Ranks distinct AI gain and problem claims from the published record. Scores reward impact, independent source strength, scale, confidence, and recency.
Researchers developed a dynamic joint prediction system for incomplete immune reconstitution risk in people living with HIV using Bayesian joint modeling of longitudinal CD4+ counts and CD4/CD8 ratios from 21,862 patients across 31 Chinese provinces between 2003 and 2024.
On August 5, 2026, a peer-reviewed study in JCO Oncology Practice evaluated AI translation of three oncology clinical trial informed consent forms from English to Spanish, comparing DeepL Pro, ChatGPT-4o, and a medically trained model Med_English2Spanish against certified translations using five equivalence domains scored by two bilingual physicians.
On August 5, 2026, a viewpoint in the Journal of Participatory Medicine described how large language model chatbots and purpose-built companion agents are being used by millions for emotional support, distress processing, and relationship-like interaction, with 48.7% of people with self-reported mental health concerns reporting use for mental health support.
On 2026-08-04, a peer-reviewed cross-sectional study reported testing ChatGPT4, Gemini 2.0, Copilot, DeepSeek V3, and Grok 3 on 12 tracheostomy care questions, with three blinded laryngologists rating responses for accuracy, completeness, clarity, and sourcing, and readability measured with nine metrics.
A February 2026 narrative review in the International Journal of Medical Informatics synthesized studies from 2018 to 2025 on human-in-the-loop AI in healthcare. It found HITL approaches applied across diagnostic imaging, clinical decision support, patient monitoring, drug discovery, and research data analysis, with evidence of improved diagnostic accuracy, reduced medical errors, and increased clinician trust versus automated or traditional care.
Published February 20 2026, this peer-reviewed paper argues that while AI offers scalable mental health support, emerging patterns suggest those same systems may undermine wellbeing. It introduces the AI-IARA framework naming six capacities - Awareness, Interpretation, Intention, Action, Relational Agency and Autonomy - as essential for wellbeing when AI mediates experience.
A peer-reviewed philosophy paper published February 11, 2026 argues that false outputs from systems like ChatGPT, Claude, Gemini, DeepSeek and Grok should be understood not only as AI hallucinating at users, but as humans hallucinating with AI. Through distributed cognition theory, it describes how routine reliance on chatbots to think, remember and narrate can embed errors and reinforce users' own distorted beliefs.
By February 2026, 53 editors-in-chief or delegates in anaesthesiology and pain medicine completed a three-round modified Delphi to produce 59 statements on responsible LLM use. The group agreed LLMs may help with limited editorial and authorial tasks if fully disclosed and human-verified, but must not create original ideas, data, references, conclusions, full manuscripts, or make editorial or peer-review decisions.
ace