All traces

In 55 realistic pediatric scenarios evaluated by the Skyer benchmark, ChatGPT-4.5-preview and Gemini-2.5_05-06 achieved higher triage accuracy and weighted scores than human triage experts, with acceptable consistency across repeats.

In a 20-question TB Q&A test generating 100 responses, GPT-5 produced the most suitable patient-education texts as measured by C-PEMAT-P among five LLMs.

In health professions OSCEs, AI applications improved grading efficiency, feedback speed, and consistency for structured observable tasks, with personalization as the dominant P4 precision education alignment observed by June 2025.

In 414 prostatectomy cases, the MMAI score derived from digitized pathology images was associated with long-term biochemical recurrence and metastasis, and when combined with genomic CCP score reached the highest discrimination for metastasis.

The Lanarkshire AI datacentre complex was presented as delivering jobs and prosperity while being powered entirely from on-site renewables with up to 1GW of new energy infrastructure by 2030.

In two Canadian diagnostic laboratories, AI-based PhenoMATRIX urine culture assessment enabled earlier availability of interpretable results and reduced time to result reporting by about 1.3 hours with automated PM+ release at a tertiary hospital and about 5.3 hours with earlier manual screening at a community lab.

GPT-4o integrated into systematic review screening substantially accelerated evidence synthesis for osteogenesis imperfecta biologics while maintaining high sensitivity.

When used thoughtfully in higher education, large language models can enhance teaching and learning by letting students generate and then critique outputs against primary sources.

Machine learning models analyzing nonlinear and high-dimensional clinical data have shown improved predictive discrimination for cardiothoracic surgical risk in selected cohorts compared with static traditional scores.

In simulated consultations about nipple discharge, AI chatbots recognized most clinical warning features and were rated safe in most responses.

Retrospective AI quantification of routinely available thin-slice chest CT reclassifies patients from CAC zero to positive, improving sensitivity for early subclinical atherosclerosis without additional radiation.

M4CXR achieved higher report consistency than ChatGPT-4o and cut reporting time from 179.2 seconds unaided to 16.3 seconds assisted when interpreting chest radiographs.

AI integration in radiologic practice improves triage and reduces report turnaround times while achieving diagnostic performance approaching or exceeding radiologists in chest imaging and breast cancer screening

Adopting AI tools in humanitarian operations was reported to improve health diagnostics, service quality, and analytical efficiency for multifactorial predictions in resource-limited and conflict settings.

Custom GPT models completed all QUIPS domain judgments and reduced data-extraction time from 30.4 to 5.7 minutes per study in rheumatology systematic reviews.