The modernization of legacy claims platforms is critical for payers that must sustain high-volume adjudication while integrating with an expanding ecosystem of vendors and delegated entities. This paper presents a distributed claim-centric platform that decomposes enrollment, eligibility, claims intake, adjudication orchestration, and payments into independently deployable microservices running on a container-based runtime. Container orchestration is used to scale claim-adjacent services, such as EDI ingestion, provider enrichment, and remittance generation, horizontally under peak loads while maintaining predictable performance and resilience. Long-running and exception-driven activities, including pends, manual reviews, and appeal cycles, are offloaded to workflow engines and asynchronous orchestration layers, ensuring that the core claims engine remains optimized for high-throughput adjudication. In addition, the platform publishes claim and accumulator events in near real time, enabling partner-facing services to consume, reconcile, and maintain synchronized benefit and cost-share views across third-party vendors and delegated entities. Empirical migration outcomes—covering scalability, failure isolation, processing latency, and reconciliation accuracy over comparable production windows—demonstrate that the proposed architecture delivers measurable improvements over a monolithic claim engine deployment, while aligning with emerging microservices migration and event-driven design practices.
Federated edge learning has emerged as a powerful paradigm for privacy-preserving analytics in large-scale agricultural and environmental Internet of Things (IoT) ecosystems, yet its real-time deployment remains constrained by multimodal heterogeneity, intermittent connectivity, and non-stationary field conditions. This paper introduces a Federated Edge Learning (FEL) framework with Multimodal Signal Fusion (MSF) that enables low-latency, privacy-preserving environmental monitoring directly at distributed farm and field nodes. The framework integrates adaptive modality encoders for soil, climate, CO2, and visual inputs with a lightweight cross-attention fusion module optimized for on-device inference. A heterogeneity-aware aggregation mechanism dynamically weights client updates based on data drift, node reliability, and resource availability, ensuring stable convergence under non-IID conditions. Implemented on distributed Jetson Nano and Raspberry Pi 4 devices with secure aggregation and quantized model exchange, FEL-MSF achieves up to 27% reduction in communication cost, 33% faster convergence, and a sub-second end-to-end latency compared with conventional FedAvg and centralized baselines, without sacrificing prediction accuracy. The results establish FEL-MSF as a scalable foundation for real-time, privacy-preserving intelligence in next-generation smart agriculture and environmental sensing networks.
Sexual minority people of color are subject to both racism and heterosexism. Prior studies have shown that, compared to people who report a single type of discrimination, those who experience intersectional discrimination exhibit heightened risk for mental health problems. The present study explored how living in a neighborhood with high coethnic density may either expose or shield sexual minority people of color from discrimination experiences. Participants (N = 90) residing in Los Angeles and New Orleans completed a survey as part of a larger study beginning in 2016. Their addresses were geotagged using 2010 census tract-level data to reflect their neighborhood ethnic composition. Participants reported their demographic characteristics and completed the two-stage Everyday Discrimination Scale. Results based on reversed odds ratios indicated that for every 10% increase in neighborhood coethnic density, participants were 49% more likely to report intersectional discrimination than heterosexist discrimination only. Additionally, for every 10% increase in coethnic density, participants were 20% more likely to report intersectional discrimination than racial discrimination only. Findings preliminarily suggest that living in coethnic neighborhoods may increase exposure to intersectional discrimination among sexual minority people of color. Given the study limitations, these findings should be interpreted with caution, and future larger studies should further examine this question. Interventions that buffer against the impact of intersectional discrimination are needed, and community-based interventions should aim to reduce experiences of intersectional minority stress. Additionally, public health research focused on neighborhood-level factors should integrate intersectional perspectives to promote equity in health.
The digital twins (DTs) are central to the next-generation communication system, and it allows real-time simulation and control. Their development under cross-domain ecosystems however reveals the fundamental weakness in interoperability, security and trust which is made worse by heterogeneous data model stipulations and antagonistic policies. This paper offers an analytical report of the DT standards environment, where major shortages are observed in security architecture and federated intelligence. To overcome these problems, we introduce a new framework of Intelligence Secure Edge Framework (SEIF) that incorporates Trusted Execution Environments (TEEs), blockchain and federated learning. Also, we introduce a road-map towards cross-domain standardization that is discussed stratified and propose a list of urgent research directions to make acquiring a secure and scalable deployment of DT in 6G and 6G.
Generative AI is increasingly embedded in enterprise decision-making processes. However, its adoption poses a variety of challenges for governance, such as so-called hallucinations, exposure of private data, policy violations, lack of explainability, and unclear accountability. This paper presents a Governance Framework for Human-in-the-Loop Enterprise Generative AI Systems. The framework consists of request classification, multidimensional risk assessment, policy management, explainability, risk-based human approval, audit logging for compliance checking, and monitoring of AI system performance throughout the complete decision-making AI workflow lifecycle. The framework is demonstrated by means of a case study in the financial services domain for the generation of regulatory reports, business summaries, and operational recommendations. A qualitative analysis evaluates the framework with respect to a set of relevant criteria, including governance coverage, human approval, policy compliance, explainability, audit completeness, and monitoring coverage. The results show that by embedding governance into the AI decision-making process of enterprises, transparent, accountable, and even risk-aware decisions can be made while keeping the efficiency of AI-driven decision-making. The architecture serves as a practical starting point for responsible AI deployment in various regulated domains and can be further extended with features such as adaptive policies, dynamic risk models, multi-agent governance, and/or federated oversight.