This review summarizes recent advances in wind speed forecasting using artificial intelligence. It systematically analyzes multi-scale signal decomposition methods and intelligent model fusion strategies, highlighting their effectiveness in addressing the challenges of nonstationary and multiscale wind data. By identifying key methodological patterns and gaps, the review offers actionable insights for designing more accurate, efficient, and robust AI-based wind forecasting systems.

A study analyzing sun protection test data from 1,036 East Asian participants found significant correlations between skin color, seasonal changes, and UV tolerance. Results show higher UV resistance in winter and greater sensitivity in lighter-skinned individuals, offering crucial insights for personalized sunscreen development.

Researchers have extracted a novel polysaccharide fraction named AJDW from the flower of Albizia julibrissin Durazz. (AJD). AJDW could not only inhibit the growth and proliferation of pancreatic cancer in vitro and in vivo, but also suppress pancreatic cancer migration, block cell cycle and induce ROS-mediated apoptosis by reducing mitochondrial membrane potential linked to PI3K/Akt/mTOR signaling pathway.

This study leverages cryo-correlative light and electron microscopy (cryo-CLEM) with triple fiducial markers to achieve precise spatial alignment between fluorescence-localized synaptic components and their ultrastructural details in cryo-preserved neurons. The method identified unique electron-dense patterns within synaptic clefts and resolved high-resolution states of presynaptic vesicles (e.g., docked, primed, or fused), demonstrating cryo-CLEM’s capability to bridge molecular specificity (via LM) with nanoscale structural dynamics (via EM). This approach advances mechanistic studies of synaptic plasticity and neurodegenerative pathologies by correlating functional protein localization with ultrastructural context.

This protocol refines co-immunoprecipitation (co-IP) for lipid droplet (LD)-specific protein interactions by isolating purified LDs prior to protein extraction, eliminating contamination from cytosolic proteins in traditional lysate-based methods. Key steps include high-purity LD isolation, optimized solubilization of LD-associated proteins, and enhanced co-IP specificity. This approach uncovers dynamic interactions (e.g., perilipin complexes) under metabolic stimuli, advancing mechanistic studies of LD dysfunction in obesity and fatty liver disease.

A 28-day clinical trial with 31 subjects compared a 0.1 % pterostilbene skincare emulsion to a control using a double-blind, split-face design. Advanced tools and self-assessments showed the pterostilbene emulsion significantly improved skin elasticity, reduced wrinkles, and increased epidermis thickness and collagen intensity. It outperformed the control in multiple metrics (p < 0.05) and gained higher subject satisfaction, highlighting its potent anti-aging effects and value in cosmetics.

Foundation models in molecular biology, leveraging their success in NLP and image generation, are revolutionizing the understanding of multi-level molecular correlations by training on vast datasets encompassing RNA/DNA/protein sequences, single-cell transcriptomics, and spatial transcriptomics. These models decode intricate relationships (e.g., gene regulatory networks, protein interaction hubs) to predict functions, design therapeutics, and infer spatial tissue dynamics. Current frameworks include ESM-2 (protein structure-function prediction), scGPT (single-cell data integration), and DNABERT (genomic variant interpretation). Future directions emphasize multimodal integration (combining sequences, structures, and omics), interpretable attention mechanisms for biological insights, and scalable architectures for high-resolution spatial-temporal data. Addressing data heterogeneity and model generalizability will unlock precision biomedicine applications.

Recently, a research team led by Dr. Tida Ge from Xinjiang University and Ningbo University conducted a pot experiment to explore this question. The study selected peas as the experimental crop and examined two types of microplastics: traditional microplastics (polypropylene (PP) and polyethylene (PE)) and biodegradable microplastics (polycaprolactone (PCL) and polybutylene adipate terephthalate (PBAT)). These were added to the soil at doses of 0%, 0.1%, and 1% (w/w) to observe the growth status of peas during three critical growth stages: seedling, flowering, and maturity. The study also analyzed soil nutrients, microbial activity, and community changes. The study has been published in Frontiers of Agricultural Science and Engineering (DOI: 10.15302/J-FASE-2025626).