Ferdous Sohel

Modeling the impact of climate change in the semiarid maize cropping system through real-time sensor–GCM integration

Climate change threatens global food security by disrupting temperature and rainfall patterns, increasing the risk of reduced crop yields. While extensive research has explored these impacts worldwide, limited studies have quantified the combined effects of climate change and management strategies on maize productivity in Pakistan’s Peshawar region. This study addresses this gap by simulating maize yield predictions using the decision support system for agrotechnology transfer (DSSAT) cropping system modeling (CSM) CERES module integrated with real-time environmental data and advanced crop modeling techniques. We conducted a two-year field experiment (2020–21) calibrating and evaluating leaf area index (LAI), leaf weight (kg ha−1), pods weight (kg ha−1), aboveground biomass (AGB, kg ha−1), and yield (kg ha−1) in Peshawar, Pakistan. Two maize hybrids, hybrid-1 (SB92-K97) and hybrid-2 (SB-909), were evaluated. Climate change scenarios were constructed from five global climate models (GCMs) under representative concentration pathways (RCPs) 4.5 and 8.5, and comprehensive simulations tracked yield performance across baseline, near-, mid-, and far-future time periods (2006–2100). This process involved the development of an iterative recalibration module that adjusted model parameters in response to environmental feedback loops, resulting in refined predictions of anthesis, maturity, and yield, particularly under nitrogen (N)-limited scenarios. Our findings demonstrate that the model showed acceptable calibration, with RMSEs of 397 kg ha−1 for hybrid-1 and 411 kg ha−1 for hybrid-2, and accurate yield predictions with RMSEs of 68 kg ha−1 and 79 kg ha−1, respectively. The implementation of dynamic recalibration led to predictions of yield declines of 14% and 16% under RCP4.5 and RCP8.5, respectively, emphasizing the model’s adaptability in forecasting under increasing temperatures. We analyzed a thermal stress coefficient that captures maize sensitivity to extreme temperature fluctuations and elucidated the transient impacts of heat stress on long-term productivity. Sensitivity analysis showed that in 2021, yield was most influenced by temperature changes, which affected genetic parameters (P1, P2, P5, G2, G3, PHINT), with the highest yield occurring under stable temperatures. The climate projections indicated the highest increases in temperature, by 4 °C and 9 °C, under RCP 8.5. The irrigation demand is expected to rise 15–22% by 2100. The optimal strategy for maximizing maize yield was found to be 175 mm of irrigation combined with 240 kg N ha−1 fertilizer.

Chemoinformatic and machine learning fusion for designing silver nanoparticles with enhanced antimicrobial properties

Advancing Multimodal Fusion on Heterogeneous Medical Data with Hybrid Geometry Attention

Multimodal fusion learning (MFL) (a framework to jointly learn from heterogeneous data sources) has shown great potential in various fields such as Medicine, Science, and Engineering. It is extremely desirable in the medical domain, where we are faced with disparate data modalities such as imaging, clinical records, and omics. However, existing MFL strategies face several major challenges. First, they struggle to capture complex cross-modal interactions effectively, which in turn limits performance improvements. Second, they incur high computational costs, restricting their applicability in resource-constrained healthcare AI applications. Finally, they are often designed and evaluated for narrow, fixed modality configurations (e.g., imaging-only, or specific pairs such as image and omics), which limits evidence of their adaptability and generalizability to broader collections of heterogeneous medical modalities. To address these challenges, we propose a novel MFL framework – Cascaded Unified Representation Learning for Efficient Fusion Network (CURE) – a lightweight and scalable framework that progressively integrates various modalities through a novel efficient Hybrid Geometry Aware Fusion layer (HyFuse), where each HyFuse layer is sequentially learned for each modality, making the framework adaptable and generalizable. Within HyFuse, an efficient residual convolution module captures rich multi-scale features to ensure cost-effective learning, while a hybrid-space aware attention mixer learns coarse-to-fine structural cues to better preserve cross-modal relationships. Complementary learnable late-fusion and shared-information refinement modules are then employed to learn robust, modality-order-invariant shared features, which in turn yields consistent performance improvements. Extensive evaluations on 16 public datasets show that CURE outperforms leading multimodal fusion methods (e.g., DRIFA-Net and HEALNet), boosting performance by up to ≈ 3.97% and lowering computational costs by up to ≈ 87.8%, ensuring more effective and reliable predictions.