DR-NTU (Digital Repository of NTU)
Not a member yet
116018 research outputs found
Sort by
Hypoxia-tolerant polymeric photosensitizer prodrug for cancer photo-immunotherapy
No full textAlthough photodynamic immunotherapy represents a promising therapeutic approach against malignant tumors, its efficacy is often hampered by the hypoxia and immunosuppressive conditions within the tumor microenvironment (TME) following photodynamic therapy (PDT). In this study, we report the design guidelines towards efficient Type-I semiconducting polymer photosensitizer and modify the best-performing polymer into a hypoxia-tolerant polymeric photosensitizer prodrug (HTPSNiclo) for cancer photo-immunotherapy. HTPSNiclo not only performs Type-I PDT process to partially overcome the limitation of hypoxic tumors in PDT by recycling oxygen but also specifically releases a Signal Transducer and Activator of Transcription-3 (STAT3) inhibitor (Niclosamide) in response to a cancer biomarker in the TME. Consequently, HTPSNiclo inhibits the phosphorylation of STAT3, and suppresses the expression of hypoxia-inducible factor-1α. The synergistic effect results in the enhanced activation of immune cells (including mature dendritic cells, cytotoxic T cells) and production of immunostimulatory cytokines compared to Type-I PDT alone. Thus, HTPSNiclo-mediated photodynamic immunotherapy enhances tumor inhibition rate from 75.53% to 91.23%, prolongs the 100% survival from 39 days to 60 days as compared to Type-I PDT alone. This study not only provides the generic approach towards design of polymer-based Type-I photosensitizers but also uncovers effective strategies to counteract the immunosuppressive TME for enhanced photo-immunotherapy in 4T1 tumor bearing female BALB/c mice.Ministry of Education (MOE)National Research Foundation (NRF)Published versionY.Z. thanks National Natural Science Foundation of China (22322406) for financial support. K.P. thanks Singapore National Research Foundation (NRF) (NRF-NRFI07-2021-0005) and the Singapore Ministry of Education Academic Research Fund Tier 2 (MOE-T2EP30220-0010, MOE-T2EP30221-0004) for financial support. This article was also supported by China postdoctoral science foundation under grant number GZC20240525 (J.Y.)
Associations between GFAP, Aβ42/40 ratio, and perivascular spaces and cognitive domains in vascular cognitive impairment
No full textPerivascular spaces (PVS) support metabolic clearance in the brain and are increasingly recognized as key contributors to dementia pathogenesis. Plasma-based biomarkers, such as glial fibrillary acidic protein (GFAP) and the amyloid β42/40 (Aβ42/40) ratio, show promise in dementia diagnosis but remain understudied in vascular cognitive impairment (VCI). VCI, a major global cause of cognitive decline, may be more prevalent in Southeast Asia. Despite its impact, it is underdiagnosed compared to Alzheimer’s, highlighting the need for early, reliable markers. This study aims to examine how these biomarkers relate to PVS burden and domain-specific cognitive outcomes in VCI. VCI was defined as global cognition as assessed by a Montreal Cognitive Assessment Score 2 or periventricular hyperintensities >3), and >1 lacuna. A total of 108 participants (mean age of 67.3 years, 51.9% female) were included. Multivariate ordinal regression assessed biomarker associations with PVS grade, adjusting for age and diastolic blood pressure. A Aβ42/40 ratio 54.1 pg/mL were used as biomarker thresholds to subgroup the participants, and the relationship between these thresholds and cognitive performance was analyzed. Elevated GFAP (p = 0.0438) and a reduced Aβ42/40 ratio (p < 0.01) were correlated with a higher PVS grade. In the subgroup with a low Aβ42/40 ratio, a greater PVS burden was associated with poorer executive function (p = 0.045, β = 0.612), while in those with high GFAP levels, it was linked to more pronounced impairments in learning and memory (p = 0.006, β = 0.375). A lower Aβ42/40 ratio and higher GFAP levels track greater PVS burden in VCI. PVS severity may be associated with domain-specific cognitive decline, highlighting the potential utility of these biomarkers in refining clinical assessments and monitoring disease progression.Ministry of Education (MOE)Nanyang Technological UniversityPublished versionThis project was funded by the Strategic Academic Initiative (SAI) Grant (SP1CLNT900- NTU-A630-PJ-03INP001400A630OOE01) from the Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore and the Ministry of Education Start-up Grant. W.J.D.J. was supported by Nanyang Technological University under the URECA Undergraduate Research Programme
InsAIghts: dynamic dashboard leveraging LLMs for data exploration
No full textIn light of the rapid proliferation of data analytics tools and the growing technical skill gap among business users, InsAIghts Dashboard was conceptualised as a cost-effective and user-friendly solution to accelerate data driven decision-making. The goal is to empower all business, regardless of financial or resource constraints, to gain actionable insights with minimal effort, all whilst ensuring privacy and security of their data.
This paper will provide a comprehensive analysis on the development of the InsAIghts Dashboard application, highlighting important architectural and design decisions made and provide greater insights into providing cost-effective data analytics tools with advanced artificial intelligence solutions. Furthermore, the report will also discuss the limitations of the project, and provide recommendations for future improvement.Bachelor's degre
The cloak of creativity: AI imaginaries and creative labour in media production
No full textWe critique the ways ‘creativity’ is harnessed by artificial intelligence (AI) companies, technologists, cultural producers, and even academics to justify a sense of urgency around the adoption of AI tools in creative work. The creative industries have seen the rapid deployment of AI-powered video generation tools such as Sora and Runway, accompanied by a broad communication and publicity campaign that emphasises AI as a co-creative tool that enables human practitioners to ‘boost’ or ‘enhance’ their productivity. These discourses not only position and frame creativity as a uniquely human, cognitive trait that can now be extended, expanded, or even fully automated by AI, but they also conjure imaginaries of AI models as entities capable of simultaneously generating ‘original creative output’ while also acting as servile ‘assistants’ taking on rote tasks (Murati, 2022: 164). Underlying these imaginaries is a model of creativity as a cognitive faculty is possessed by humans or replicated by machine algorithms; yet, this imagination dehumanises actual humans (Bender, 2024). Reflecting on our study of European film practitioners using AI tools in their creative workflows, we argue that the dominant narrative of creativity as a faculty serves merely to cloak the material, networked, and relational nature of creativity. From this approach, the issue of labour is paramount to the conceptualisation of ‘creativity’ in the algorithmic age and we argue for a de-cloaking and re-materialisation of human creative labour. ‘Creativity’ is neither human nor machinic, but rather a networked and distributed form of agency.Submitted/Accepted versio
Parallel processing framework for analyzing taxi mobility data
No full textThe unprecedented growth of data generation has created a demand and need for efficient data processing frameworks that can handle large volumes of raw, structured and unstructured data. These data then undergo transformation and preprocessing to turn them into a suitable format that is easy to perform data analysis and machine learning. This report examines the limitations of using pandas with multiprocessing for large-scale data processing and explores parallel processing frameworks such as Hadoop and Apache Spark.
A mobility data set spanning 253GB, consisting of taxi trips data between January to December 2022 in Singapore was provided by Land Transport Authority (LTA) and used to study the performance of different data processing frameworks. A Spark cluster was set up to process the raw mobility data, utilizing profiling techniques to identify and resolve logical and hardware bottlenecks, ensuring optimal utilization of the compute resources Spark ran on. A comparison between performance revealed that Apache Spark consistently outperforms pandas with multiprocessing when processing large mobility datasets across multiple months. Subsequently, the processed mobility data was structured using different data models, namely One Big Table and Fact-Dimension and their query performance were evaluated. Eventually, these data models were hosted on Google BigQuery, where curated multi-layered data models were implemented to optimize data retrieval, improve accessibility and support different analytics and machine learning applications.Bachelor's degre
Integrated project allocation and examiner assignment system for final year projects part 4
No full textThe Final Year Project (FYP) is a major academic milestone for students at the College of Computing and Data Science (CCDS), encompassing project allocation, student-supervisor matching, and examiner assignment. Historically, these processes were carried out manually, leading to inefficiencies and inconsistent workload distribution.
To address these challenges, three core systems were initially introduced: the Supervision Computation Allocation (SCA), Student Selection System (SSS), and Final Year Project Examiner Allocation System (FEAS). This project builds upon those foundations by modernising and consolidating them into a unified web-based platform, the FYP Application (FYP-APP).
The system integrates multiple modules to automate key FYP lifecycle stages. The SCA calculates the number of project proposals each faculty member should submit based on workload. The SSS uses the Hungarian algorithm to optimise student-project allocations based on ranked preferences (Kuhn, 1955). Finally, the FEAS module ensures fair examiner assignments, considering faculty availability and teaching load.
This report outlines enhancements made in streamlining examiner allocation, refining student project assignment, improving the user interface, and migrating legacy logic into a scalable, maintainable platform. The project ultimately aims to improve the efficiency, accuracy, and transparency of the FYP management process.Bachelor's degre
Sentic API testing for polarity classification
No full textSentiment analysis also known as opinion mining, uses Natural Language Processing (NLP) to analyze a given text and identify the underlying emotion associated with the text. The sentiments are identified as negative, neutral or positive. Sentiment analysis helps to provide valuable insights with regards to opinions, feedbacks as well trends. The insights can be performed on social media, online shopping platforms and many more providing insights.Bachelor's degre
Mitigating core energy losses in Fe-Si alloys fabricated by direct energy deposition through oxide inclusions and abnormal Goss grain growth
No full textIn traditional electrical steel production oxide inclusions are conventionally perceived as deleterious elements for the functional and structural properties. The present work describes the fabrication of a high silicon content electrical steel alloy (Fe-6.5wt%Si) using directed energy deposition (DED), coupled with oxide inclusions to mitigate core energy losses. Abnormal Grain Growth (ABG) was observed after thermal post-processing at 1000 °C for 24 h (1000–24), together with the creation of oxide inclusions mainly around the grain boundaries. Magnetic properties were assessed through dynamic and quasi-static measurements for both as-printed (AP) and 1000–24 samples. The quasi-static analysis revealed hysteresis losses of 206.9 J/m3 for the AP and 19.02 J/m3 for the 1000–24, with maximum flux densities of 1.295 T and 1.031 T, at the magnetic field of 3000 A/m. Dynamic magnetic analysis demonstrated an improvement of 39.2% in the total core losses of the 1000–24 sample (2088.8 J/m3), compared to the AP sample (3436.9 J/m3). The microstructure of the 1000–24 sample revealed the formation of Goss texture via ABG, ultimately decreasing the static hysteresis loss. Furthermore, an improved electrical resistivity compare to conventional electrical steel alloys was demonstrated at 119 μΩcm for the 1000-24 sample, and 105 μΩcm for the AP sample. This work introduces a promising avenue to minimize core energy losses by incorporating oxide inclusions and ABG Goss texture in additively manufactured soft magneitc components after thermal post-processing.Agency for Science, Technology and Research (A*STAR)National Research Foundation (NRF)Published versionThe work was supported by the Panasonic Industrial Devices Singapore Pte. Ltd, Singapore, under its Joint Lab Programme 04IDS001408C140, National Research Foundation (NRF) Singapore under its NRF Fellowship Grant NRF-NRFF12- 2020-0003. The work is also supported by the Agency for Science, Technology and Research (A*STAR) of Singapore via the Individual Research Grant (Grant reference No. A20E7c0109)
Trustworthy of large language models
No full textLarge language models (LLMs) are widely used nowadays across various applications. Despite their usefulness, there have been growing concerns about the integrity of the information they generate. Hallucination, the phenomenon where LLMs produce content that contradicts factual knowledge, has been observed in many real-world scenarios. This paper investigates the hallucination problem and explores the use of Retrieval-Augmented Generation (RAG) and Prompt Engineering (PE) as mitigation strategies. Experimental results indicate that RAG alone is effective in reducing hallucinations in fact-based question-answering (QA) tasks. However, for long-form texts, more advanced methods are required for the model to develop a deeper semantic understanding of the retrieved content.Bachelor's degre
The battle of narratives: China's global brand revolution vs Western dominance
No full textThe emergence of DeepSeek AI has sparked a narrative clash between China and the West. Western critics argue that DeepSeek reflects Chinese state censorship and poses security risks due to data regulations, threatening free speech and privacy. China counters by positioning DeepSeek as an affordable, open alternative to Western AI, promoting it as a tool for technological independence in emerging markets. This rivalry frames the debate between China and the West as a struggle for global technological leadership and ideological dominance.Published versio