Ten Rules for Keeping AI Honest in Computational Biology
A new PLOS Computational Biology perspective sets out ten practical rules defining where AI assistance in computational biology is legitimate ...
Tag: Large Language Models
AI Decision Aid for Juvenile Arthritis Scores Well on Safety but Diverges on Treatment Choices
A feasibility study of a large language model decision aid for juvenile idiopathic arthritis found high ratings for safety and ...
Integrated digital health prototype shows generally good usability in healthy volunteers
A formative laboratory study of 10 healthy volunteers found generally good usability for a platform integrating wearables, conversational coaching and ...
Chatbots Deliver Less Diverse Knowledge Than Web Search, Study Finds
A study of 27 language models across 155 topics found that every tested chatbot presented less varied information than Google, ...
Structured prompts improve Chinese LLM radiology impressions in retrospective validation
A retrospective study found that structured prompts improved several MDQA measures for Chinese LLMs generating diagnostic impressions from liver MRI ...
ChatGPT configuration led expert ratings of dental radiology advice
In an expert evaluation of 60 patient-facing dental radiology answers, a reasoning-enabled ChatGPT configuration received the strongest ratings for accuracy, ...
Single-run benchmark finds AI liver-injury extraction errors
A single-run benchmark of 120 liver injury narratives found that AI systems produced parseable output while retaining potentially causality-distorting errors, ...
AI improves detection but may reshape clinical reasoning
A critical narrative review finds that AI improves detection in mammography and colonoscopy but does not improve complex diagnostic reasoning, ...
LLM study of 67 abstracts finds strong local QA, weak global consistency
An evaluation of 67 complete scientific abstracts found high automated scores for local question-answering checks but only 0.07 for document-wide ...
Agentic AI framework reaches 91.4% resolution success in cloud benchmarks
A cloud-native agentic AI framework combining reinforcement learning, large-language-model artifact generation and policy-as-code controls achieved 91.4% mean resolution success across ...
Artificial intelligence for symptom assessment and patient-reported outcomes in neuro-oncology
A narrative review of 53 studies finds that artificial intelligence can extract and analyze brain-tumor symptom data, but limited validation ...
Researchers propose computational framework for citizen participation discourse
A theoretical article proposes and partially specifies a three-level computational framework for analyzing citizen participation through content, interactional synchrony and ...
LLMs underperform specialized platform in 475-variant clinical classification
A study of 475 variants in 185 genes found that four general-purpose language models achieved 69.0% mean concordance with expert ...
AI chatbots differ in reliability, usefulness and readability for dental implant questions
A comparative study of 99 dental-implant questions found that DeepSeek led in expert-rated reliability, ChatGPT in usefulness, and DeepSeek and ...
Medical chatbots resist unnecessary tests, but waver where doctors disagree
Six large language models rarely discouraged medically indicated care but became less consistent when recommending unnecessary tests in areas where ...
Large language model prompting strategies show no stable advantage for predicting SDS screen-positive status in university students
A study of 6,395 Chinese university students found competitive performance from some LLM prompting strategies but no stable advantage over ...
Large Language Models Could Help Detect Pharmacogenomic Phenoconversion
A 2026 perspective proposes that large language models could integrate genetic and clinical records to identify possible pharmacogenomic phenoconversion, while ...
SpecLLM feature-based VNIR representations improve selected LLM classification tasks
A Scientific Reports abstract reports that SpecLLM’s compact VNIR features improved selected LLM blood-substance classification over raw numerical encodings, although ...
Automated Prompt Optimization Improves Same-Day Discharge Prediction in a Retrospective Inpatient Cohort Study
A retrospective cohort study found that automated prompt optimization improved a large language model’s same-day discharge prediction performance on a ...
Controlled robot experiment shows trigger attack and defense for LLM planners
A controlled Binghamton University robot experiment showed that a hidden trigger word could alter LLM-generated task plans and led to ...
POPULAR NEWS
-
VR Outperforms Screens for Medical Students Learning Facial Anatomy
-
Defenses Only Partially Protect Thyroid Ultrasound Segmentation From Adversarial Attacks
-
Survey Evaluates Safe Medication Practices Among Pregnant Women in Suzhou, China
-
Scoping Review Maps Cannabis-Related Emergency Visits in Youth