AI language models sharpen chest CT diagnoses, speeding surgical decisions
Peer-Reviewed Publication
Updates every hour. Last Updated: 3-Aug-2025 19:11 ET (3-Aug-2025 23:11 GMT/UTC)
A head-to-head study of five large language models (LLMs) shows that GPT-4, Claude-3.5 and Qwen-Max can correctly identify up to 75 percent of 13 thoracic diseases described in chest CT reports—outperforming earlier models and improving further with prompt design and fine-tuning. The findings point to faster, more reliable pre-operative planning for surgery.
A comprehensive review finds that GLP-1 receptor agonists, popular diabetes medications, show varying effects on cancer risk - protective against some cancers while potentially increasing risk for others. The findings emphasize the need for personalized treatment approaches.
A research team has developed a powerful two-stage screening pipeline that integrates hyperspectral imaging and metabolomic profiling to rapidly and accurately identify salt-tolerant crop mutants before visible symptoms appear.
This review elucidates the specific roles of the above mechanisms in the development of obesity-associated cancers at different sites, and outline intervention strategies to prevent obesity-associated cancers.
This review elucidates the dysfunction of B cells in tumor microenvironment, summarizes the discovery and mechanism of action of B cell ICs in tumors, and discusses urgent issues and future perspectives.
Researchers review controversial effects of fecal microbiota transplantation (FMT) on immune checkpoint inhibitor therapy. While some studies show improved cancer outcomes, others reveal no benefit or adverse effects, highlighting the complex relationship between gut bacteria and cancer treatment success.