Researchers at the Boyce Thompson Institute and Cornell University have developed AIMe (AI Molecule Explorer), a neuro-symbolic AI tool that predicts, organizes, and searches mass spectra for more than 100 million known small organic molecules—creating a searchable map of chemical space roughly a thousand times larger than existing experimental libraries. Applying AIMe to gut microbiome data, the team identified a previously unknown, ring-shaped polyamine compound also found in human fecal samples, and used the tool to annotate millions of unidentified spectra in a major public mass spectrometry database.

To overcome the fragility of soft sensing materials under long-term cyclic loading, researchers have developed a highly durable "eutectogel" utilizing a novel nanoscale cavitation mechanism. By integrating this fatigue-resistant material into a smart glove with deep learning algorithms, the team achieved a 99.9% accuracy rate in real-time sign language recognition. This innovation paves the way for seamless communication and advances the durability of next-generation human-machine interfaces.

This study explored the link between cuproptosis, a type of cell death, and unexplained infertility (UI) using gene expression data. Researchers analyzed 27 cuproptosis-related genes in endometrial samples from UI patients, identifying two molecular clusters linked to immune cell infiltration. They tested seven machine learning models to predict UI, finding the support vector machine (SVM) model most effective, with high accuracy (AUC 0.918). The study suggests five cuproptosis-related genes could be potential biomarkers for UI, offering insights into the disease's molecular basis and aiding in future diagnostic and treatment strategies.

Researchers in China developed a novel two-terminal β-Ga₂O₃ photo-synapse based on self-trapped holes engineering, achieving low weight update nonlinearity and excellent anti-interference characteristics. The device enables diversified in-sensor computing tasks from image classification to motion recognition, demonstrating great potential for neuromorphic machine vision systems in the artificial intelligence era.