| 1 |
Author(s):
Sihoon Kim.
Page No : 1-6
|
An Empirical Evaluation of DragGAN’s Efficacy Across Distinct Subject Categories
Abstract
Generative Adversarial Networks (GANs) have reshaped the landscape of synthetic media, enabling
the creation of hyper-realistic imagery through adversarial learning. Within this domain, DragGAN
has emerged as a notable innovation, offering intuitive point-based manipulation of generated images
by translating user-specified handle points to target spatial locations. Despite its qualitative success in
published demonstrations, a rigorous quantitative evaluation of its performance across varying semantic
categories remains absent from the literature. This study addresses that gap by assessing DragGAN’s
efficacy in maintaining structural integrity and generating diverse outputs across four distinct subject
categories: human faces, dog faces, cat faces, and whole dog bodies. Using a curated dataset of images
generated from pre-trained StyleGAN2 checkpoints (FFHQ, AFHQ, and a Self-Distilled StyleGAN
body model), the Structural Similarity Index (SSIM) was applied to measure fidelity and a decomposed
Inception Score (IS) was used to evaluate perceptual quality and diversity. All categories exhibited
substantial structural degradation under point-based manipulation, with mean SSIM scores ranging
from 0.21 (cat faces) to 0.33 (dog bodies). The full-body dog category achieved the highest structural
preservation, while facial categories—particularly cat and dog faces—showed the greatest degradation.
Decomposed Inception Score analysis indicated consistently low classifier confidence across all
categories, a pattern attributable to domain mismatch between the generated subjects and the ImageNettrained
Inception-v3 classifier. These findings establish a quantitative baseline indicating that DragGAN’s
point-based manipulation introduces significant structural distortion across all tested domains, with
relative performance differences suggesting that full-body manipulation may be more tractable than finegrained
facial editing.
| 2 |
Author(s):
Faith Koh.
Page No : 7-15
|
Sex Differences in Thermoregulation During Exercise in Hot-Humid Environments: A Narrative Review with Implications for Athletic Populations in Singapore
Abstract
This narrative review synthesizes current evidence on sex differences in thermoregulatory responses
during exercise, with specific application to athletes training in Singapore’s hot-humid climate. The
review examines physiological mechanisms underlying sex-based differences in sweating, cutaneous
vasodilation, and hormonal influences on thermal regulation. Evidence indicates that males generally
exhibit higher sweat rates and earlier sweating onset, whereas females demonstrate lower sweat output
per gland despite higher gland density, with additional thermoregulatory variability introduced by
menstrual cycle phases. In hot-humid environments characteristic of Singapore, reduced evaporative
cooling efficiency may attenuate these sex differences, as high ambient humidity limits heat dissipation
regardless of sweat production capacity. This review evaluates Singapore’s athletic heat stress
frameworks, particularly Sport Singapore’s Heat Stress Management Plan, identifying gaps in sexspecific
provisions. Recommendations include individualized hydration guidelines based on sweat rates,
educational initiatives regarding menstrual cycle effects on thermoregulation, and Singapore-specific
research to validate laboratory findings in field conditions.
| 3 |
Author(s):
Blossom Patel, Hangpeng Li.
Page No : 16-19
|
The Biological Significance of cfDNA Methylation Patterns in Early Ovarian Cancer Detection and Analytical Methods for Detection
Abstract
Ovarian cancer is one of the deadliest gynecological cancers; survival over five years drops sharply,
from about 90% when found early to under 30% if detected late. Although ovarian cancer detection based
on blood tests that analyze cell-free DNA (cfDNA) could provide a less invasive option, finding consistent
signals in the limited data found in blood samples is difficult. To address the high-dimensional nature
of this data, the critical challenge of data leakage in machine learning pipelines was investigated. Taken
together, combining sophisticated AI methods with highly sensitive methylation tests appears to offer the
best chance for developing practical early-detection screening tools. A synthetic dataset was generated
that mirrors plasma cfDNA methylation fragments, and a simulation was performed to determine the
number of cancerous versus noncancerous differentially methylated regions (DMRs). First, a standard
“global” selection model was simulated, which displayed the artificial inflation (overfitting) of accuracy
that occurs due to data leakage. Second, a nested-CV elastic-net model was tested to isolate the true
biological signal from cfDNA methylation. After modeling this procedure, the leakage-safe model could
successfully distinguish early ovarian cancer from non-tumor samples (AUROC 0.938; AUPRC 0.914).
This project lays the foundation for future exploration of theory-driven, AI-powered liquid biopsy models.
| 4 |
Author(s):
Emily Avram, Rakesh Chand.
Page No : 20-27
|
Global Surveillance Evidence of Rabies as a Threat to Wildlife Conservation
Abstract
Rabies is a fatal zoonotic disease that is widely recognized as a public health priority, yet its
implications for wildlife conservation remain insufficiently examined. This study assessed whether
rabies constitutes a conservation threat by analyzing international surveillance data alongside a narrative
synthesis of published evidence on urban–sylvatic transmission. Reported wildlife rabies deaths were
extracted from the World Organisation for Animal Health’s World Animal Health Information System
(WAHIS) to describe long-term global trends from 2005 to 2024 and recent regional patterns from 2020
to 2024, with species-level records mapped to International Union for Conservation of Nature Red List
categories. Surveillance data showed that wildlife rabies deaths have been consistently reported across
all major global regions over the past two decades, with a general decline in reported deaths since 2012.
Between 2020 and 2024, 4,850 wildlife rabies deaths were reported, including 53 deaths (1.09%) in species
classified as endangered, critically endangered, or vulnerable, encompassing multiple taxonomic groups
and geographic regions. Although the proportion of reported deaths in conservation-priority species was
small, their distribution across vulnerable taxa indicates that rabies exposure extends into populations
where even limited mortality may have disproportionate conservation consequences. Interpretation of
these findings is constrained by substantial underreporting and uneven surveillance capacity, particularly
in resource-limited settings. Overall, the results indicate that rabies represents an underrecognized but
meaningful risk to wildlife conservation, especially at domestic animal–wildlife interfaces, and that
integrating rabies control measures, particularly mass dog vaccination, with conservation planning and
surveillance within a One Health framework may support biodiversity protection while advancing global
rabies elimination efforts.
| 5 |
Author(s):
Thondari Cho Thar.
Page No : 28-37
|
Organizational Reconfiguration in Response to Technological Disruption: A Case Study of Microsoft Under Satya Nadella
Abstract
When Satya Nadella became CEO of Microsoft in 2014, the company faced a defining shift: moving
from desktop software to cloud computing. This case study examines how Microsoft transformed
between 2014 and 2018, applying theories of dynamic capabilities, organizational ambidexterity, and
identity to understand how real change happens. The findings show that Microsoft recognized the shift
to the cloud, invested in Azure, and restructured its workforce. Change spread through distributed
adaptation, where employees at all levels experimented and learned. The Stratus team served as a teacher
model, building skills within departments rather than isolating innovation. Microsoft’s identity shifted
through daily work as identity emerged from practice; new habits reshaped how the company saw itself.
Underlying these changes was psychological safety, the trust that enabled people to take risks. This
study contributes to adaptation theory by introducing distributed adaptation, questioning the assumption
that ambidexterity requires separate units through a teacher model, and suggesting that organizational
identity can emerge from behavioral change rather than precede it. The results show that lasting change
arises from habits embedded in workplace culture, offering useful insights for organizations facing
technological disruption.
| 6 |
Author(s):
Aarav P. Narang.
Page No : 38-43
|
The Gut–Brain Axis and Insulin Resistance: Evaluating Alzheimer’s Disease as Type 3 Diabetes
Abstract
Alzheimer’s disease is a progressive neurodegenerative disorder characterized by memory loss and
cognitive decline. While traditionally associated with amyloid-beta plaques and tau protein tangles,
recent research suggests that metabolic dysfunction, particularly impaired insulin signaling in the brain,
may play a central role in its development. This supports the hypothesis that Alzheimer’s disease may
represent “Type 3 Diabetes.” At the same time, the gut–brain axis has emerged as an important system
linking gut microbiota to brain function through immune, metabolic, and neural pathways. This paper
examines how gut microbiome dysbiosis may contribute to insulin resistance and Alzheimer’s disease.
Evidence suggests that dysbiosis promotes inflammation and disrupts insulin signaling, but causality
remains unclear.
| 7 |
Author(s):
Raudeen Roodgarmi.
Page No : 44-51
|
Measuring AI Accuracy on Standardized Tests: A Comparative Study of ChatGPT, Copilot and Gemini
Abstract
This study evaluates the performance of three widely used artificial intelligence systems, ChatGPT,
Microsoft Copilot, and Google Gemini, on standardized test questions in Math, Reading, and English
derived from SAT and ACT examinations. A total of 90 questions (30 per subject) were selected from
multiple test forms across different years to reduce potential bias and ensure broad content coverage. All
questions, including those with visual components, were presented to each AI system in a standardized
format, and responses were scored for accuracy. A chi-square test for homogeneity was conducted
to assess differences in performance among the models. Results indicate that all three AI systems
performed strongly in language-based tasks, Reading and English. In contrast, performance in Math was
notably lower across all models, with common errors involving advanced mathematical concepts and
misinterpretation of visual and graphical information. Despite observable differences in error patterns,
statistical analysis revealed no significant differences in overall performance among the three systems.
These findings suggest that current AI models are highly proficient in processing and interpreting
textual information but remain less reliable in mathematical reasoning and multimodal tasks. The study
highlights both the capabilities and limitations of AI in standardized testing contexts and underscores
the importance of prompt design and continued model development.
| 8 |
Author(s):
Aarit Das.
Page No : 52-61
|
Dynamic Multi-Asset Portfolio Optimization: Evaluating Risk-Return Tradeoffs Under Time-Varying Volatility
Abstract
Prior literature on the low-volatility anomaly suggests that portfolios composed of lower-volatility
assets often achieve superior risk-adjusted returns compared to their higher-volatility counterparts over
long investment horizons. This phenomenon challenges the traditional risk-return trade-off implied by
classical asset pricing models, which associate higher risk with higher expected returns. As a result, lowvolatility
strategies have gained attention for their ability to deliver more efficient return profiles with
reduced downside risk. This study investigates this claim by comparing alternative portfolio allocation
strategies in optimizing the risk-return tradeoff over the 2021-2025 period. Using volatility forecasts
generated through a GARCH (1,1) model and evaluating two allocation frameworks - volatility targeting
and Sharpe ratio-constrained optimization, we examine performance across varying market conditions.
The findings indicate that Sharpe ratio-constrained optimization produces higher returns during strong
market recoveries; however, it is vulnerable to significant drawdowns and elevated tail risk during market
downturns. In contrast, the volatility-targeting strategy demonstrates greater stability, lower maximum
drawdowns, and more consistent risk-adjusted performance in adverse market environments. Overall,
the results suggest that dynamic rebalancing and active risk management are critical determinants of
long-term portfolio performance. High-volatility assets, such as Bitcoin, may enhance returns, but only
when their exposure is carefully managed within a diversified portfolio framework.