JMIR AI, Journal Year: 2024, Volume and Issue: unknown
Published: Dec. 17, 2024
Language: Английский
International Journal for Equity in Health, Journal Year: 2025, Volume and Issue: 24(1)
Published: Feb. 26, 2025
Abstract Background Large language models are increasingly evaluated for use in healthcare. However, concerns about their impact on disparities persist. This study reviews current research demographic biases large to identify prevalent bias types, assess measurement methods, and evaluate mitigation strategies. Methods We conducted a systematic review, searching publications from January 2018 July 2024 across five databases. included peer-reviewed studies evaluating models, focusing gender, race, ethnicity, age, other factors. Study quality was assessed using the Joanna Briggs Institute Critical Appraisal Tools. Results Our review 24 studies. Of these, 22 (91.7%) identified biases. Gender most prevalent, reported 15 of 16 (93.7%). Racial or ethnic were observed 10 11 (90.9%). Only two found minimal no certain contexts. Mitigation strategies mainly prompt engineering, with varying effectiveness. these findings tempered by potential publication bias, as negative results less frequently published. Conclusion Biases various medical domains. While detection is improving, effective still developing. As LLMs influence critical decisions, addressing resultant essential ensuring fair artificial intelligence systems. Future should focus wider range factors, intersectional analyses, non-Western cultural Graphic
Language: Английский
Citations
1
Published: Sept. 9, 2024
Abstract Background Large language models (LLMs) are increasingly evaluated for use in healthcare. However, concerns about their impact on disparities persist. This study reviews current research demographic biases LLMs to identify prevalent bias types, assess measurement methods, and evaluate mitigation strategies. Methods We conducted a systematic review, searching publications from January 2018 July 2024 across five databases. included peer-reviewed studies evaluating LLMs, focusing gender, race, ethnicity, age, other factors. Study quality was assessed using the Joanna Briggs Institute Critical Appraisal Tools. Results Our review 24 studies. Of these, 22 (91.7%) identified LLMs. Gender most prevalent, reported 15 of 16 (93.7%). Racial or ethnic were observed 10 11 (90.9%). Only two found minimal no certain contexts. Mitigation strategies mainly prompt engineering, with varying effectiveness. these findings tempered by potential publication bias, as negative results less frequently published. Conclusion Biases various medical domains. While detection is improving, effective still developing. As influence critical decisions, addressing resultant essential ensuring fair AI systems. Future should focus wider range factors, intersectional analyses, non-Western cultural
Language: Английский
Citations
5
Frontiers in Public Health, Journal Year: 2025, Volume and Issue: 13
Published: Jan. 27, 2025
Background Health equity and access to essential medical information remain significant challenges, especially for the Spanish-speaking Hispanic population, which faces barriers in accessing living kidney donation opportunities. ChatGPT, an AI language model with sophisticated natural processing capabilities, has been identified as a promising tool translating critical health into Spanish. This study aims assess ChatGPT’s translation efficacy ensure provided is accurate culturally relevant. Methods T his utilized ChatGPT versions 3.5 4.0 translate 27 frequently asked questions (FAQs) from English Spanish, sourced Donate Life America’s website. The translated content was reviewed by native nephrologists using standard rubric scale (1–5). assessment focused on linguistic accuracy cultural sensitivity, emphasizing retention of original message, appropriate vocabulary grammar, relevance. Results mean scores were 4.89 ± 0.32 GPT-3.5 5.00 0.00 GPT-4.0 ( p = 0.08). percentage excellent-quality translations (score 5) 89% 100% 0.24). sensitivity both 1.00). Similarly, achieved cases Conclusion demonstrates strong potential enhance improving patients’ LKD through sensitive translations. These findings highlight role mitigating healthcare disparities underscore need integrating AI-driven tools systems. Future efforts should focus developing accessible platforms establishing guidelines maximize AI’s impact equitable delivery patient education.
Language: Английский
Citations
0
PLOS Digital Health, Journal Year: 2025, Volume and Issue: 4(2), P. e0000746 - e0000746
Published: Feb. 21, 2025
Language: Английский
Citations
0
Published: Nov. 19, 2024
Aim: This study aims to evaluate the accuracy and readability of responses generated by two large language models (LLMs) (ChatGPT-4 Gemini) frequently asked questions lay persons (the general public) about signs symptoms, risk factors, screening, diagnosis, treatment, prevention, survival in relation oral cancer. Methods: The each response given LLMs was rated four cancer experts, blinded source responses. as 1: complete, 2: correct but insufficient, 3: includes incorrect/outdated information, 4: completely incorrect. Frequency, mean scores for question, overall were calculated. Readability analyzed using Flesch Reading Ease Flesch-Kincaid Grade Level (FKGL) tests. Results: ChatGPT-4 ranged from 1.00 2.00, with an score 1.50 (SD 0.36), indicating that usually sometimes insufficient. Gemini had ranging 1.75, 1.20 0.27), suggesting more complete Mann-Whitney U test revealed a statistically significant difference between models’ (p = 0.02), outperforming terms completeness accuracy. ChatGPT generally produces content at lower grade level (average FKGL: 10.3) compared 12.3) 0.004). Conclusions: provides accurate people may seek answers ChatGPT-4, although its less readable. Further improvements model training evaluation consistency are needed enhance reliability utility healthcare settings.
Language: Английский
Citations
1
Communications in computer and information science, Journal Year: 2024, Volume and Issue: unknown, P. 390 - 404
Published: Nov. 26, 2024
Language: Английский
Citations
1
Journal of Medicine Surgery and Public Health, Journal Year: 2024, Volume and Issue: 3, P. 100131 - 100131
Published: Aug. 1, 2024
Chronic kidney disease (CKD) is defined as the presence of damage persisting for 3 months or more. Kidney transplantation stands a vital intervention individuals grappling with end-stage renal (ESRD) in Africa, offering promise extended life and improved quality life. However, numerous challenges hinder its widespread implementation across continent. This paper explored aiming to illuminate key strategies bridging gaps building pathways enhanced care. There disproportionate burden CKD on region's population. Therefore, there critical need early diagnosis intervention. outlines comprehensive improving Africa. Results indicate that financial support systems, infrastructure enhancement, public awareness campaigns, legal frameworks are essential addressing care barriers. Active measures such government subsidy programs, international funding collaboration, engagement community leaders highlighted effective approaches. Drawing from global standards best practices, shows importance tailored approaches address Africa's unique socio-economic healthcare landscape. By leveraging collaborative efforts, regulatory frameworks, engagement, African nations can overcome barriers pave way equitable access life-saving treatment.
Language: Английский
Citations
0
Proceedings of the Association for Information Science and Technology, Journal Year: 2024, Volume and Issue: 61(1), P. 1083 - 1085
Published: Oct. 1, 2024
ABSTRACT In this paper we examine the determinants (trust and social influence) that influence users' future use intentions of generative artificial intelligence (AI) applications among black, indigenous other people color (BIPOC) users AI such as ChatGPT large language models. An online survey instrument was administered to 119 BIPOC identified individuals residing in United States. Descriptive inferential statistics were used analyze data through SPSS. Results indicate no statistically significant differences ethnic/ racial groups continued intentions. However, with respect trust, Hispanic/LatinX perceive information from chatbots more trustworthy accurate, their search activity secure than populations. The study also found trust factors driving
Language: Английский
Citations
0
medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown
Published: Nov. 1, 2024
Abstract Background Large Language Models (LLMs) are emerging as promising tools in healthcare. This systematic review examines LLMs’ potential applications nephrology, highlighting their benefits and limitations. Methods We conducted a literature search PubMed Web of Science, selecting studies based on Preferred Reporting Items for Systematic Reviews Meta-Analyses (PRISMA) guidelines. The focuses the latest advancements LLMs nephrology from 2020 to 2024. PROSPERO registration number: CRD42024550169. Results Fourteen met inclusion criteria were categorized into five key areas nephrology: Streamlining workflow, disease prediction prognosis, laboratory data interpretation management, renal dietary patient education. showed high performance various clinical tasks, including managing continuous replacement therapy (CRRT) alarms (GPT-4 accuracy 90-94%) reducing intensive care unit (ICU) alarm fatigue, predicting chronic kidney diseases (CKD) progression (improved positive predictive value 6.7% 20.9%). In education, GPT-4 excelled at simplifying medical information by readability complexity, accurately translating transplant resources. Gemini provided most accurate responses frequently asked questions (FAQs) about CKD. Conclusions While incorporation shows promise across levels care, broad implementation is still premature. Further research required validate these terms accuracy, rare critical conditions, real-world performance.
Language: Английский
Citations
0
Journal of Nephrology, Journal Year: 2024, Volume and Issue: unknown
Published: Dec. 2, 2024
Language: Английский
Citations
0