Category: Artificial Intelligence
- Interpretable AI supports small liver-lesion assessment on multiphase CT ()
Published October 30, 2026. Researchers describe dLI-RADS, an interpretable AI method for estimating hepatocellular carcinoma probability from multiphase CT scans. It combines neural-network predictions with quantitative measurements related to established LI-RADS imaging features. The retrospective multicenter study used training, internal evaluation and external evaluation groups of 143, 62 and 133 patients, respectively. Lesions larger than […]
- Large screening study compares AI-assisted cytology and HPV testing strategies ()
Published October 30, 2026. A Nature Communications study compares cervical screening strategies using records from 284,002 women screened in Wuhan, China. It evaluates artificial intelligence-assisted liquid-based cytology alongside high-risk HPV testing. The researchers modeled nine strategies, comparing detection of CIN2 or worse—precancerous changes or more serious disease—with referral for colposcopy. After correction for verification bias, […]
- Agentic AI helps human reviewers identify immunotherapy adverse events in clinical notes ()
Published October 30, 2026. A Nature Cancer study published October 29 evaluates an AI system that helps reviewers identify immune-related adverse events in oncology notes. It extracts event presence, timing, severity, treatment attribution and diagnostic certainty across six toxicity types. Retrospective testing produced a macro-averaged F1 score of 0.92 for current-event detection, but only 0.66 […]
- AbOptimizer uses AI to improve multiple antibody properties in laboratory tests ()
Published October 29, 2026. Researchers have introduced AbOptimizer, an artificial intelligence framework that searches for antibody variants with several desirable properties at once. The study appeared in Nature Communications on October 26. The system combines protein language models with reinforcement learning to choose editing sites and amino-acid changes. It can edit several positions together and, […]
- Impression-R1 improves radiology summary writing in reader study ()
Published October 29, 2026. Impression-R1 is a specialized reasoning model designed to write diagnostic summary sections from radiology findings. Wang and colleagues describe its development and evaluation in Nature Communications, published October 28. The model combines supervised learning, clinically conditioned reasoning and radiology-specific reward optimization. Evaluation covered 11,884 cases from six institutions and three public […]
- Clinical reasoning AI benchmark finds accuracy gains alongside more false positives ()
Published October 29, 2026. A diagnostic AI benchmark finds that stronger reasoning can improve primary diagnoses while increasing false positives. Wu and colleagues report the results today in Nature Communications. The AIMS-LDR benchmark uses long electronic health records and compares six matched pairs of reasoning-enabled and standard language models across 12 specialties. Reasoning models produced […]
- Deep learning predicts IDH1 mutations from leukemia bone marrow images ()
Published October 29, 2026. A deep-learning model predicts IDH1 mutation status from digitized bone marrow smears in acute myeloid leukemia (AML). Eckardt and colleagues report the multicenter study today in npj Precision Oncology. The researchers used whole-slide images from 436 patients across four centers in Germany and the United States. Their attention-based multiple-instance learning approach […]
- Inflammatory markers improve research models for mild frailty classification ()
Published October 29, 2026. Blood inflammatory markers may add information to research models that classify mild frailty. A study published in BMC Geriatrics on October 28 examined participants in the GAINS cohort using clinical assessments and a panel of 27 serum inflammatory markers. The initial analysis included 165 participants, of whom 31 met the study’s […]
- ProBEiT combines protein sequence and structure for efficient engineering predictions ()
Published October 28, 2026. An October 26 paper in Communications Biology introduces ProBEiT, a protein language model that combines amino-acid sequences with structural information. Li and colleagues aim to improve protein-property prediction without relying solely on increasing model size. ProBEiT jointly represents sequences and tokenized structures through an architecture that specializes parts of the model […]
- AI-assisted screening identifies TFEB activators for adipose remodeling in mice ()
Published October 28, 2026. An October 27 study in Communications Biology combines mechanistic experiments with AI-assisted image screening to investigate healthier remodeling of adipose tissue. Zeng and colleagues focused on TFEB, a transcription factor involved in cellular regulation. In mice fed a high-fat diet, deleting Tfeb in myeloid cells worsened oxygen deprivation, fibrosis and insulin […]