Using MODIS observations from 2003 to 2024, researchers evaluated candidate regions and seasonal windows for marine cloud brightening (MCB) worldwide by examining how key cloud properties varied with aerosol optical depth (AOD). Among five representative ocean regions, the North Pacific during boreal autumn (SON) showed a favorable coordinated pattern: cloud fraction and cloud optical thickness were positively sensitive to AOD, whereas cloud droplet effective radius was negatively sensitive. The study provides an observational basis for further evaluation of MCB candidate regions and seasonal windows. The findings were published in Science China Earth Sciences.

Many vitreoretinal disorders, including retinal detachment (rhegmatogenous, exudative, or tractional), macular holes, proliferative vitreoretinopathy, diabetic retinopathy, and vitreous hemorrhage, often require vitrectomy with retinal repair, followed by intraocular tamponade to stabilize the retina and preserve ocular integrity. The tamponade agents most used in clinical practice are expansile gases and silicone oil. Although these materials can effectively support postoperative retinal reattachment, they remain limited in terms of long-term stability and biosafety. Complications such as cataract formation, elevated intraocular pressure, intraocular inflammation, and retinal injury associated with silicone oil emulsification may still occur. To address this challenge, Prof. Jianhai Yang and Prof. Wenguang Liu (Tianjin University), together with Prof. Xiaorong Li and Prof. Xiaoli Xing (Tianjin Medical University Eye Hospital) developed an injectable zwitterionic polymer hydrogel based on poly(2-methacryloyloxyethyl phosphorylcholine) (PMPC).

Published in Engineering, this research from PolyU and Guangzhou Institute of Industrial Technology develops a ConvLSTM-Fire deep learning model to repair broken sensor data and reconstruct 2D building temperature fields. Paired with a ChatGPT-4o-based LLM self-driven agent, the hybrid system converts real-time fire thermal data into actionable emergency guidance. Quantitative tests prove full 2D temperature inputs greatly improve the accuracy of LLM firefighting decisions, supporting voice-aided on-site response for occupants and firefighters.

Published in Engineering, this study introduces HCGCDM, an LLM-powered cognitive diagnosis model built on heterogeneous concept graphs. Unlike traditional tools that only capture prerequisite knowledge links, it automatically mines three types of concept relations via RAG. Tailored graph encoders and adaptive weighting boost student proficiency assessment accuracy, with ablation and LLM comparison experiments verifying its effectiveness and the impact of LLM capability on diagnostic performance.

Published in Engineering, this report introduces LLM-IDA, a three-tier multi-agent LLM framework for automated industrial design. It divides design workflows into four functional modules to address LLM hallucination and poor robustness via closed-loop code generation and self-verification. Benchmarked against RAG with pass@10 metrics and ablation tests, the modular system cuts manual engineering overhead and supports cross-domain mechanical design optimization, with its current limits and future research directions covered.

Published in Engineering, this news report covers a comprehensive survey exploring how large language models tackle long-standing flaws in two mainstream autonomous driving frameworks: modularized pipelines and end-to-end neural networks. It details LLM applications across perception, prediction, planning and data construction, lists core deployment barriers like hallucination and latency, and unpacks ongoing debates over AGI and the future architecture of intelligent driving systems.