Autonomous driving systems increasingly rely on data-driven approaches, yet many still struggle with reasoning, handling rare scenarios, and transparently explaining their actions. A new study introduces DriveMLM, a multi-modal large language model framework that aligns language-based reasoning with structured behavioral planning states, enabling full closed-loop driving in realistic simulators. By integrating multi-view images, LiDAR inputs, traffic rules, and natural-language instructions, DriveMLM generates both driving decisions and human-readable explanations that map directly to vehicle control. The system significantly improves safety, adaptability, and interpretability, demonstrating how large language models (LLMs) can advance the next generation of autonomous driving technology.

Understanding how sound travels through the middle ear is essential for designing reliable hearing implants. Traditional measurements of the middle ear transfer function (METF) can be affected by inner ear impedance and surgical manipulation, limiting their accuracy. This study introduces a new technique that eliminates inner ear interference while precisely capturing stapes footplate vibrations at multiple points. By accessing the vestibular side of the stapes through a trans-petrous approach, the researchers generated a stable and reproducible METF reference range using human temporal bone specimens. This refined method offers more reliable data for evaluating implant performance and enhances the biomechanical understanding of auditory transmission.

Eustachian tube dysfunction often determines whether a routine ear infection clears quickly or develops into a persistent condition. This study reveals that the mitochondrial enzyme SIRT3 plays a crucial protective role during acute inflammation. Using a mouse model of otitis media, the researchers show that loss of SIRT3 transforms a typical inflammatory reaction into a more severe pathological process—characterized by excessive mucus buildup, ciliary damage, increased resistance to tube opening, and impaired mucociliary transport. These findings highlight SIRT3 as an unrecognized stabilizer of middle-ear physiology and help explain why some infections resolve smoothly while others progress toward chronic disease.

A comprehensive review in Current Molecular Pharmacology highlights how machine learning, molecular dynamics simulations, and hybrid AI-MD platforms are revolutionizing antimicrobial discovery. These computational strategies can drastically reduce development time and costs while enabling the design of novel antibiotics and antimicrobial peptides. The authors propose an integrated “computation-experiment-clinical translation” framework to overcome current bottlenecks and combat the global AMR crisis.

A study of 13,126 participants from the UK Biobank found that individuals diagnosed with diabetes after age 50 had a lower risk of all-cause dementia compared to those diagnosed before 50. The protective effect was strongest among those with high genetic risk for dementia or carrying the APOE ε4 allele. These results highlight the importance of early diabetes management and personalized prevention strategies for high-risk groups.

Helical pulses, shaped like flying electromagnetic conches, long predicted but never directly observed, have now been experimentally realized in both optical and microwave bands. The work shows that light can twist through space and time within a single cycle, forming truly space time nonseparable wave packets. This breakthrough confirms a decades old theoretical prediction and opens new paths for ultrafast communication, topological field control, and advanced imaging.

A research team from the Songshan Lake Materials Laboratory has developed an AI-guided "Recommendation System" to discover new metallic glasses (MG). By combining element embeddings learned from Wikipedia by a language model with graph neural networks analyzing hidden material relationships. This approach addresses longstanding challenges related to the vast chemical space and limited experimental datasets, opening new horizons for materials design and accelerating the development of next-generation MGs.

Schematic of the early-stage characteristics of house dust mites-induced bronchial fibroblasts, which are characterized by suppressed extracellular matrix production alongside activated inflammatory signaling.