A Systematic Review of ChatGPT and Other Conversational Large Language Models in Healthcare

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

Published: April 27, 2024

Abstract Background The launch of the Chat Generative Pre-trained Transformer (ChatGPT) in November 2022 has attracted public attention and academic interest to large language models (LLMs), facilitating emergence many other innovative LLMs. These LLMs have been applied various fields, including healthcare. Numerous studies since conducted regarding how employ state-of-the-art health-related scenarios assist patients, doctors, health administrators. Objective This review aims summarize applications concerns applying conversational healthcare provide an agenda for future research on Methods We utilized PubMed, ACM, IEEE digital libraries as primary sources this review. followed guidance Preferred Reporting Items Systematic Reviews Meta-Analyses (PRIMSA) screen select peer-reviewed articles that (1) were related both (2) published before September 1 st , 2023, date when we started paper collection screening. investigated these papers classified them according their concerns. Results Our search initially identified 820 targeted keywords, out which 65 met our criteria included most popular LLM was ChatGPT from OpenAI (60), by Bard Google (1), Large Language Model Meta AI (LLaMA) (5). into four categories terms applications: 1) summarization, 2) medical knowledge inquiry, 3) prediction, 4) administration, concerns: reliability, bias, privacy, acceptability. There are 49 (75%) using summarization and/or 58 (89%) expressing about reliability bias. found exhibit promising results providing patients with a relatively high accuracy. However, like not able reliable answers complex tasks require specialized domain expertise. Additionally, no experiments reviewed thoughtfully examine lead bias or privacy issues research. Conclusions Future should focus improving tasks, well investigating mechanisms brought issues. Considering vast accessibility LLMs, legal, social, technical efforts all needed address promote, improve, regularize application

Language: Английский

---

Applications and Concerns of ChatGPT and Other Conversational Large Language Models in Healthcare: A Systematic Review (Preprint)

Journal of Medical Internet Research, Journal Year: 2024, Volume and Issue: 26, P. e22769 - e22769

Published: Oct. 4, 2024

The launch of ChatGPT (OpenAI) in November 2022 attracted public attention and academic interest to large language models (LLMs), facilitating the emergence many other innovative LLMs. These LLMs have been applied various fields, including health care. Numerous studies since conducted regarding how use state-of-the-art health-related scenarios.

Language: Английский

---

Large language models in healthcare: from a systematic review on medical examinations to a comparative analysis on fundamentals of robotic surgery online test

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(9)

Published: Aug. 6, 2024

Abstract Large language models (LLMs) have the intrinsic potential to acquire medical knowledge. Several studies assessing LLMs on examinations been published. However, there is no reported evidence tests related robot-assisted surgery. The aims of this study were perform first systematic review and establish whether ChatGPT, GPT-4, Bard can pass Fundamentals Robotic Surgery (FRS) didactic test. A literature search was performed PubMed, Web Science, Scopus, arXiv following Preferred Reporting Items for Systematic Reviews Meta-Analyses (PRISMA) approach. total 45 analyzed. GPT-4 passed several national qualifying with questions in English, Chinese, Japanese using zero-shot few-shot learning. Med-PaLM 2 obtained similar scores United States Medical Licensing Examination more refined prompt engineering techniques. Five different 2023 releases one tested FRS. Seven attempts each release. score 79.5%. ChatGPT achieved a mean 64.6%, 65.6%, 75.0%, 78.9%, 72.7% respectively from fifth release FRS vs 91.5% 79.5% Bard. outperformed all corresponding statistically significant difference (p < 0.001), but not = 0.002). Our findings agree other included review. We highlighted challenges transform education healthcare professionals stages learning, by assisting teachers preparation teaching contents, trainees acquisition knowledge, up becoming an assessment framework leaners.

Language: Английский

---

Comparing Meta-Analyses with ChatGPT in the Evaluation of the Effectiveness and Tolerance of Systemic Therapies in Moderate-to-Severe Plaque Psoriasis

Journal of Clinical Medicine, Journal Year: 2023, Volume and Issue: 12(16), P. 5410 - 5410

Published: Aug. 20, 2023

Background: Meta-analyses (MAs) and network meta-analyses (NMAs) are high-quality studies for assessing drug efficacy, but they time-consuming may be affected by biases. The capacity of artificial intelligence to aggregate huge amounts information is emerging as particularly interesting processing the volume needed generate MAs. In this study, we analyzed whether chatbot ChatGPT able summarize in a useful fashion providers patients way that matches up with results MAs/NMAs. Methods: We included 16 (13 NMAs 3 MAs) evaluate biologics (n = 6) both biologic systemic treatment 10) moderate-to-severe psoriasis, published between January 2021 May 2023. Results: conclusions MAs/NMAs were compared ChatGPT’s answers queries about molecules evaluated selected reproducibility was random regarding safety. Regarding reached same conclusion 5 out (four four when three compared), gave acceptable 7 studies, inconclusive 4 studies. Conclusions: can similar MAs efficacy fewer drugs still unable more than compared.

Language: Английский

---

Applications of machine learning methods for design and characterization of high-performance fiber-reinforced cementitious composite (HPFRCC): a review

Journal of Sustainable Cement-Based Materials, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 24

Published: Feb. 6, 2025

Language: Английский

---

Aplicaciones de inteligencia artificial en la escritura y la corrección académicas en la universidad: una revisión sistemática

Íkala Revista de Lenguaje y Cultura, Journal Year: 2025, Volume and Issue: 30(1)

Published: Jan. 1, 2025

La corrección y la edición son esenciales para mejorar calidad de los textos. Aunque existe abundante bibliografía sobre herramientas tecnológicas identificar errores semánticos léxico-gramaticales, las pruebas eficacia real inteligencia artificial (ia) en este proceso siguen siendo limitadas, estudios varían alcance rigor. Este estudio examina si existentes apoyan o contradicen hipótesis que aplicaciones basadas ia ayudan a editar corregir textos enseñanza superior. Se realizó una revisión bases datos Scopus Web of Science, abarcó artículos científicos inglés español, publicados entre 2019 2024, relacionados con escritura universitaria el uso mayoría fueron exploratorios descriptivos. observó un notable aumento publicaciones relacionadas académica 2022 Estados Unidos, China, Australia Canadá cabeza ámbito. Los hallazgos sugieren mejora lingüística retroalimentación escritura. También se destacan problemas integridad académica, privacidad incapacidad resolver complejos. Son necesarias conexiones más explícitas complementar estrategias pedagógicas tradicionales. necesidad investigación ámbito es urgente, ya cuestiones acceso equitativo integración responsable apoyar al desarrollo académica.

---

SEETrials: Leveraging Large Language Models for Safety and Efficacy Extraction in Oncology Clinical Trials

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

Published: Jan. 20, 2024

ABSTRACT Background Initial insights into oncology clinical trial outcomes are often gleaned manually from conference abstracts. We aimed to develop an automated system extract safety and efficacy information study abstracts with high precision fine granularity, transforming them computable data for timely decision-making. Methods collected key conferences PubMed (2012-2023). The SEETrials was developed four modules: preprocessing, prompt modeling, knowledge ingestion postprocessing. evaluated the system’s performance qualitatively quantitatively assessed its generalizability across different cancer types— multiple myeloma (MM), breast, lung, lymphoma, leukemia. Furthermore, of innovative therapies, including CAR-T, bispecific antibodies, antibody-drug conjugates (ADC), in MM were analyzed a large scale studies. Results achieved (0.958), recall (sensitivity) (0.944), F1 score (0.951) 70 elements present studies Generalizability tests on additional cancers yielded precision, recall, scores within 0.966-0.986 range. Variation distribution efficacy-related entities observed diverse certain adverse events more common specific treatments. Comparative analysis using overall response rate (ORR) complete (CR) highlighted differences among therapies: CAR-T (ORR: 88%, 95% CI: 84-92%; CR: 95%, 53-66%), antibodies 64%, 55-73%; 27%, 16-37%), ADC 51%, 37-65%; 26%, 1-51%). Notable heterogeneity identified (>75% I 2 index scores) several outcome therapy subgroups. Conclusion demonstrated highly accurate extraction versatility therapeutics various domains. Its processing datasets facilitates nuanced comparisons, promoting swift effective dissemination insights.

Language: Английский

---

Seetrials: Leveraging Large Language Models for Safety and Efficacy Extraction in Oncology Clinical Trials

Published: Jan. 1, 2024

Language: Английский

---

Topical Cellular/Tissue and Molecular Aspects Regarding Nonpharmacological Interventions in Alzheimer’s Disease—A Systematic Review

International Journal of Molecular Sciences, Journal Year: 2023, Volume and Issue: 24(22), P. 16533 - 16533

Published: Nov. 20, 2023

One of the most complex and challenging developments at beginning third millennium is alarming increase in demographic aging, mainly—but not exclusively—affecting developed countries. This reality results one harsh medical, social, economic consequences: continuously increasing number people with dementia, including Alzheimer’s disease (AD), which accounts for up to 80% all such types pathology. Its large progressive disabling potential, eventually leads death, therefore represents an important public health matter, especially because there no known cure this disease. Consequently, periodic reappraisals different therapeutic possibilities are necessary. For purpose, we conducted systematic literature review investigating nonpharmacological interventions AD, their currently cellular molecular action bases. endeavor was based on PRISMA method, by selected 116 eligible articles published during last year. Because unfortunate lack effective treatments it necessary enhance efforts toward identifying improving various rehabilitative approaches, as well related prophylactic measures.

Language: Английский

---

Examining the Effect of Problem-Based Learning Approach on Learners` Mathematical Creativity: A Meta-Analysis

International Journal of Research in Education and Science, Journal Year: 2024, Volume and Issue: 10(3), P. 653 - 668

Published: Aug. 12, 2024

Problem-based learning (PBL) is linked to developing learners` creative thinking in mathematics. This process-oriented approach capitalizes on using problems stimulate through independent and collaborative investigations. meta-analysis looked at the effectiveness of PBL influence mathematical creativity. Fifteen results from 13 studies were analyzed which a medium effect (g=0.580) was computed random effects model. Further, analysis heterogeneity statistics suggests conducting subgroup only strategy used comparison group educational level, among identified characteristics, moderates Future research should expand geographically encompass more diverse landscape include broader demographic validate efficacy across different age groups cultural contexts.

Language: Английский

---