All traces

In a 34-case glaucoma reasoning test, LLM systems produced structured reasoning with weighted scores overlapping attending ophthalmologists and often included safety-critical diagnostic and management elements.

Researchers used genome language models Evo1 and Evo2 trained on 2 million bacteriophage genomes to design functional bacteriophage genomes, and a cocktail of the resulting viruses killed E. coli strains resistant to natural phages in lab dishes.

Large language models such as ChatGPT can provide personalized learning experiences when integrated into medical education.

AI-driven systems improve dietary tracking accuracy and enable personalized diet recommendations and disease-specific nutrition management in clinical and public health practice.

TrialTriage achieved perfect concordance with ground truth on 90 synthetic phase I oncology cases and reclassified ambiguous cases to definitive eligibility after capturing investigator email replies, processing cases in seconds compared to slower manual review.

General-purpose AI tool ChatGPT-4o achieved near-certified translation equivalence for English-to-Spanish oncology informed consent forms.

AI companion agents used for emotional support can ease loneliness and produce real symptom reduction for users with mental health concerns.

In a cross-sectional analysis of 5 leading chatbots, AI responses to 12 tracheostomy care questions were rated accurate and comprehensive, with Gemini 2.0 scoring higher on completeness than a senior laryngologist, indicating potential to support patient education where guidance is critical for safety.

Manus architecture using raw 3D CBCT data detected and correctly diagnosed 95% of jaw lesions in 97 patients, outperforming 2D panoramic inputs.

AI-assisted interpretation using Queen of Hearts software improved STEMI diagnostic accuracy, sensitivity, specificity, and interrater agreement among certified physician assistants interpreting 12-lead ECGs.

Projected gain that AI, integrated with telehealth and clinical decision support, could enable earlier identification of distress and more timely, safer triage for regional, rural and remote Australians.

AI applications in healthcare could expand access and improve disease surveillance and health system planning for marginalised populations in Sub-Saharan Africa.

AI software-as-a-medical-device platforms cleared since 2019 that estimate sleep parameters improve accessibility to obstructive sleep apnea diagnosis for patients unable or unwilling to undergo in-laboratory polysomnography.

A random forest model trained on 16S rRNA microbial community data from a field mesocosm predicted antifouling paint particle presence in sediment, with perfect detection of presence in test set and correct classification of all uncontaminated and 3 of 5 contaminated real-world Baltic Sea sites.

AI models for CT, MRI, and ultrasound diagnosis of urological cancers achieved higher pooled specificity and AUC than clinicians in a 110-study meta-analysis.