De Anza College

With the rapid proliferation of Internet of Things (IoT) devices, intrusion detection in resource-constrained environments has become a critical challenge. Traditional centralized machine learning approaches are infeasible due to bandwidth limitations, privacy concerns, and the heterogeneous nature of IoT deployments. In this paper, we propose LF-IDS, a Lightweight Federated Learning Intrusion Detection System tailored specifically for IoT environments. The framework integrates a LightGBM-based local model with a differential privacy mechanism and a communication-efficient aggregation strategy to achieve high detection accuracy while minimizing model size, communication overhead, and memory consumption. Experiments conducted on the NSL-KDD and N-BaIoT benchmark datasets demonstrate that LF-IDS achieves a detection accuracy of 98.347%, a precision of 97.915%, recall of 98.621%, and an F1-score of 98.267%, surpassing four state-of-the-art baselines. Moreover, the proposed framework reduces the model size by 71.5% and per-round communication cost by 84.7% compared to existing federated learning approaches, confirming its practical viability on edge IoT devices such as Raspberry Pi.

BackgroundEvidence regarding the combined influence of healthy plant-based dietary patterns and lifestyle behaviours on obesity-related cancer outcomes remains limited. We prospectively investigated the associations of the healthy plant-based diet-lifestyle (hPDI-Lifestyle) score with obesity-related cancer incidence and mortality in the UK Biobank.MethodsAmong 104,713 eligible participants, 93,763 were included in the primary complete-case analyses. The hPDI-Lifestyle score integrated healthy plant-based dietary quality, physical activity, smoking status, and sleep duration. Restricted cubic spline analyses were used to examine dose–response relationships, and Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for the three outcomes, including sex-stratified analyses. Several sensitivity analyses were conducted to assess the robustness of the findings.ResultsDuring median follow-up periods of 16.3, 15.0, and 15.1 years, respectively, 2,258 incident obesity-related cancers, 1,095 obesity-related cancer deaths, and 6,215 all-cause deaths were documented. Higher hPDI-Lifestyle scores were associated with lower risks of obesity-related cancer mortality and all-cause mortality. For obesity-related cancer incidence, an inverse association was observed overall, whereas sex-stratified analyses showed no statistically significant association among women. Compared with participants in the lowest tertile, those in the highest tertile had lower risks of obesity-related cancer incidence (HR: 0.82, 95% CI: 0.74–0.92), obesity-related cancer-specific mortality (HR: 0.66, 95% CI: 0.56–0.77), and all-cause mortality (HR: 0.64, 95% CI: 0.60–0.68). Each 1-SD increment in the hPDI-Lifestyle score was associated with 8, 18, and 22% lower risks of obesity-related cancer incidence, obesity-related cancer-specific mortality, and all-cause mortality, respectively.ConclusionHigher hPDI-Lifestyle scores were associated with lower risks of obesity-related cancer incidence, cancer-specific mortality, and all-cause mortality. These findings require confirmation in more representative populations and intervention studies before preventive implications can be established.

Gender-based violence (GBV) within LGBTQ+ communities raises challenges for protection order (PO) systems that were largely developed around heterosexual intimate partner violence (IPV) and physical harm. This paper compares the PO systems of California in the United States and the United Kingdom (UK) through a qualitative socio-legal approach. It examines how evidentiary standards, institutional bias, and narrow understandings of abuse affect LGBTQ+ victims’ access to protection. The analysis shows that identity-based abuse, coercive control, psychological violence, and threats of forced outing are often difficult to prove under conventional legal categories. California’s model formally permits LGBTQ+ applicants to seek relief, but it places heavy burdens on individuals through private filings, judicial discretion, and uneven enforcement. The UK system adopts broader definitions of domestic abuse and stronger multi-agency coordination, although bureaucratic complexity remains a limitation. The paper argues that effective reform requires expanded legal definitions of abuse, improved institutional awareness of LGBTQ+ IPV, and closer coordination between courts, police, social services, and community support organizations.

BACKGROUND:Coronary artery disease (CAD) remains a major cause of morbidity and mortality worldwide. Variants in the lipid metabolism gene SCARB1 may influence CAD susceptibility, but existing evidence is inconsistent. METHODS:We systematically searched PubMed, Embase, Web of Science, the Cochrane Library, and Scopus up to 13 February 2026 for case-control studies on SCARB1 polymorphisms and CAD risk. Three polymorphisms, rs5888, rs4238001, and rs10846744, were included. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were calculated under multiple genetic models. Subgroup analyses were conducted for rs5888 by sex, ethnicity, and clinical outcome. Heterogeneity was assessed using I2. RESULTS:12 eligible case-control studies involving 3947 CAD cases and 5076 controls were included. Overall, rs5888 was not significantly associated with CAD in any genetic model, either in the pooled analysis or in subgroup analyses by ethnicity and clinical outcome. In sex-stratified analyses, males carrying the TT genotype had a significantly lower CAD risk under the recessive model (OR = 0.73, 95% CI: 0.57-0.93), whereas no significant association was observed in females. No significant association was found between rs4238001 and CAD under any model. In contrast, the rs10846744 G allele was significantly associated with reduced CAD risk under the allelic (OR = 0.78, 95% CI: 0.64-0.94), dominant (OR = 0.68, 95% CI: 0.50-0.93), homozygote (OR = 0.65, 95% CI: 0.44-0.94), and additive models (OR = 0.80, 95% CI: 0.67-0.96). CONCLUSIONS:SCARB1 rs5888 showed a male-specific association with CAD, while rs10846744 showed a suggestive inverse association that requires further validation.

This research investigates the divergent and convergent methodologies employed in the construction of sacredness within Chinese and American religious art paintings, specifically examining the interplay between "divine narrative" and "humanistic expression." Through a comparative analysis, the study identifies dominant narrative strategies in Chinese religious art, characterized by adherence to established divine accounts, rich symbolism, and authoritative depictions of celestial beings, thereby prioritizing a "divine narrative" paradigm. In contrast, American religious art is analyzed for its emphasis on emotional resonance, personal spiritual journeys, and the relatable human experience of faith, aligning with a "humanistic expression" paradigm. The research explores the distinct theological underpinnings that inform these contrasting artistic approaches, revealing how cultural contexts shape the visual language and intended spiritual impact of religious artworks. Despite these differences, both traditions utilize art as a potent vehicle for religious pedagogy, fostering communal identity, and facilitating a connection with the divine. This study contributes a nuanced understanding of how diverse cultural and theological frameworks influence the artistic representation and experiential construction of sacredness, bridging existing gaps in comparative scholarship on religious art.