BRAC University Institutional Repository
Preserving Knowledge, Advancing Research, Sharing Scholarship
A digital platform for collecting, preserving, and sharing BRAC University’s scholarly, academic, and institutional outputs.
Contribute to the Repository
Faculty members and students are invited to submit their research publications, theses, dissertations, and scholarly works to increase visibility, access, and long-term preservation.
Recent Submissions
Covid-19 pandemic: Review of contemporary and forthcoming detection tools
(Dove Medical Press Ltd, 2021-01-01) Oishee, Mumtarin Jannat; Ali, Tamanna; Jahan, Nowshin; Khandker, Shahad Saif; Haq, Md Ahsanul; Khondoker, Mohib Ullah; Sil, Bijon Kumar; Lugova, Halyna; Krishnapillai, Ambigga; Abubakar, Abdullahi Rabiu; Kumar, Santosh; Haque, Mainul; Jamiruddin, Mohd Raeed; Adnan, Nihad; School of Pharmacy
Recent severe acute respiratory syndrome 2 (SARS-CoV-2) known as COVID-19, presents a deadly challenge to the global healthcare system of developing and developed countries, exposing the limitations of health facilities preparedness for emerging infectious disease pandemic. Opportune detection, confinement, and early treatment of infected cases present the first step in combating COVID-19. In this review, we elaborate on various COVID-19 diagnostic tools that are available or under investigation. Consequently, cell culture, followed by an indirect fluorescent antibody, is one of the most accurate methods for detecting SARS-CoV-2 infection. However, restrictions imposed by the regulatory authorities prevented its general use and implementation. Diagnosis via radiologic imaging and reverse transcriptase PCR assay is frequently employed, considered as standard proce-dures, whereas isothermal amplification methods are currently on the verge of clinical introduction. Notably, techniques such as CRISPR-Cas and microfluidics have added new dimensions to the SARS-CoV-2 diagnosis. Furthermore, commonly used immunoassays such as enzyme-linked immunosorbent assay (ELISA), lateral flow immunoassay (LFIA), neutralization assay, and the chemiluminescent assay can also be used for early detection and surveillance of SARS-CoV-2 infection. Finally, advancement in the next generation sequencing (NGS) and metagenomic analysis are smoothing the viral detection further in this global challenge.
Real-time Bangladeshi Sign Language recognition and natural language interpretation using hybrid deep learning and LLMs
(BRAC University, 2026-06) Fairuz, Fabliha Akther; Roy, Santonu; Mahmud, Sajid; Tasnim, Sumiya; Rafey, Inkiad Bin Ershad; Alam, Md. Ashraful; Department of Computer Science and Engineering
Deaf and hard-of-hearing communities in Bangladesh face a persistent communication
gap: existing sign language recognition systems process hand gestures but ignore the
facial expressions and head movements that carry grammatical meaning in Bangladeshi
Sign Language (BdSL). This study introduces BdSL-NMM, the first BdSL dataset annotated
with non-manual marker labels alongside word-level glosses, covering 62 sign classes
and 5 expression categories across 4,170 recordings from seven native signers. To evaluate
on this dataset, a multi-stream Transformer architecture called SignNet-V2 was developed,
which simultaneously recognises sign words and non-manual expression markers
from skeletal landmark sequences. The system processes four input modalities, namely
body pose, left hand, right hand, and face mesh, through dedicated stream-specific encoders,
cross-stream attention fusion, hierarchical temporal encoding, and a multi-task
classification head. All models were trained and evaluated under leave-one-signer-out
cross-validation, where the test signer never appeared during training, providing a realistic
measure of cross-signer generalisation. Under this protocol, SignNet-V2 achieves
38.44% top-1 word recognition accuracy. For expression recognition, a 56-dimensional
framework of normalised geometric and temporal features achieves 66.39% overall accuracy
under signer-independent evaluation, with neutral and negation recall of 97.22% and
91.18% on the unseen test signer. A second interpretation stage passes recognised signs
and expression tags to Google Gemini, which generates grammatically correct Bengali
sentences. This work establishes the first published benchmarks for simultaneous word
and non-manual marker recognition in BdSL under signer-independent evaluation.
Sustainable procurement practices in Ashuganj Power Station Company Limited (an enterprise of BPDB)
(BRAC University, 2025-12) Mamun, Abdullah Al; Chowdhury, Mahbub Ahmed; BRAC Institute of Governance and Development
The given report will analyse sustainable procurement practices of Ashuganj Power Station Company Limited (APSCL) in terms of how the environmental, social and governance (ESG) principles are integrated into the procurement process and the operations of the company in general. The paper assesses the efforts of APSCL in adopting renewable energy, environmental and safety compliance due to the ISO requirements, digital procurement governance and community engagement. With the help of document analysis, field observations, and stakeholder interviews, the report determines that most of the advancement has been made in replacing outdated power units with high-efficiency combined cycle plants and using tools that ensure transparency, including e-GP, d-Nothi and the ASCL E-Auction system. The strengths identified in the analysis are policy support and alignment of the ESG, and the weaknesses are the gaps in compliance with the suppliers, and the cost restraints. To improve the sustainability of APSCL, the report recommends the adoption of a roadmap, which embraces ESG, better governance, improved vendor management and increased social and environmental interventions.
Breast cancer management in the era of Covid-19; Key issues, contemporary strategies, and future implications
(Dove Medical Press Ltd, 2023-01-01) Prodhan, A. H. M. Safayet Ullah; Islam, Dewan Zubaer; Khandker, Shahad Saif; Jamiruddin, Mohd Raeed; Abdullah, Adnan; Godman, Brian; Opanga, Sylvia; Kumar, Santosh; Sharma, Paras; Adnan, Nihad; Pisana, Alice; Haque, Mainul; School of Pharmacy
During the COVID-19 pandemic, several priority diseases were not getting sufficient attention. Whilst breast cancer is a fatal disease affecting millions worldwide, identification and management of these patients did not initially attract critical attention to minimize the impact of lockdown, post-lockdown, and other measures. Breast cancer patients’ conditions may not remain stable without proper care, worsening their prognosis. Proper care includes the timely instigation of surgery, systemic therapy, and psychological support. This includes low-and middle-income countries where there are already concerns with available personnel and medicines to adequately identify and treat these patients. Consequently, there was a need to summarize the current scenario regarding managing breast cancer care during COVID-19 across all countries, including any guidelines developed. We systematically searched three scientific databases and found 76 eligible articles covering the medical strategies of high-income countries versus LMICs. Typically, diagnostic facilities in hospitals were affected at the beginning of the pandemic following the lockdown and other measures. This resulted in more advanced-stage cancers being detected at initial presentation across countries, negatively impacting patient outcomes. Other than increased telemedicine, instigating neo-adjuvant endocrine therapy more often, reducing non-essential visits, and increasing the application of neo-adjuvant chemotherapy to meet the challenges, encouragingly, there was no other significant difference among patients in high-income versus LMICs. Numerous guidelines regarding patient management evolved during the pandemic to address the challenges posed by lockdowns and other measures, which were subsequently adopted by various high-income countries and LMICs to improve patient care. The psychological impact of COVID-19 and associated lockdown measures, especially during the peak of COVID-19 waves, and the subsequent effect on the patient’s mental health must also be considered in this high-priority group. We will continue to monitor the situation to provide direction in future pandemics.
Students’ experiences of using AI conversational tools to enhance their speaking confidence in English
(BRAC University, 2026-06) Farisha, Nujhat Kashfia; Rubaiat, Shakika; Department of English and Humanities
Free versions of AI conversation apps are widely used for English speaking practice, but we know
very little about how their limits (like daily time caps, annoying pop‑up ads, and locked features)
affect learning. I interviewed five high school students one‑on‑one using Zoom. Three of them
only used the free version of Duolingo; the other two also paid for subscriptions to Lingoda or
Cambly. I analysed the interview transcripts using thematic analysis (Braun & Clarke, 2006). The
students said they liked the privacy and safety of AI apps because it made them less anxious than
speaking in class. But free‑version users often got frustrated when the app didn't understand their
accent, sounded robotic, or showed an ad right after they made a mistake. Paid users got live
feedback from real teachers and felt much more confident. Everyone agreed that apps can't replace
talking to real people. Free‑version problems – especially ads popping up after errors – broke their
focus, killed motivation, and made practice inconsistent. Paid users struggled with high
subscription costs that sometimes forced them to pause their learning. The study shows that
punishing mistakes with ads works against learning. App designers should offer ad‑free basic
speaking practice and cheaper student plans.