We also have X and podcasts
Daily AI medical research digest in English
Researchers have developed a machine learning approach that combines transfer learning with physical knowledge to design high-performance conjugated polymers for low-voltage organic electrochemical transistors, overcoming the fundamental challenge of limited training data in emerging materials fields. Original paper: Transfer-learning guided…
Researchers have developed BUSGen, a generative AI model trained on 3.5 million breast ultrasound images that outperforms board-certified radiologists in early breast cancer detection. The model also generates synthetic data that could enable privacy-preserving research. Original paper: A foundation generative…
A new lightweight artificial intelligence model combined with salivary metabolic analysis achieved 91.9% accuracy in identifying patients with both type 2 diabetes and periodontitis in just 0.7 minutes per test, offering a rapid non-invasive screening tool. Original paper: Lightweight liquid…
A major study combining large-scale genetic analysis with advanced deep learning has identified hundreds of new genetic variants associated with asthma in Europeans, paving the way for improved risk prediction and personalized medicine approaches. Original paper: Deep learning and statistical…
A new deep learning model called SuRImage uses smartphone photos of surgical tissue to instantly diagnose lung cancer invasiveness, potentially eliminating 30+ minute waits for traditional analysis while enabling junior surgeons to match senior-level diagnostic performance. Original paper: Deep learning…
A secondary analysis of the LUNAR trial reveals that neoadjuvant 177Lu-PSMA-I&T delivers well-tolerated radiation doses while significantly improving progression-free survival in patients with low-volume oligometastatic prostate cancer when combined with stereotactic radiotherapy. Original paper: Dosimetry Analysis of 177Lu-PSMA-I&T in Patients…
Decipher-MR represents a significant advance in medical imaging AI, offering a reusable foundation model trained on 200,000 diverse MRI scans that consistently outperforms existing approaches across disease detection, anatomy localization, and cross-modal retrieval tasks. Original paper: Decipher-MR: a vision-language foundation…
A new study demonstrates that custom AI models fine-tuned on institutional data can generate radiology impressions comparable to human radiologists, while general-purpose language models are unsuitable for the task. Original paper: Comparison of AI-generated radiology impressions: a multi-stakeholder evaluation. —…
Researchers have developed a label-free method using autofluorescence imaging and deep learning to diagnose non-small cell lung cancer subtypes without conventional tissue staining, achieving expert-validated accuracy while reducing diagnostic time and costs. Original paper: Label-free pathological subtyping of non-small cell…
A new validated risk model combining gestational age, postmenstrual age, and vascular severity assessment can predict treatment-requiring retinopathy of prematurity (TR-ROP) within 2 weeks, potentially reducing screening examinations by up to 39% without missing critical cases. Original paper: Precision Risk…
