Assessing metal(loid)s-Induced long-term spatiotemporal health risks in Coastal Regions, Bay of Bengal: A chemometric study
(Scopus, 2024) Aktar, Shammi; Islam, Abu Reza Md. Towfiqul; Mia, Md Yousuf; Jannat, Jannatun Nahar; Islam, Md Saiful; Masud, Md Abdullah Al; Idris, Abubakr M.; Pal, Subodh Chandra; Senapathi, Venkatramanan
Despite sporadic and irregular studies on heavy metal(loid)s health risks in water, fish, and soil in the coastal areas of the Bay of Bengal, no chemometric approaches have been applied to assess the human health risks comprehensively. This review aims to employ chemometric analysis to evaluate the long-term spatiotemporal health risks of metal(loid)s e.g., Fe, Mn, Zn, Cd, As, Cr, Pb, Cu, and Ni in coastal water, fish, and soils from 2003 to 2023. Across coastal parts, studies on metal(loid)s were distributed with 40% in the southeast, 28% in the south-central, and 32% in the southwest regions. The southeastern area exhibited the highest contamination levels, primarily due to elevated Zn content (156.8 to 147.2 mg/L for Mn in water, 15.3 to 13.2 mg/kg for Cu in fish, and 50.6 to 46.4 mg/kg for Ni in soil), except for a few sites in the south-central region. Health risks associated with the ingestion of Fe, As, and Cd (water), Ni, Cr, and Pb (fish), and Cd, Cr, and Pb (soil) were identified, with non-carcinogenic risks existing exclusively through this route. Moreover, As, Cr, and Ni pose cancer risks for adults and children via ingestion in the southeastern region. Overall non-carcinogenic risks emphasized a significantly higher risk for children compared to adults, with six, two-, and six-times higher health risks through ingestion of water, fish, and soils along the southeastern coast. The study offers innovative sustainable management strategies and remediation policies aimed at reducing metal(loid)s contamination in various environmental media along coastal Bangladesh.
BnText2Table – dataset and Text-to-Table generation in Bangla
(BRAC University, 2024-01) Zariyat, Tahreema Rahman; Ahmed, Fahim Irfan; Oishi, Tahsina Tajrim; Morshed, Maruf; Islam, Md Saiful
"In this fast-paced world, everyone relies on technology to get their work done quickly and efficiently, since using technology greatly simplifies every task that needs to be done. The majority of the publications are lengthy and packed with crucial data. However, in many instances, extra words are also added to boost the word count, which causes a number of difficulties when trying to get the desired information. For the English language, numerous tools are available to summarize the text and present it in tabular form. However, it is not the same for our mother tongue, Bangla. Despite being the 5th most-spoken native language in the world, there is no tool available to ease the workload in Bengali language. Our research will assist in such circumstances by summarizing the given information in tabular form within the shortest possible time. Since there is no dataset available that will be suitable for our research, we have prepared the dataset ourselves. Then, we have used the mBART-50-large, mT5-base, mT5-m2m-CrossSum and BanglaT5 models for the implementation. Finding the appropriate table headers in light of the context and order of the data is the most important task in this study. To sum up, our main goal is to develop a benchmark dataset for a text-to-table model for the betterment of the NLP research community."
Detecting students’ state of mind through facial expression in real time educational environment
(Daffodil International University, 2018-11-08) Ahmed, Sazzad; Hossain, S.M. Shabriar; Islam, Md Saiful
Monitoring a student is tough when you only have one hour and thirty hour of class time. So this project can help a teacher or the varsity authority to monitor a student individually. Here we tried to find out the state mind of a student. When a student is in the classroom it’s very tough to find out and understand, if a student is happy or confused in the class. It’s also can give an indication that student is attentive or not. With our system which will detect Students state of mind through facial expression it will be easy to find out those who need counseling. So this system can create a new diminution to this educational environment.
Effects of Microplastic and Heavy Metals on Coral Reefs
(Elsevier, 2023-11-19) Islam, Md Saiful; Islam, Abu Reza Md Towfiqul; Ismail, Zulhilmi; Ahmed, Md Kawser; Ali, Mir Mohammad; Kabir, Md Humayun; Ibrahim, Khalid A.; Al-Qthanin, Rahmah N.; Idris, Abubakr M.
In the modern world, plastic trash has been recognized as a global issue, and studies on microplastics (MPs) in the marine and inland environments have previously been conducted. Marine ecosystems act as a bio-diverse ecosystem where coral reefs contribute to make a sound living of the coastal people by gathering natural resources. The current study indicates that MPs and heavy metals (HMs) accumulation to biofilm and organic matter through sedimentation, precipitation, adsorption, and desorption that may have potential effect on growth and development of coral reefs in the marine ecosystems. However, the knowledge of distribution, impact, mechanism, degradation, and association mechanisms between MPs and HMs in the natural environment may open a new window for conducting analytical research from an ecological viewpoint. The current study thus summarizes the types of marine samples with the analytical techniques, polymers of MPs, and their impact on corals and other marine biota. This study also identifies existing knowledge gaps and recommends fresh lines of inquiry in light of recent developments in MPs and HMs research on the marine ecosystems. Overall, the present study suggests a sustainable intervention for reducing MPs and HMs from the marine ecosystems by demonstrating their existence in water, sediment, fish, corals, and other biota, and their impending ecotoxicological impacts on the environment and human health. The impacts of MPs and HMs on coral reefs are critically assessed in this study in light of the most recent scientific knowledge, existing laws, and new suggestions to minimize their contamination in the marine ecosystems.
Encryption Based Image Watermarking Algorithm in 2DWT-DCT Domains
(Sensors, 2021-08-17) Hasan, Nayeem; Islam, Md Saiful; Chen, Wenyu; Kabir, Muhammad Ashad; Al-Ahmadi, Saad
In this work, we examine the privacy and safety issues of Internet of Things (IoT)-based Precision Agriculture (PA), which could lead to the problem that industry is currently experiencing as a result of Intellectual Property Theft (IPT). Increasing IoT-based information flow in PA will make a system less secure if a proper security mechanism is not ensured. Shortly, IoT will transform everyday lives with its applications. Intellectual Property (IP) is another important concept of an intelligent farming system. If the IP of a wise farming system leaks, it damages all intellectual ideas like cultivation patterns, plant variety rights, and IoT generated information of IoT-based PA. Thus, we proposed an IoT enabled SDN gateway regulatory system that ensures control of a foreign device without having access to sensitive farm information. Most of the farm uses its devices without the use of its integrated management and memory unit. An SDN-based structure to solve IP theft in precision farming has been proposed. In our proposed concept, a control system integrates with the cloud server, which is called the control hub. This hub will carry out the overall PA monitoring system. By hiring the farm devices in the agricultural system, these devices must be tailored according to our systems. Therefore, our proposed PA is a management system for all controllable inputs. The overall goal is to increase the probability of profit and reduce the likelihood of IPT. It does not only give more information but also improves information securely by enhancing the overall performance of PA. Our proposed PA architecture has been measured based on the throughput, round trip time, jitter, packet error rate, and the cumulative distribution function. Our achieved results reduced around (1.66–6.46)% compared to the previous research. In the future, block chain will be integrated with this proposed architecture for further implementation.
End to end Bangla handwritten and scene text detection using convolutional neural network
(BRAC University, 8/21/2017) Mahal, Somania Nur; Abir, B M; Bakhtiar, Fahim; Chakrabarty, Amitabha; Islam, Md Saiful
Handwritten text detection from a natural image has a large set of difficulties. A systematic approach that can automatically recognise text from handwriting, printed books, road signs and also classifies text and nontext blocks from natural image has many significant applications. For instance, visual assistance for visually impaired people, image understanding, classification of text in image, implementing autonomous navigation system. Recent development of deep learning approach has strong capabilities to extract high level feature from a kernel(patch) of an Image. In this thesis we will demonstrate an alternate approach that integrates a multilayer convolutional neural network (CNN) with supervised feature learning .This approach allows a higher recall rate for the text in an image and thus increases the overall performances of the system. And we have used these methodologies to create a learning model using synthetic and real-world data that is capable to process bangla and english handwritten and scene text in natural image.
Exploring the non-Political factors behind young voter enthusiasm: A machine learning approach
(BRAC University, 2025-02) Hossain, Md Israk; Subah, Raya; Utsho, Mashrur Ahmed; Islam, Md Saiful; Hasan Shoumo, Syed Zamil
To build a democratic nation, a voting system provides the foundation and it represents the fundamental rights of citizens to voice their decisions. It also represents the responsibility of citizens for shaping their government. The participation of young voters is important as they form a significant portion of a country and are able to bring new perspectives regarding decision-making or policies revolving around a country. However, a lot of young people refrain from voting due to political violence, thinking it will not make a difference or due to various other reasons. They can be engaged through awareness on social media, political campaigns or discussions in classrooms. Their involvement can increase voter enthusiasm which is pivotal in determining higher voter turnout. With the rise of Artificial Intelligence, the effort of people has decreased significantly in the extraction of data and finding meaningful insights. Hence, we will utilize machine learning to make future decisions based on the primary data we have collected. The aim of this research is to propose a multi-modal agent that can help to infer voter turnout and understand the factors that influence the behavior of our youths when participating in voting using artificial intelligence. We aim to focus only on the non-political factors that affect the voting behavior of young people. We have carried out a survey on the students of BRAC university who were asked to fill out a questionnaire containing both multiple choice questions and opinion based questions. The answers to the multiple choice questions will be processed as tabular data and fed to an artificial neural network for inference. Similarly, the answers to the opinion based questions will be fed to an extreme gradient boosting model for sentiment analysis. The true label for both levels of inference will be whether a person would participate in voting or not. The proposed multi-modal agent will concatenate the outputs from the artificial neural network and the extreme gradient boosting model and provide a final level of prediction. Moreover, to assess the predictions of the artificial neural network, XAI models such as SHAP and LIME will be used to produce global and local level explanations. A further analysis of these explanations will be made to understand which features are affecting our model. Therefore, the proposed model can help us to gauge young voter turnout and shed a light into the influencing factors that steer their decisions.
Fuzzy Logic, Geostatistics, and Multiple Linear Models To Evaluate Irrigation Metrics and Their Influencing Factors in a Drought-Prone Agricultural Region
(Springer, 2023-10-01) Zihad, S.M. Rabbi Al; Islam, Abu Reza Md Towfiqul; Siddique, Md Abu Bakar; Mia, Md Yousuf; Islam, Md Saiful; Islam, Md Aminul; Bari, A.B.M. Mainul; Bodrud-Doza, Md.; Yakout, Sobhy M.; Senapathi, Venkatramanan; Chatterjee, Sumanta
The quality of water used for irrigation is one of the major threats to maintaining the long-term sustainability of agricultural practices. Although some studies have addressed the suitability of irrigation water in different parts of Bangladesh, the irrigation water quality in the drought-prone region has yet to be thoroughly studied using integrated novel approaches. This study aims to assess the suitability of irrigation water in the drought-prone agricultural region of Bangladesh using traditional irrigation metrics such as sodium percentage (NA%), magnesium adsorption ratio (MAR), Kelley's ratio (KR), sodium adsorption ratio (SAR), total hardness (TH), permeability index (PI), and soluble sodium percentage (SSP), along with novel irrigation indices such as irrigation water quality index (IWQI) and fuzzy irrigation water quality index (FIWQI). Thirty-eight water samples were taken from tube wells, river systems, streamlets, and canals in agricultural areas, then analyzed for cations and anions. The multiple linear regression model predicted that SAR (0.66), KR (0.74), and PI (0.84) were the primary important elements influencing electrical conductivity (EC). Based on the IWQI, all water samples fall into the “suitable” category for irrigation. The FIWQI suggests that 75% of the groundwater and 100% of the surface water samples are excellent for irrigation. The semivariogram model indicates that most irrigation metrics have moderate to low spatial dependence, suggesting strong agricultural and rural influence. Redundancy analysis shows that Na+, Ca2+, Cl−, K+, and HCO3− in water increase with decreasing temperature. Surface water and some groundwater in the southwestern and southeastern parts are suitable for irrigation. The northern and central parts are less suitable for agriculture because of elevated K+ and Mg2+ levels. This study determines irrigation metrics for regional water management and pinpoints suitable areas in the drought-prone region, which provides a comprehensive understanding of sustainable water management and actionable steps for stakeholders and decision-makers.
Hate Speech Detection in the Bengali Language
(Daffodil International University, 2022-06-20) Romim, Nauros; Ahmed, Mosahed; Talukder, Hriteshwar; Islam, Md Saiful
Hate speech is a common problem in the current time of social media and the internet as it is very easy to be in touch with everything through the internet and social media. Hate speech detection research is not very rare but in terms of Bengali language there are very few works related to hate speech in Bengali language. The proposed research experiment has developed a machine learning based project to detect hate speech from Bengali language data or comments, posts in social media that are in Bengali language. This research work has used 3006 pure Bengali data from social media pages (such as Facebook, YouTube) groups, comment sections of news portals. Further, this research work has categorized them in 0 for non-Hate-Speech and 1 for Hate-Speech to classify the data between non-abusive and abusive data. This research work has used several algorithms to find the best possible result in order to determine whether the sentence is abusive or non-abusive such as Logistic Regression, Naive Bayes, Random Forest, Support Vector Machine, K Nearest Neighbor Classifier. From these algorithms, the best result for detecting non-abusive data is the Random Forest [RF] algorithm, which is 67%. © 2022 IEEE.
Heart Disease Prediction Based on External Factors
(International Journal of Advanced Computer Science and Applications, 2019) Tamal, Maruf Ahmed; Islam, Md Saiful; Ahmmed, Md Jisan; Aziz, Md. Abdul; Miah, Pabel; Rezaul, Karim Mohammed
Technology has immensely changed the world over the last decade. As a consequence, the life of the people is undergoing multiple changes that directly have positive and negative effects on health. Less physical activity and a lot of virtual involvements are pushing people into various health-related issues and heart disease is one of them. Currently, it has gained a great deal of attention among various life-threatening diseases. Heart disease can be detected or diagnosed by different medical tests by considering various internal factors. However, this type of approach is not only time-consuming but also expensive. Concurrently, there are very few studies conducted on heart disease prediction based on external factors. To bridge this gap, we proposed a heart disease prediction model based on the machine learning approach which enables predicting heart disease with 95% accuracy. To acquire the best result, 6 distinct machine learning classifiers (Decision Tree, Random Forest, Naive Bayes, Support Vector Machine, Quadratic Discriminant, and Logistic Regression) were used. At the same time, sklearn.ensemble. Extra Trees Classifier has been used to extract relevant features to improve predictive accuracy and control over-fitting. Findings reveal that Support Vector Machine (SVM) outperforms the others with greater accuracy (95%)
Hospitalization and Mortality by Vaccination Status among COVID-19 Patients Aged ≥ 25 Years in Bangladesh Results From a Multicenter Cross-Sectional Study
(Scopus, 22-11-23) Rahman, Md Saydur; Harun, Md Golam Dostogir; Sumon, Shariful Amin; Mohona, Tahrima Mohsin; Abdullah, Syed Abul Hassan Md; Khan, Md Nazuml Huda; Gazi, Md Ismail; Islam, Md Saiful; Anwar, Md Mahabub Ul
The COVID-19 pandemic has inflicted a massive disease burden globally, involving 623 million confirmed cases with 6.55 million deaths, and in Bangladesh, over 2.02 million clinically confirmed cases of COVID-19, with 29,371 deaths, have been reported. Evidence showed that vaccines significantly reduced infection, severity, and mortality across a wide age range of populations. This study investigated the hospitalization and mortality by vaccination status among COVID-19 patients in Bangladesh and identified the vaccine's effectiveness against severe outcomes in real-world settings. Between August and December 2021, we conducted this cross-sectional survey among 783 RT-PCR-confirmed COVID-19 hospitalized patients admitted to three dedicated COVID-19 hospitals in Bangladesh. The study used a semi-structured questionnaire to collect information. We reviewed the patient's records and gathered COVID-19 immunization status from the study participants or their caregivers. Patients with incomplete or partial data from the record were excluded from enrollment. Logistic regression analyses were performed to determine the association between key variables with a patient's vaccination status and mortality. The study revealed that overall hospitalization, severity, and morality were significantly high among unvaccinated study participants. Only one-fourth (25%) of hospitalized patients were found COVID-19 vaccinated. Morality among unvaccinated COVID-19 study participants was significantly higher (AOR: 7.17) than the vaccinated (11.17% vs. 1.53%). Severity was found to be seven times higher among unvaccinated patients. Vaccination coverage was higher in urban areas (29.8%) compared to rural parts (20.8%), and vaccine uptake was lower among female study participants (22.7%) than male (27.6%). The study highlighted the importance of COVID-19 vaccines in reducing mortality, hospitalization, and other severe consequences. We found a gap in vaccination coverage between urban and rural settings. The findings would encourage the entire population toward immunization and aid the policymakers in the ground reality so that more initiatives are taken to improve vaccination coverage among the pocket population.
MoS2 thin film hetero-interface as effective back surface field in CZTS-based solar cells
(Elsevier, 2024-11-16) Islam, Md Saiful; Doroody, Camellia; Sieh Kiong, Tiong; Rahman, Kazi Sajedur; Mahmood Zuhdi, Ahmad Wafi; Yap, Boon Kar; Alam, Mohammad Nur-E; Amin, Nowshad
In this review article, we explore the insertion possibility of molybdenum disulfide (MoS2) thin-film heterostructures into copper, zinc, and tin sulfide (CZTS) based thin film solar cells for improved performance. CZTS has gained prominence as a naturally occurring, non-toxic alternative to conventional solar energy system materials, necessitating a focus on the study of integrating MoS2 (as a back contact) with thin-film solar cells with CZTS integration, as well as understanding the impact on device efficiency and stability to advance, upscale, and commercialize products. By analyzing the MoS2-CZTS interface, critical insights into MoS2's functioning in optimizing charge carrier dynamics, lowering recombination losses, and enhancing overall device performance have been framed in this work. Furthermore, the necessity of optimizing process parameters and characterizing MoS2 back contacts in the context of CZTS-based solar cells is discussed. This thorough study intends to highlight the revolutionary potentials of MoS2 back contact structures, pave the way for future developments in optoelectronics, and contribute to the continued-evolution of sustainable energy technology. This article will be a valued resource for understanding and coupling the synergies between MoS2 back surface field (BSF) and CZTS in thin film solar cell applications for future advancement.
Post-Slaughter Physiochemical Properties and Meat Quality Evaluation of Gayal (Bos frontalis)
(Faculty of Veterinary Medicine, Chattogram Veterinary and Animal Sciences University, Khulshi, Chattogram-4225, Bangladesh, 2024-09) Islam, Md Saiful
Gayal (Bos frontalis) is a semi-domesticated bovine species valued for its meat. Yet, there is limited research on its nutritional composition compared to other bovine species. Assessing the proximate and mineral content of Gayal meat can provide insights into its dietary potential and guide its use as a protein source. The study aims to evaluate the proximate composition (dry matter, ash, ether extract, and crude protein) and mineral content (calcium, magnesium, potassium, and phosphorus) across different cuts of Gayal meat. These analyses provide a better understanding of the nutritional differences between cuts and compare Gayal with other commonly consumed cattle species. This study investigates the proximate composition and mineral content of Gayal meat, a valuable yet under-researched protein source. Conducted in Bakalia, Chittagong, Bangladesh, the research focuses on the nutritional properties of various cuts of Gayal (Bos frontalis) meat, including their moisture, protein, fat, and mineral levels. Utilizing standardized methods for proximate and mineral analyses, the study reveals significant differences across cuts, highlighting the loin as a particularly rich source of protein and low in fat, while the rump offers essential minerals such as calcium and phosphorus. The findings underscore the potential of Gayal meat as a nutritious alternative to other bovine species, providing insights into its suitability for human consumption and dietary recommendations. Statistical analyses confirm the significance of these variations, paving the way for future research into the health benefits and market potential of Gayal meat in sustainable food systems. Significant variations in dry matter, fat, protein, and mineral content were identified among the cuts (p < 0.05). ANOVA and Kruskal-Wallis tests confirmed statistically significant differences, highlighting the influence of anatomical location on the nutritional profile of the meat. The loin had the lowest protein content (25.1±0.01%) and the lowest fat content (0.49±0.03%), making it a suitable choice for lean meat consumers. The brisket showed the highest fat content (0.84±0.02%). Mineral analysis revealed that the rump contained the highest calcium (0.6±0.06%) and phosphorus (3.5±0.06%) levels, while the chuck was richest in potassium (1.7±0.06%). These results indicate substantial differences in nutritional composition across the cuts. The differences in protein and fat content between cuts align with known patterns in other bovine species, where loin cuts generally contain more protein and less fat. The higher mineral content in the rump, particularly calcium and phosphorus, is likely due to its proximity to bones and its role in supporting muscle function. These findings suggest that Gayal meat offers a nutritionally rich alternative to other cattle species, particularly in terms of protein and essential minerals. Gayal meat presents a valuable 2 nutritional option, especially for consumers seeking high-protein, low-fat meat. The loin cut is especially rich in protein, while the rump offers high mineral content. Future studies should investigate the fatty acid and amino acid composition of Gayal meat to further explore its health benefits. Sensory analysis and consumer preference studies could also provide insights into its market potential. This study contributes to a broader understanding of underutilized livestock species and their role in sustainable food production systems.
Pre-COVID-19 Knowledge, Attitude and Practice Among Nurses Towards Infection Prevention and Control in Bangladesh: A Hospital-Based Cross-Sectional Survey
(Scopus, 22-12-01) Harun, Md. Golam Dostogir; Anwar, Md Mahabub Ul; Sumon, Shariful Amin; Abdullah-Al-Kafi, Md; Datta, Kusum; Haque, Md. Imdadul; Chowdhury, A. B. M. Alauddin; Sharmin, Sabrina; Islam, Md Saiful
Introduction Hospital-acquired infections endanger millions of lives around the world, and nurses play a vital role in the prevention of these infections. Knowledge of infection prevention and control (IPC) best practices among nurses is a prerequisite to maintaining standard precautions for the safety of patients. Aim The study aims to assess knowledge, attitudes, and practices (KAP) towards IPC including associated factors among the nurses of a tertiary care hospital in Bangladesh. Methods We conducted this hospital-based cross-sectional study from October 2017 to June 2018 at Dhaka Medical College Hospital among 300 nurses working in all departments. We calculated three KAP scores for each participant reflecting their current state of knowledge and compliance towards IPC measures. Descriptive, bivariate and multivariable analyses were conducted to determine KAP scores among nurses and their associated factors. Results Average scores for knowledge, attitudes, and practices were 18.6, 5.4, and 15.5 (out of 26, 7, and 24), respectively. The study revealed that the majority (85.2%) of the nurses had a good to moderate level of knowledge, half (51%) of them showed positive attitudes, and only one fifth (17.1%) of the nurses displayed good practices in IPC. The respondents’ age, education, monthly income and years of experience were found to have statistical associations with having moderate to adequate level of KAP scores. Aged and experienced nurses were found more likely to have poor knowledge and unfavorable attitude toward IPC practices. Conclusion The majority of nurses had good IPC knowledge, but their practices did not reflect that knowledge. In particular, nurses needed to improve the proper IPC practice for better patient care and to protect themselves. Regular IPC training and practice monitoring can enhance the IPC practice among nurses.
Resilience Strategies of Tour Operators During the Uncertainty of COVID-19: Evidence from Bangladesh
(Emerald Publishing Limited, 2022-11-16) Islam, Md Saiful; Kabir, Md. Mishkatul; Hassan, Kamrul
Since the outbreak of COVID-19 pandemic, tour operators have been going through uncertain times as they depend directly on supply-side (e.g. airlines, hotels) and demand-side (e.g. tourists) of tourism as well as on destination management organizations. This study explores resilience strategies made by tour operators in Bangladesh that ultimately helped them survive through the COVID-19 pandemic. Drawing on qualitative semi-structured interviews with 25 tour operators, findings of the study show that resilience-building depends not only on strategies of tour operators but also on supports from external agencies. The study further shows that a multi-dimensional understanding of resilience strategies is essential in tourism research and proposes that the resilience-building of tour operators can be conceptualized as a three-dimensional mechanism including innate resilience, internally-induced resilience, and externally-induced resilience. The study would facilitate improved resilience strategy and informed policy making to better address uncertainties during and after a major crisis for tour operators.
Using unsupervised machine learning models to drive groundwater chemistry and associated health risks in Indo-Bangla Sundarban region
(Scopus, 2024) Jannat, Jannatun Nahar; Islam, Abu Reza Md Towfiqul; Mia, Md Yousuf; Pal, Subodh Chandra; Biswas, Tanmoy; Jion, Most Mastura Munia Farjana; Islam, Md Saiful; Siddique, Md Abu Bakar; Idris, Abubakr M.; Khan, Rahat; Islam, Aznarul; Kormoker, Tapos; Senapathi, Venkatramanan
Groundwater is an essential resource in the Sundarban regions of India and Bangladesh, but its quality is deteriorating due to anthropogenic impacts. However, the integrated factors affecting groundwater chemistry, source distribution, and health risk are poorly understood along the Indo-Bangla coastal border. The goal of this study is to assess groundwater chemistry, associated driving factors, source contributions, and potential non-carcinogenic health risks (PN-CHR) using unsupervised machine learning models such as a self-organizing map (SOM), positive matrix factorization (PMF), ion ratios, and Monte Carlo simulation. For the Sundarban part of Bangladesh, the SOM clustering approach yielded six clusters, while it yielded five for the Indian Sundarbans. The SOM results showed high correlations among Ca2+, Mg2+, and K+, indicating a common origin. In the Bangladesh Sundarbans, mixed water predominated in all clusters except for cluster 3, whereas in the Indian Sundarbans, Cl−-Na+ and mixed water dominated in clusters 1 and 2, and both water types dominated the remaining clusters. Coupling of SOM, PMF, and ionic ratios identified rock weathering as a driving factor for groundwater chemistry. Clusters 1 and 3 were found to be influenced by mineral dissolution and geogenic inputs (overall contribution of 47.7%), while agricultural and industrial effluents dominated clusters 4 and 5 (contribution of 52.7%) in the Bangladesh Sundarbans. Industrial effluents and agricultural activities were associated with clusters 3, 4, and 5 (contributions of 29.5% and 25.4%, respectively) and geogenic sources (contributions of 23 and 22.1% in clusters 1 and 2) in Indian Sundarbans. The probabilistic health risk assessment showed that NO3− poses a higher PN-CHR risk to human health than F− and As, and that potential risk to children is more evident in the Bangladesh Sundarban area than in the Indian Sundarbans. Local authorities must take urgent action to control NO3− emissions in the Indo-Bangla Sundarbans region.
Using Unsupervised Machine Learning Models To Drive Groundwater Chemistry and Associated Health Risks in Indo-Bangla Sundarban Region
(Elsevier, 2024-03-20) Jannat, Jannatun Nahar; Islam, Abu Reza Md Towfiqul; Mia, Md Yousuf; Pal, Subodh Chandra; Biswas, Tanmoy; Jion, Most Mastura Munia Farjana; Islam, Md Saiful; Siddique, Md Abu Bakar; Idris, Abubakr M.; Khan, Rahat; Islam, Aznarul; Kormoker, Tapos; Senapathi, Venkatramanan
Groundwater is an essential resource in the Sundarban regions of India and Bangladesh, but its quality is deteriorating due to anthropogenic impacts. However, the integrated factors affecting groundwater chemistry, source distribution, and health risk are poorly understood along the Indo-Bangla coastal border. The goal of this study is to assess groundwater chemistry, associated driving factors, source contributions, and potential non-carcinogenic health risks (PN-CHR) using unsupervised machine learning models such as a self-organizing map (SOM), positive matrix factorization (PMF), ion ratios, and Monte Carlo simulation. For the Sundarban part of Bangladesh, the SOM clustering approach yielded six clusters, while it yielded five for the Indian Sundarbans. The SOM results showed high correlations among Ca2+, Mg2+, and K+, indicating a common origin. In the Bangladesh Sundarbans, mixed water predominated in all clusters except for cluster 3, whereas in the Indian Sundarbans, Cl−-Na+ and mixed water dominated in clusters 1 and 2, and both water types dominated the remaining clusters. Coupling of SOM, PMF, and ionic ratios identified rock weathering as a driving factor for groundwater chemistry. Clusters 1 and 3 were found to be influenced by mineral dissolution and geogenic inputs (overall contribution of 47.7%), while agricultural and industrial effluents dominated clusters 4 and 5 (contribution of 52.7%) in the Bangladesh Sundarbans. Industrial effluents and agricultural activities were associated with clusters 3, 4, and 5 (contributions of 29.5% and 25.4%, respectively) and geogenic sources (contributions of 23 and 22.1% in clusters 1 and 2) in Indian Sundarbans. The probabilistic health risk assessment showed that NO3− poses a higher PN-CHR risk to human health than F− and As, and that potential risk to children is more evident in the Bangladesh Sundarban area than in the Indian Sundarbans. Local authorities must take urgent action to control NO3− emissions in the Indo-Bangla Sundarbans region.
Verifying online signatures through an iterative device independent model
(BRAC University, 2023-01) Tahsin, Samiha; Molla, Robin; Jamal, Omran; Islam, Md Saiful; Rahman, Rafeed
Hand signatures are getting used from as early as we invented writing. In 3100 BC, we found examples of people using words and symbols to denote their identity. It has also been used as a method of identification. Modern society kept hand signatures for many purposes like the authentication of banking and real estate fields. The recent trend of working from home and business on the go created a necessity to bring the signature from paper to smartphone. Statistics also indicated that it is a user-preferred method of verification. In this paper, we proposed a novel method to verify online signatures using an iterative approach that is device independent. It will be helpful to bring the signatures from paper to smartphones. In this method, we have created a model per signatory, based on their behavioral pattern on each point based on time and distance from the start of the signature. We also considered the defference between the signatory’s own signatures while training. We worked with defferent derived datapoints like velocity, angular velocity etc. We have achieved 8% EER on the MCYT dataset and 20% EER on the Mobisig dataset.