AM-IA brings together news, research papers and applications relating to artificial intelligence published around the world into a single feed — and makes them available in your European language at the click of a button.
Scroll to browse · Click to filter
A study shows that Google's AI summaries favor commercial publishers, and that nearly one in two quoted claims is not supported by its source.
A survey of clinicians in two departments shows that prediction is mainly important in complex cases, that it is discussed among colleagues, and that use is changing with experience.
960 conversations of 20 simulating people in crisis show that recognizing distress and speaking with empathy is not enough to respond safely.
Interpreting a genetic variant requires crossing recommendations, bases and calculations. An agent who assembles executable skills significantly improves the accuracy and speed of 45 professionals.
UCLA researchers tested their model on 16 diseases and over 2 million patients in two health systems never seen in training, setting themselves at the realistic alert threshold for screening.
CEDAR, a self-checking chain of LLM agents, is ahead of machine learning on paediatric type 1 and 2 diabetes cases, with only 4 erroneous quotes out of over 70,000.
Caltech, Ohio State, LSU, Indiana University and UTHEalth Houston will use Meta's models, including open versions deployed on site, on patient files, scanners and MRIs.
Driven across 28 million images and 70,000 volumes from nine modes, it reaches the state of art on 24 test benches, from segmentation to reporting.
The OpenGray test bench shows that LLM-based agents, well rated overall, report only a quarter to half of the anatomical contradictions they should raise.
CAR-T therapy shortened by 70%, 35,000 antibodies generated in one month, regulatory files written in four hours: Anthropic takes stock of Claude in the development of drugs.
The systems that entrust questionable cases to physicians are based on uncertainty measures. Of three neurological cohorts, these measures primarily reflect class imbalance, resulting in a "safety illusion".
Driven by 1,632 patients classified as BI-RADS 4 and 5, the multimodal model reaches an AUC of 0.72, above single or text versions, and shows on which areas it is based.
Nine models tested change their diagnosis to match the mistaken conviction of the patient, sometimes even for an invented disease. The " rare self-proclaimed patient" profile wins 43% of the time.
NPC-RubricBench evaluates 30 models at each stage of support. Medical models forget decisions more often than generalist models.
Based on Rosenhan (1973)'s experience, a study of 20,000 responses from five models showed that with identical symptoms, the diagnostic label leans towards hospital maintenance.
The test detects BRAF and RET gene rearrangements in solid tumours; DNA-associated RNA sequencing increases patients' access to targeted therapies by 24%.
An imitation learning framework anchored on anatomy obtains standard kidney, vesicle and liver cuts from 16 participants never seen in training.
142 questions from australian repositories (reimbursement, triage, vaccination): most errors come from a poorly-memorized rule, not badly applied, and some models are wrong with almost every other tip.
Tested on more than 5,000 simulated chat tours in India, South Africa and Kenya, three health chatbots on WhatsApp yield when the user relativizes an emergency.
Brazilian researchers also show that enriching the reference annotations changes the scores displayed without changing the ranking of the document search systems.