Search Articles

View query in Help articles search

Search Results (1 to 5 of 5 Results)

Download search results: CSV END BibTex RIS


Evaluating Bard Gemini Pro and GPT-4 Vision Against Student Performance in Medical Visual Question Answering: Comparative Case Study

Evaluating Bard Gemini Pro and GPT-4 Vision Against Student Performance in Medical Visual Question Answering: Comparative Case Study

Standardized prompts used for medical visual question-answering tasks in English and German. This table presents the system prompts provided to GPT-4 1106 Vision Preview and Bard Gemini Pro for answering 1070 image-based multiple-choice questions (605 English and 465 German) from the AMBOSS learning platform. The standardized format ensures consistency in the models’ approach to interpreting medical images and providing diagnoses or answers across both languages. Question template with language and prompt.

Jonas Roos, Ron Martin, Robert Kaczmarczyk

JMIR Form Res 2024;8:e57592

Authors’ Reply: “Evaluating GPT-4’s Cognitive Functions Through the Bloom Taxonomy: Insights and Clarifications”

Authors’ Reply: “Evaluating GPT-4’s Cognitive Functions Through the Bloom Taxonomy: Insights and Clarifications”

Traditionally in educational testing, “difficulty” is quantified by the ratio of correct responses against the number of students taking the test [3]; thus, a rating of 1 indicates an extremely simple question (100% correct responses), and a rating of 0 indicates a significantly challenging question (0% correct responses). Throughout the manuscript, we used “difficulty” as a measurement scale. Consequently, “higher difficulty” means it is higher on the scale and thus easier. This also applies to Figure 3.

Anne Herrmann-Werner, Teresa Festl-Wietek, Friederike Holderried, Lea Herschbach, Jan Griewatz, Ken Masters, Stephan Zipfel, Moritz Mahling

J Med Internet Res 2024;26:e57778

Assessing ChatGPT’s Mastery of Bloom’s Taxonomy Using Psychosomatic Medicine Exam Questions: Mixed-Methods Study

Assessing ChatGPT’s Mastery of Bloom’s Taxonomy Using Psychosomatic Medicine Exam Questions: Mixed-Methods Study

In addition to question stems, answers, and distractors, the system also offers quality criteria for each individual question. Illustration of our methodological approach. API: application programming interface; GPT-4: Generative Pre-trained Transformer 4; MCQ: multiple-choice question. Each examination consists of 20 MCQs, with each MCQ having 1 answer and 4 distractors. Diagnostic and therapeutic questions cover topics concerning anxiety disorders, depression, eating, and somatoform and trauma disorders.

Anne Herrmann-Werner, Teresa Festl-Wietek, Friederike Holderried, Lea Herschbach, Jan Griewatz, Ken Masters, Stephan Zipfel, Moritz Mahling

J Med Internet Res 2024;26:e52113

Examining the Public’s Most Frequently Asked Questions Regarding COVID-19 Vaccines Using Search Engine Analytics in the United States: Observational Study

Examining the Public’s Most Frequently Asked Questions Regarding COVID-19 Vaccines Using Search Engine Analytics in the United States: Observational Study

Google assigns each FAQ a link to information that “answers” the question [20]. Google uses its webmaster guidelines to remove low-quality spam websites from search results and prioritize high-quality sources using a system called Page Rank [19]. Taken together, these FAQs represent millions of common inquiries regarding medical information. Linked answers to each FAQ reveal which information sources individuals are likely to encounter when searching Google for medical information.

Nicholas B Sajjadi, Samuel Shepard, Ryan Ottwell, Kelly Murray, Justin Chronister, Micah Hartwell, Matt Vassar

JMIR Infodemiology 2021;1(1):e28740

Analysis of Keywords Used in Internet Searches for Melanoma Information: Observational Study

Analysis of Keywords Used in Internet Searches for Melanoma Information: Observational Study

Using advertising competition data (paid competition and CPC) from multiple databases for the most popular keywords and questions in internet searches for melanoma and skin cancer, the keyword or question associated with the highest or one of the highest indicators of advertising competition (in the case of a tie) was related to treatment in 7 out of 10 cases. The keyword or question associated with the lowest indicator of advertising competition was related to diagnosis in 8 out of 10 cases.

Japbani K Nanda, Jennifer L Hay, Michael A Marchetti

JMIR Dermatol 2021;4(1):e25720