ChatGPT in medical imaging higher education

Currie, Geoffrey; Singh, Chandralekha; Nelson, Timothy S.; Nabasenja, Caroline; Al-Hayek, Yazan; Spuur, Kelly

doi:10.1016/j.radi.2023.05.011

Cited by 44 publications

(26 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Echoing Currie et al's observations, our study also noted GPT-4's variable accuracy in diagnosing complex cardiac conditions, January 2024. Highlighting the critical role of contextual knowledge and expertise in medical diagnostics (27). Finally, in comparison to highly specialized and research-oriented deep learning tools, our results show lower accuracy.…”

Section: Discussionmentioning

confidence: 75%

Beyond Text: The Impact of Clinical Context on GPT-4’s 12-lead ECG Interpretation Accuracy

Lisicic,

Jordan,

Serman

et al. 2024

Preprint

View full text Add to dashboard Cite

Introduction Artificial intelligence (AI) and large language models (LLMs), such as OpenAI's Chat Generative Pre-trained Transformer – version 4 (GPT-4), are being increasingly explored for medical applications, including clinical decision support. The introduction of the capability to analyze graphical inputs marks a significant advancement in the functionality of GPT-4. Despite the promising potential of AI in enhancing diagnostic accuracy, the effectiveness of GPT-4 in interpreting complex 12-lead electrocardiograms (ECGs) remains to be assessed. Methods This study utilized GPT-4 to interpret 150 12-lead ECGs from the Cardiology Research Dubrava (CaRD) registry, spanning a wide range of cardiac pathologies. The ECGs were classified into four categories for analysis: Arrhythmias (Category 1), Conduction System abnormalities (Category 2), Acute Coronary Syndrome (Category 3), and Other (Category 4). Two experiments were conducted: one where GPT-4 interpreted ECGs without clinical context and another with added clinical scenarios. A panel of experienced cardiologists evaluated the accuracy of GPT-4's interpretations. Statistical significance was determined using the Shapiro-Wilk test for distribution, Mann-Whitney U test for continuous variables, and Chi-square/Fisher's exact tests for categorical variables. Results In this cross-sectional, observational study, GPT-4 demonstrated a correct interpretation rate of 19% without clinical context and a significantly improved rate of 45% with context (p < 0.001). The addition of clinical scenarios significantly enhanced interpretative accuracy, particularly in the Category 3 (Acute Coronary Syndrome) (10 vs. 70%, p < 0.0.01). Unlike Category 4 (Other) which showed no impact (51 vs. 59%, p = 0.640), an impact with a trend toward significance was observed in Category 1 (Arrhythmias) (9.7 vs. 32%, p = 0.059) and Category 2 (Conduction System abnormalities) (4.8 vs. 19%, p = 0.088) when tasked with context. Conclusion While GPT-4 shows some potential in aiding ECG interpretation, its effectiveness varies significantly depending on the presence of clinical context. The study suggests that, in its current form, GPT-4 alone may not suffice for accurate ECG interpretation across a broad spectrum of cardiac conditions.

show abstract

Section: Discussionmentioning

confidence: 75%

Beyond Text: The Impact of Clinical Context on GPT-4’s 12-lead ECG Interpretation Accuracy

Lisicic,

Jordan,

Serman

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Several studies have examined the ability of ChatGPT to be used as a replacement for student examinations or assessments in a variety of elds of accounting (Wood et al, 2023), anatomy (Talan & Kalinkara, 2023), computer science (Bordt & von Luxburg, 2023), cultural heritage (Spennemann, 2023a), dentistry (Ali et al, 2023), economics (Geerling, Mateer, Wooten, & Damodaran, 2023), engineering (Nikolic et al, 2023), English (high school) (de Winter, 2023), information systems (Busch & Hausvik, 2023), law (US) (OpenAI, 2023b), medical licensing boards (Gilson et al, 2023;Kung et al, 2023), medicine (Gravel, D'Amours-Gravel, & Osmanlliu, 2023), nursing (Irwin et al, 2023), operations management (Terwiesch, 2023), ophthalmology ( (Currie et al, 2023). While ChatGPT is able to provide coherent text that at rst sight presents as a well-informed answer, the responses are formulaic, mechanistic and essentially shallow (Bishop, 2023).…”

Section: The Use Of Generative Ai For Student Assessment and Associat...mentioning

confidence: 99%

“…Several studies have found that ChatGPT has been found to be ineffective at reviewing literature (Ali et al, 2023), or to present complex concepts in a coherent and logical fashion (Gilson et al, 2023;Spennemann, 2023a). A common characteristic of text written by generative AI was that the answers and mini essays lacked depth (Busch & Hausvik, 2023;Chaudhry et al, 2023;Currie et al, 2023;Spennemann, 2023a;Stokel-Walker, 2022). For many professions, such as medicine, the fallibility of ChatGPT is of concern as it can provide factually wrong answers that appear authoritative (Currie, 2023;Heng, Teo, & Tan, 2023).…”

Section: The Use Of Generative Ai For Student Assessment and Associat...mentioning

confidence: 99%

ChatGPT giving advice on how to cheat in university assignments—how workable are its suggestions?

Spennemann,

Biles,

Lachlan Brown

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

The generative artificial intelligence (AI) language model ChatGPT is programmed not to provide answers that are unethical or that may cause harm to people. By setting up user-created role-plays designed to alter ChatGPT’s persona, ChatGPT can be prompted to answer with inverted moral valence supplying unethical answers. In this inverted moral valence mode ChatGPT was asked to provide suggestions on how to avoid being detected when commissioning and submitting contract written assignments. We conducted 30 iterations of the task, we examine the types of the suggested strategies and their likelihood of avoiding detection by markers, or, if detected, escaping a successful investigation of academic misconduct. Suggestions made by ChatGPT ranged from communications with contract writers and the general use of contract writing services to content blending and innovative distraction techniques. While the majority of suggested strategies has a low chance of escaping detection, recommendations related to obscuring plagiarism and content blending as well as techniques related to distraction have a higher probability of remaining undetected. We conclude that ChatGPT can be used with success as a brainstorming tool to provide cheating advice, but that its success depends on the vigilance of the assignment markers and the cheating student’s ability to distinguish between genuinely viable options and those that appear to be workable but are not. In some cases the advice given would actually decrease probability of remaining undetected.

show abstract

“…Adding to this reality, the recent emergence of AI-based language models, such as ChatGPT, Google Bard, Microsoft Bing Chat, among others, has triggered unprecedented interrogations to the world's higher education systems (Currie et al, 2023;Strzelecki, 2023). These inquiries are mainly centered around concomitant epistemological dilemmas (Benuyenah, 2023), their intertwining with strategies to foster student engagement (Fuchs, 2023), the development of personalized learning models (Lo, 2023), and the limitation of creative capacities (Gilson et al, 2023;Wang et al, 2023).…”

Section: Introductionmentioning

confidence: 99%

Bibliometric analysis of global scientific literature on the application and implementation of artificial intelligence in higher education

Espinoza-Hurtado,

Bastidas-Santivañez,

Sarmiento-Peralta

2024

Preprint

View full text Add to dashboard Cite

This study aimed to analyze the bibliometric characteristics of global scientific literature on the application and implementation of artificial intelligence in higher education. A bibliometric study was conducted using a search strategy applied in Scopus. Productivity, visibility, and impact indicators were estimated using the SciVal tool. The majority of the world's scientific production on the subject belongs to the field of computer science and engineering. Additionally, high-impact journals play a significant role in dissemination. Articles published in recent years received higher citations. China and India have institutions actively contributing to productivity, although it is essential to strengthen international, inter-institutional, and interprofessional cooperation to promote and enhance the use of artificial intelligence in teaching and learning processes in higher education.

show abstract

ChatGPT in medical imaging higher education

Cited by 44 publications

References 1 publication

Beyond Text: The Impact of Clinical Context on GPT-4’s 12-lead ECG Interpretation Accuracy

Beyond Text: The Impact of Clinical Context on GPT-4’s 12-lead ECG Interpretation Accuracy

ChatGPT giving advice on how to cheat in university assignments—how workable are its suggestions?

Bibliometric analysis of global scientific literature on the application and implementation of artificial intelligence in higher education

Contact Info

Product

Resources

About