Smarter skies: A new AI model turns street cameras into rainfall sensors
Peer-Reviewed Publication
This month, we’re focusing on artificial intelligence (AI), a topic that continues to capture attention everywhere. Here, you’ll find the latest research news, insights, and discoveries shaping how AI is being developed and used across the world.
Updates every hour. Last Updated: 5-Nov-2025 13:11 ET (5-Nov-2025 18:11 GMT/UTC)
Medulloblastoma is the most common malignant pediatric brain tumor with a high risk of metastasis and poor survival outcomes. To delineate the metastatic microenvironment, researchers in China have developed an explainable machine learning model that identifies key immune cells and cytokine markers linked to tumor dissemination and prognosis. Their model offers a transparent, data-driven approach that could help clinicians better assess risk and personalize treatment for children with this life-threatening disease.
As the most seismically active regions in the world, oceanic subduction zones show contrasting seismicity in different regions. A review by a team from Nanjing University explored the relationships among the thermal structure, metamorphism, deformation and fluid activity of oceanic subduction zones. The thermal structure of subduction zones not only controls dehydration depths of hydrous minerals, but also affects the mechanical coupling state of the subduction plate interface, eclogitization of the subducted oceanic crust, and phase transition of olivine in the subducted lithospheric mantle. Fault stick-slip behavior is the major mechanism of shallow earthquakes (<70 km), while dehydration embrittlement of hydrous minerals and other mechanisms may trigger intermediate-depth earthquakes (70–300 km). The amounts of earthquakes decrease sharply below 160 km in warm subduction zones, but below ~300 km in cold subduction zones where hydrous minerals release water at greater depths. Cold slabs could carry water down to depths >300 km, resulting in localized water enrichment in the mantle transition zone. More studies are needed to decipher how the fluid activity triggers slow earthquakes and deep-focus earthquakes. The review is published in Science China Earth Sciences.
A research team from Kumamoto University has developed a promising deep learning model that significantly enhances the accuracy of subgraph matching — a critical task in fields ranging from drug discovery to natural language processing.