Large Language Models in Cardiology: A Systematic Review DOI Open Access

Moran Gendler,

Girish N. Nadkarni, Karin Sudri

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 1, 2024

Abstract Purpose This review analyzes the application of large language models (LLMs), in field cardiology, with a focus on evaluating their performances across various clinical tasks. Methods We conducted systematic literature search PubMed for studies published up to April 14, 2024. Our used wide range keywords related LLMs and cardiology capture relevant terms. The risk bias was evaluated using QUADAS-2 tool. Results Fifteen met inclusion criteria, categorized into four domains: chronic progressive cardiac conditions, acute events, education, monitoring. Six addressing conditions demonstrated variability accuracy depth LLM-generated responses. In scenarios, three articles showed that provided medical advice mixed effectiveness, particularly delivering CPR instructions. Two educational revealed high answering assessment questions interpreting cases. Finally, diagnostics multimodal displayed capabilities ECGs interpretation, some performing at or exceeding level human specialists. Conclusion demonstrate considerable potential applications routine diagnostics. However, performance remains inconsistent care settings where precision is critical. Enhancing real-world complex data emergency response guidance imperative before integration practice.

Language: Английский

ChatGPT is a comprehensive education tool for patients with patellar tendinopathy, but it currently lacks accuracy and readability DOI Creative Commons
Jie Deng, Lun Li,

Jan Oosterhof

et al.

Musculoskeletal Science and Practice, Journal Year: 2025, Volume and Issue: 76, P. 103275 - 103275

Published: Jan. 31, 2025

Generative artificial intelligence tools, such as ChatGPT, are becoming increasingly integrated into daily life, and patients might turn to this tool seek medical information. To evaluate the performance of ChatGPT-4 in responding patient-centered queries for patellar tendinopathy (PT). Forty-eight were collected from online sources, PT patients, experts then submitted ChatGPT-4. Three board-certified independently assessed accuracy comprehensiveness responses. Readability was measured using Flesch-Kincaid Grade Level (FKGL: higher scores indicate a grade reading level). The Patient Education Materials Assessment Tool (PEMAT) evaluated understandability, actionability (0-100%, information with clearer messages more identifiable actions). Semantic Textual Similarity (STS score, 0-1; similarity) variation meaning texts over two months (including ChatGPT-4o) different terminologies related PT. Sixteen (33%) 48 responses rated accurate, while 36 (75%) comprehensive. Only 17% treatment-related questions received accurate Most written at college level (median interquartile range [IQR] FKGL score: 15.4 [14.4-16.6]). median PEMAT understandability 83% (IQR: 70%-92%), actionability, it 60% 40%-60%). medians STS across all ≥ 0.9. provided generally comprehensive response but lacked difficult read individuals below level.

Language: Английский

Citations

2
Assessing the Readability of Patient Education Materials on Cardiac Catheterization From Artificial Intelligence Chatbots: An Observational Cross-Sectional Study DOI Open Access
Benjamin J. Behers, Ian Vargas, Brett M. Behers

et al.

Cureus, Journal Year: 2024, Volume and Issue: unknown

Published: July 4, 2024

Artificial intelligence (AI) is a burgeoning new field that has increased in popularity over the past couple of years, coinciding with public release large language model (LLM)-driven chatbots. These chatbots, such as ChatGPT, can be engaged directly conversation, allowing users to ask them questions or issue other commands. Since LLMs are trained on amounts text data, they also answer reliably and factually, an ability allowed serve source for medical inquiries. This study seeks assess readability patient education materials cardiac catheterization across four most common chatbots: Microsoft Copilot, Google Gemini, Meta AI.

Language: Английский

Citations

14
Large language models in patient education: a scoping review of applications in medicine DOI Creative Commons
Serhat Aydın, Mert Karabacak,

Victoria Vlachos

et al.

Frontiers in Medicine, Journal Year: 2024, Volume and Issue: 11

Published: Oct. 29, 2024

Large Language Models (LLMs) are sophisticated algorithms that analyze and generate vast amounts of textual data, mimicking human communication. Notable LLMs include GPT-4o by Open AI, Claude 3.5 Sonnet Anthropic, Gemini Google. This scoping review aims to synthesize the current applications potential uses in patient education engagement.

Language: Английский

Citations

9
Adopting artificial intelligence for health information literacy: A literature review DOI
Godwin Dzangare,

Thabo Ayibongwe Gulu

Information Development, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 25, 2025

Purpose – Artificial Intelligence (AI) is increasingly becoming a popular source of information, including health information. It essential to explore the adoption AI achieve Health Information Literacy (HIL) and ensure that users maximise use This study explores AI's in advancing HIL. identifies gaps, concerns, challenges suggests areas where could be improved. Approach The retrieved papers were initially assessed based on title abstract inclusion criteria. full text relevant was verified following exclusion Additionally, comprehensive assessment reference lists included performed. extracted from selected articles, bibliometric thematic analysis applied for thorough examination. Methodology Key details about author, publication year, type, purpose, key findings, collected using standardised format. As themes emerged, information publications address main research questions. All articles reviewed English published between 2019 2024. Findings growing HIL can accounted by growth 128.13% publications. However, concerns must addressed as continuous guaranteed. Originality likely first assess current findings will provide clear landscape investing, identifying partners, providing gap.

Language: Английский

Citations

1
ChatGPT-4 Can Help Hand Surgeons Communicate Better With Patients DOI Creative Commons
Robert Browne,

Khadija Gull,

Ciarán Hurley

et al.

Journal of Hand Surgery Global Online, Journal Year: 2024, Volume and Issue: 6(3), P. 441 - 443

Published: April 6, 2024

The American Society for Surgery of the Hand and British produce patient-focused information above sixth-grade readability recommended by Medical Association. To promote health equity, content should be aimed at an appropriate level literacy. Artificial intelligence-driven large language models may able to assist hand surgery societies in improving provided patients. was calculated all articles written English on websites, terms seven commonest formulas. Chat Generative Pre-Trained Transformer version 4 (ChatGPT-4) then asked rewrite each article a level. response compared with unedited articles. improve across chosen formulas successful achieving mean Flesch Kincaid Grade Level Simple Measure Gobbledygook calculations. It increased Reading Ease score, higher scores representing more readable material. This study demonstrated that ChatGPT-4 can used material surgery. However, is interested primarily sounding natural, not seeking truth, hence, must evaluated surgeon ensure accuracy being sacrificed sake this powerful tool.

Language: Английский

Citations

8
Assessing the response quality and readability of chatbots in cardiovascular health, oncology, and psoriasis: A comparative study DOI Creative Commons
Robert Olszewski, Klaudia Watros, Małgorzata Mańczak

et al.

International Journal of Medical Informatics, Journal Year: 2024, Volume and Issue: 190, P. 105562 - 105562

Published: Oct. 1, 2024

Chatbots using the Large Language Model (LLM) generate human responses to questions from all categories. Due staff shortages in healthcare systems, patients waiting for an appointment increasingly use chatbots get information about their condition. Given number of currently available, assessing they is essential.

Language: Английский

Citations

7
Evaluating ChatGPT’s Accuracy in Responding to Patient Education Questions on Acute Kidney Injury and Continuous Renal Replacement Therapy DOI
M. Salman Sheikh, Charat Thongprayoon, Supawadee Suppadungsuk

et al.

Blood Purification, Journal Year: 2024, Volume and Issue: 53(9), P. 725 - 731

Published: April 26, 2024

Introduction: Acute kidney injury (AKI) and continuous renal replacement therapy (CRRT) are critical areas in nephrology. The effectiveness of ChatGPT simpler, patient education-oriented questions has not been thoroughly assessed. This study evaluates the proficiency 4.0 responding to such questions, subjected various linguistic alterations. Methods: Eighty-nine were sourced from Mayo Clinic Handbook for educating patients on AKI CRRT. These categorized as original, paraphrased with different interrogative adverbs, resulting incomplete sentences, containing misspelled words. Two nephrologists verified medical accuracy. A χ2 test was conducted ascertain notable discrepancies 4.0’s performance across these formats. Results: provided accuracy handling a variety question formats education Across all types, demonstrated an 97% both original adverb-altered 98% sentences or misspellings. Specifically AKI-related consistently maintained at versions. In subset CRRT-related tool achieved 96% this increased statistical analysis revealed no significant difference varied types (p value: 1.00 CRRT), there disparity between artificial intelligence (AI)’s responses CRRT 0.71). Conclusion: demonstrates consistent high interpreting queries related CRRT, irrespective modifications. findings suggest that potential be reliable support delivery education, by accurately providing information range Further research is needed explore direct impact AI-generated understanding outcomes.

Language: Английский

Citations

5
Comparative Analysis of Accuracy, Readability, Sentiment, and Actionability: Artificial Intelligence Chatbots (ChatGPT and Google Gemini) versus Traditional Patient Information Leaflets for Local Anesthesia in Eye Surgery DOI Creative Commons
Prakash Gondode, Sakshi Duggal, Neha Garg

et al.

British and Irish Orthoptic Journal, Journal Year: 2024, Volume and Issue: 20(1), P. 183 - 192

Published: Jan. 1, 2024

Eye surgeries often evoke strong negative emotions in patients, including fear and anxiety. Patient education material plays a crucial role informing empowering individuals. Traditional sources of medical information may not effectively address individual patient concerns or cater to varying levels understanding. This study aims conduct comparative analysis the accuracy, completeness, readability, tone, understandability generated by AI chatbots versus traditional Information Leaflets (PILs), focusing on local anesthesia eye surgery.

Language: Английский

Citations

5
ChatGPT Responses to Frequently Asked Questions on Ménière's Disease: A Comparison to Clinical Practice Guideline Answers DOI Creative Commons
Rebecca Ho, Ariana L. Shaari, Paul T. Cowan

et al.

OTO Open, Journal Year: 2024, Volume and Issue: 8(3)

Published: July 1, 2024

Abstract Objective Evaluate the quality of responses from Chat Generative Pre‐Trained Transformer (ChatGPT) models compared to answers for “Frequently Asked Questions” (FAQs) American Academy Otolaryngology–Head and Neck Surgery (AAO‐HNS) Clinical Practice Guidelines (CPG) Ménière's disease (MD). Study Design Comparative analysis. Setting The AAO‐HNS CPG MD includes FAQs that clinicians can give patients MD‐related questions. ability ChatGPT properly educate regarding is unknown. Methods ChatGPT‐3.5 4.0 were each prompted with 16 questions FAQs. Each response was rated in terms (1) comprehensiveness, (2) extensiveness, (3) presence misleading information, (4) resources. Readability assessed using Flesch‐Kincaid Grade Level (FKGL) Flesch Reading Ease Score (FRES). Results comprehensive 5 whereas ChatGPT‐4.0 9 (31.3% vs 56.3%, P = .2852). extensive all ( 1.0000). 3 18.75%, .6851). had resources 10 (62.5% 100%, .0177). FRES (62.4 ± 16.6) demonstrated an appropriate readability score at least 60, while both (39.1 7.3) (42.8 8.5) failed meet this standard. All platforms FKGL means exceeded recommended level 6 or lower. Conclusion While significantly better resource reporting, have room improvement being more comprehensive, readable, less patients.

Language: Английский

Citations

4
Can Generative AI Improve The Readability Of Patient Education Materials At A Radiology Practice? DOI
Mahesh Gupta, Pooja Gupta,

Cynthia K. Y. Ho

et al.

Clinical Radiology, Journal Year: 2024, Volume and Issue: 79(11), P. e1366 - e1371

Published: Aug. 23, 2024

Language: Английский

Citations

4