Nazeer Hussain University

Global food security faces escalating threats from climate variability and resource constraints. Accurate crop yield forecasting is essential; however, existing methods frequently overlook complex spatial dependencies driven by climate teleconnections, such as the ENSO, and lacks rigorous uncertainty quantification. This paper presents HSE-GNN-CP, a novel framework integrating heterogeneous stacked ensembles, graph neural networks (GNNs), and conformal prediction (CP). Domain-specific features are engineered, including growing degree days and climate suitability scores, and explicitly model spatial patterns via rainfall correlation graphs. The ensemble combines random forest and gradient boosting learners with bootstrap aggregation, while GNNs encode inter-regional climate dependencies. Conformalized quantile regression ensures statistically valid prediction intervals. Evaluated on a global dataset spanning 15 countries and six major crops from 1990 to 2023, the framework achieves an R2 of 0.9594 and an RMSE of 4882 hg/ha. Crucially, it delivers calibrated 80% prediction intervals with 80.72% empirical coverage, significantly outperforming uncalibrated baselines at 40.03%. SHAP analysis identifies crop type and rainfall as dominant predictors, while the integrated drought classifier achieves perfect accuracy. These contributions advance agricultural AI by merging robust ensemble learning with explicit teleconnection modeling and trustworthy uncertainty quantification.

Coordinated transmission-distribution (T&D) optimization with AC power flow is computationally prohibitive for real-time dispatch. This paper proposes a physics-informed graph neural network (PI-GNN) surrogate that replaces iterative AC-OPF solvers inside a bilevel T&D co-optimization framework. A heterogeneous message-passing GNN mirrors the physical grid topology, while differentiable power-flow projection layers enforce Kirchhoff's laws and operational limits to operationally acceptable residual levels (power balance residual < 3.2 & times; 10(-4 )p.u., voltage deviation < 6.3 & times; 10(-3 ) p.u.), with a Newton-Raphson safety-net fallback guaranteeing hard feasibility for the rare ( < 0.05% in-distribution) cases that exceed user-defined thresholds. The surrogate is embedded in an end-to-end differentiable bilevel formulation whose gradients are validated against analytical implicit-function-theorem sensitivities ( similar to 3% relative error), eliminating iterative Jacobian computation at inference time. Experiments on three coupled TS-DS benchmarks spanning 448 to 3,334 buses with five independent seeds demonstrate voltage MAE of 0.0053-0.0079 p.u., generation cost gaps below 0.41%, constraint violation rates under 0.9%, and GNN inference times of 0.05-0.38 ms-over 4, 000 & times; faster than iterative AC-OPF solvers-with end-to-end bilevel dispatch completing in 4.2-12.7 ms. The PI-GNN consistently outperforms DNN-OPF, GNN-OPF, and DC3-OPF baselines across all system sizes, and its accuracy advantage widens on larger networks.

Maternal mortality in Bangladesh remains a critical public health challenge, with recent evidence indicating stagnation in mortality reduction despite expanded facility-based delivery and skilled birth attendance. Accurate identification of high-risk cases is essential to enable targeted intervention and resource allocation. This study develops an interpretable machine learning framework for maternal mortality prediction using the nationally representative Bangladesh Maternal Mortality Survey 2016 (BMMS-2016). A comprehensive data integration and feature engineering pipeline was implemented across demographic, socioeconomic, and maternal healthcare domains. Given the severe class imbalance inherent in maternal death outcomes, multiple resampling strategies---including Tomek Links undersampling, SMOTE, ADASYN, and CTGAN---were systematically evaluated in conjunction with diverse classifiers and ensemble methods. Among all configurations, Random Forest combined with SMOTE achieved the best overall performance (ROC-AUC: 0.9635; Accuracy: 0.9016; F1-score: 0.8928), demonstrating superior precision--recall balance suitable for rare-event prediction. Tree-based ensemble models consistently outperformed baseline classifiers. Model interpretability was ensured through SHAP analysis, revealing region type, maternal age, and administrative division as the most influential predictors and highlighting substantial geographic and demographic disparities. An interactive Tableau dashboard translates predictive insights into accessible visual analytics for policy and decision support. The findings underscore the importance of explicit imbalance handling and explainable artificial intelligence frameworks in maternal health risk modeling, offering a scalable and transparent approach for data-driven maternal mortality reduction strategies.

Silver nanoparticles (AgNPs) have attracted considerable interest in pharmaceutical and biomedical areas due to their distinct physicochemical characteristics and extensive biological properties. Nevertheless, the current synthesis techniques usually require the use of toxic chemicals and energy-intensive processes, making them unsustainable and impractical for biomedical applications. Consequently, this study sought to explore the synthesis of silver nanoparticles using the plant Moringa oleifera leaf extract as a green reductant and stabilizer for silver nanoparticles. Silver nanoparticles were characterized by means of UV–visible spectrometry, FTIR, XRD, SEM, EDS, DLS, and zeta potential. Results showed a characteristic surface plasmon resonance peak at 432 nm. The XRD studies showed that the nanoparticles have crystallinity properties with crystallite size ranging between 18–22 nm, whereas the SEM studies displayed mainly spherical shape of the particles. The colloidal stability was high with a zeta potential of −31.4 mV. The biological studies indicated good antimicrobial activity against Staphylococcus aureus, Escherichia coli, Pseudomonas aeruginosa, and Klebsiella pneumoniae, as there were inhibition zones ranging between 17.9 and 22.8 mm. Moreover, the AgNPs had antioxidant property (81.6%) as well as cytotoxicity towards MCF-7 cell lines whereas no toxicity was observed towards HEK-293 normal cells. This suggests that Moringa oleifera AgNPs can be used as an efficient nanoparticle-based pharmaceutical platform in the future.

The rapid diffusion of generative artificial intelligence (AI) is reshaping how organizations design and deliver sustainability-oriented marketing communications, yet little empirical evidence exists on whether such AI-generated content builds genuine consumer trust or brand value. This study develops and tests an integrated model examining the influence of Generative AI-Enabled Sustainable Marketing (GAISM) on Green Brand Equity (GBE), with Green Brand Trust (GBT) proposed as a mediating mechanism and AI Transparency (AIT) proposed as a moderator of the GAISM–GBT relationship. Grounded in signaling theory, brand equity theory, and the stimulus-organism-response framework, five hypotheses were formulated and tested using Partial Least Squares Structural Equation Modeling (PLS-SEM) on survey data collected from consumers with direct experience of AI-assisted digital platforms and sustainability-related marketing communication. The measurement model demonstrated strong reliability and validity, with all constructs exceeding recommended thresholds for Cronbach's alpha, composite reliability, and average variance extracted, and discriminant validity confirmed via the Fornell-Larcker criterion and HTMT ratio. Structural model results supported all five hypotheses: GAISM significantly enhanced GBT (β = 0.621) and GBE (β = 0.377); GBT strongly predicted GBE (β = 0.505) and partially mediated the GAISM–GBE relationship (indirect effect β = 0.314); and AI Transparency positively moderated the GAISM–GBT relationship (β = 0.182), with simple slope analysis confirming that this relationship strengthened progressively from low to high transparency. The model explained substantial variance in both GBT (R² = 0.674) and GBE (R² = 0.676). These findings demonstrate that generative AI, when deployed with genuine transparency and authenticity, can meaningfully strengthen consumer trust and brand equity in sustainability contexts rather than triggering skepticism or perceived greenwashing. The study contributes to emerging AI-marketing and green branding literature and offers practical guidance for organizations seeking to responsibly integrate generative AI into sustainability communication strategies.