e.g. mhealth
Search Results (1 to 10 of 29 Results)
Download search results: CSV END BibTex RIS
Skip search results from other journals and go to results- 7 JMIR Medical Education
- 5 Journal of Medical Internet Research
- 4 JMIR Formative Research
- 4 JMIRx Med
- 3 JMIR Public Health and Surveillance
- 2 JMIR Research Protocols
- 1 JMIR Dermatology
- 1 JMIR Infodemiology
- 1 JMIR Medical Informatics
- 1 JMIR Serious Games
- 0 Medicine 2.0
- 0 Interactive Journal of Medical Research
- 0 iProceedings
- 0 JMIR Human Factors
- 0 JMIR mHealth and uHealth
- 0 JMIR Mental Health
- 0 JMIR Rehabilitation and Assistive Technologies
- 0 JMIR Preprints
- 0 JMIR Bioinformatics and Biotechnology
- 0 JMIR Cancer
- 0 JMIR Challenges
- 0 JMIR Diabetes
- 0 JMIR Biomedical Engineering
- 0 JMIR Data
- 0 JMIR Cardio
- 0 Journal of Participatory Medicine
- 0 JMIR Pediatrics and Parenting
- 0 JMIR Aging
- 0 JMIR Perioperative Medicine
- 0 JMIR Nursing
- 0 JMIRx Bio
- 0 Transfer Hub (manuscript eXchange)
- 0 JMIR AI
- 0 JMIR Neurotechnology
- 0 Asian/Pacific Island Nursing Journal
- 0 Online Journal of Public Health Informatics
- 0 JMIR XR and Spatial Computing (JMXR)
Go back to the top of the page Skip and go to footer section
However, for decades in the social sciences, this response format has been replaced by instructions requiring parameterization of one's state with numbers. For example, the overwhelming majority of screening measures for gaming disorder are based on closed questions with numerical responses. King et al [1] reviewed 32 measures of gaming disorder based on predefined test items, of which 23 used multiple response scales and 9 used binary responses (yes/no).
JMIR Serious Games 2024;12:e56663
Download Citation: END BibTex RIS
The accuracy of each LLM’s response was then evaluated by comparing it to the benchmark answers provided by a medical professor.
Using this benchmark pipeline, we compared the answers of the generative AI tools, such as GPT-3.5-Turbo-1106 (June 11th version), GPT-4-0613 (November 6th version), GPT-4-1106 (June 11th version), Pa LM 2 (chat-bison), Claude v1, and Gemini Pro, with the benchmark in 15 questions for 39 medical research papers (Table 2).
JMIR Med Inform 2024;12:e59258
Download Citation: END BibTex RIS
If the regenerated response matched the initial answer, the process was halted. However, if the 2 responses differed, the question was posed once more to Chat GPT.
The first and second responses from Chat GPT were directly assessed against the given standard answers for accuracy. For the final response (referred to as joint response), if 2 of the 3 answers were consistent, this was taken as the conclusive answer and evaluated against the standard.
JMIR Med Educ 2024;10:e52784
Download Citation: END BibTex RIS
Assessing GPT-4’s Performance in Delivering Medical Advice: Comparative Analysis With Human Experts
The distribution of speculative or inaccurate information would have had a detrimental effect on the pandemic response strategies. It is paramount to emphasize that inaccuracies or misconceptions in cardiological advice can lead to severe consequences. Hence, there is a pressing need for rigorous validation of all sources of information, whether derived from human experts or advanced computational models such as GPT-4.
JMIR Med Educ 2024;10:e51282
Download Citation: END BibTex RIS
Go back to the top of the page Skip and go to footer section
Go back to the top of the page Skip and go to footer section
Capability of GPT-4V(ision) in the Japanese National Medical Licensing Examination: Evaluation Study
One of the authors, TN, who has 10 years of experience as a medical doctor, reviewed the outputs to interpret the response output by Chat GPT.
A new chat session was created for each question and each condition (ie, with or without images). For questions that comprised multiple subquestions, the background information part and each subquestion were entered into Chat GPT in this order within the same chat session.
JMIR Med Educ 2024;10:e54393
Download Citation: END BibTex RIS
A level of P
A total of 2 authors (TFW and FH) separately coded each text response. The answers from GPT-4 were analyzed inductively and iteratively according to Mayring’s [21] qualitative content analysis, as described previously by our group [22]. The goal of the analysis was defined in line with the answers to the examination questions. For the main category, we used the correct or incorrect answer to the question, then further focused primarily on incorrect answers.
J Med Internet Res 2024;26:e52113
Download Citation: END BibTex RIS
As this was the first study using GPT as a simulated patient, we focused on 1 language model (ie, GPT-3.5, which we chose for its free availability and fast response time) and 1 patient case. Although we perceived our case as representative for history taking, our data did not allow for generalization to more specialized medical fields, and further studies are required to verify scalability to other medical specialties.
JMIR Med Educ 2024;10:e53961
Download Citation: END BibTex RIS