Stefano Gianolini

👉🏼 Evaluation of ChatGPT-Generated Educational Patient Pamphlets for Common Interventional Radiology Procedures 🤓 Soheil Kooraki https://lnkd.in/epdGuXb3 🔍 Focus on data insights: - Identified shortcomings in 30% of pamphlets - Total of 34 specific inaccuracies found - Key-word co-occurrence network revealed consistent themes 💡 Main outcomes and implications: - Potential clinical relevance of ChatGPT-generated pamphlets - Fairly consistent terminology observed - Need for future improvements and refinements in large language models 📚 Field significance: - Importance of maintaining human supervision and expert validation - Ensuring patient safety through accurate educational materials 🗄️: #InterventionalRadiology #ChatGPT #PatientEducation #DataInsights

1

1 Kommentar

Important step towards improving transparency, reproducibility, and scalability of the AI analysis tools in biomedical imaging data. AI-generated segmentations are now available along with data in the National Lung Screening Trial (NLST) collection. The results were generated by applying Totalsegmentator in cloud to 126,088 CT series to segment the total of 9,565,554 anatomic structures.

34

1 Kommentar

Both #ChatGPT and #Gemini can accurately triage critical and urgent patients in the Emergency Severity Index (ESI) 1&2 groups at a high rate. Furthermore, ChatGPT is more successful in ESI triage for all patients. These results suggest that large language models can assist in accurately triaging patients in the Emergency Room https://lnkd.in/dDYpvPzj

3

♥️ Validation of the myocardial-ischaemic-injury-index machine learning algorithm for the early diagnosis of myocardial infarction #MI #cardiology #AI #ML #digitalhealth #medicine

12

1 Kommentar

Today the most important article in chatgpt and "Enhancing readability of USFDA patient communications through large language models: a proof-of-concept study" - Large language models (LLMs) significantly improved the readability of USFDA patient communications, reducing grade levels to more accessible readings. - The study showed that LLM outputs retained technical accuracy and key messages while simplifying complex health-related information. - Future research is recommended to extend these findings to other languages and patient groups in real-world settings. #healthcare #patientcommunication #languagemodels #readability #research

1

1 Kommentar

👉🏼 How artificial intelligence can provide information about subdural hematoma: Assessment of readability, reliability, and quality of ChatGPT, BARD, and perplexity responses 🤓 Şanser Gül 👇🏻 https://lnkd.in/edFBc47J 🔍 Focus on data insights: - The study evaluated the quality, reliability, and readability of responses from ChatGPT, Bard, and perplexity regarding subdural hematoma. - Readability scores were compared with the sixth-grade reading level, showing significant differences in all formulas. - Responses from all three chatbots were found to be difficult to read. 💡 Main outcomes and implications: - Bard responses were more readable than ChatGPT's and perplexity's responses. - Perplexity responses had the best Journal of American Medical Association and modified DISCERN scores. - Current capabilities of ChatGPT, Bard, and perplexity are inadequate in terms of quality and readability of content related to subdural hematoma. 📚 Field significance: - The study highlights the challenges in achieving the recommended readability levels for patient education materials in the context of artificial intelligence applications. - Insights from this research can guide improvements in the quality and readability of AI-generated medical information. 🗄️: [#subduralhematoma, #AI, #readability, #medicalinformation]

1

👉🏼 Applications of ChatGPT in Otolaryngology-Head Neck Surgery: A State of the Art Review 🤓 Jérôme R Lechien 👇🏻 https://lnkd.in/gqkiCCNZ 🔍 Focus on data insights: - ChatGPT provides imperfect patient information or general knowledge related to diseases found in Otolaryngology-Head and Neck Surgery. - Studies reported that the model is more accurate in providing diagnoses than in suggesting additional examinations and treatments. - ChatGPT has been used to improve scientific reports, elaborate study protocols, and for student or resident exams. 💡 Main outcomes and implications: - Most applications of ChatGPT are limited to generating disease or treatment information and improving clinical case management. - Lack of comparison with other large language models is a key limitation. - Potential for analyzing clinical images in otolaryngology remains unexplored. 📚 Field significance: - ChatGPT shows promise in enhancing medical information retrieval and clinical decision-making in Otolaryngology-Head and Neck Surgery. 🗄️: #ChatGPT #Otolaryngology #DataInsights #ClinicalDecisionMaking

2

Updated the module for quickly delineating arbitrary three-dimensional tumors. The user only needs to inform the model of the approximate location of the tumor in the Axial View of the CT or MRI images, and the model will automatically outline the contours in three-dimensional space. Moreover, this three-dimensional generation is essentially instantaneous. This feature has already been updated on service.neuralrad.com https://lnkd.in/dcfUQN2j

12

Conclusion In classical mathematics, Ricci curvature and Ricci flow are among the most important tools for analyzing and decomposing manifolds according to their geometric and topological properties. What is interesting is that the corresponding discrete counterparts are shown to be powerful for detecting community structures. Interesting future work includes improving the theoretical understanding of discrete curvature on graphs and applying our methods for real-world applications. https://lnkd.in/g5x4p3R4

3

1 Kommentar

👉🏼 ChatGPT performance on radiation technologist and therapist entry to practice exams 🤓 Ryan Duggan 👇🏻 https://lnkd.in/eBH-hh6m 🔍 Focus on data insights: - ChatGPT passed the radiological technology exam only once (20%) - MRI exam passed all five times (100%) - Nuclear medicine exam passed three times (60%) - Radiation therapy exam passed all five times (100%) - Best performance on knowledge questions across disciplines except radiation therapy - Worst performance on critical thinking questions - Substantial to almost perfect agreement in responses across multiple attempts 💡 Main outcomes and implications: - ChatGPT's varying performance across different disciplines - Algorithm's substantial to almost perfect agreement in responses - Importance of considering repeated measures in evaluating ChatGPT's performance on standardized tests 📚 Field significance: - Advancements in AI for medical certification exams - Potential for improving exam preparation and performance - Implications for future research on AI in standardized testing 🗄️: #radiationtechnologists #therapist #AIperformance #certificationexams

1

🌟European Radiology Experimental – 2023’s Most Cited Papers! Spotlight the three most cited papers from European Radiology Experimental in 2023: 1. “Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem” by Johannes Hofmanninger, Forian Prayer, Jeanny Pan, Sebastian Röhrich, Helmut Prosch & Georg Langs. This landmark paper, which has garnered in 2023 71 citations, challenges existing paradigms by emphasizing the significance of data diversity over methodology in enhancing automatic lung segmentation. Full-text link: https://rdcu.be/dFbZu  2. “MRI with ultrahigh field strength and high-performance gradients: challenges and opportunities for clinical neuroimaging at 7 T and beyond” by Behroze Vachha & Susie Huang. With 12 citations in 2023, this paper explores the frontiers of ultrahigh-field MRI, discussing both the technical challenges and the immense potential for clinical applications, especially in neuroimaging. Full-text link: https://rdcu.be/dFbY9  3. “On the usage of average Hausdorff distance for segmentation performance assessment: hidden error when used for ranking” by Orhun Utku Aydin, Abdel Aziz Taha, Adam Hilbert, Ahmed A. Khalil, Ivana Galinovic, Prof. Dr. Jochen Fiebach, Dietmar Frey, MD JD MBA & Vince Istvan Madai, also cited 12 times in 2023. This insightful research critiques the reliability of using average Hausdorff distance as a standalone metric for ranking segmentation performance, highlighting potential inaccuracies. Full-text link: https://rdcu.be/dFbYY Kudos to all the authors for their remarkable contributions! #EuropeanRadiologyExperimental

59

Excellent study you may not have seen published in NEJM AI, a randomized control trial conducted in Taiwan to compare the treatment times for ST-Elevated MI (STEMI) patients triaged both with and without AI-enabled ECG. Time from arrival to treatment of a catheter treatment for heart attack is critical because every minute the STEMI is untreated costs the patient heart muscle due to localized hypoxia and lowers the effectiveness of their heart (ejection fraction). 💔 STEMI heart attacks are rare, making up just 0.5% of the 44k patients considered in the study. This limited sample of just 209 STEMI patients is further limited in that just 145 had a STEMI with an occluded vessel. 💝 Door-to-Balloon times were 14 minutes faster for ED patients using AI-ECG vs. without, and 5.6 minutes faster for all ED and inpatients. 🖥 Also compared were a Deep Learning Model and the Philips Automatic System, with Sensitivity/Specificity of 89.5%/99.9% and 61.7%/98.0%, respectively. #ai #medicalai #healthcare #responsibleai #heartattacks #heartdisease #medicalemergency #medicalresearch #medicaldiagnosis American Heart Association American College of Cardiology https://lnkd.in/e7dn3nNK

11

1 Kommentar

There is a lot of talk and papers about #AI, but not many practical implementations. Tomorrow at 11:45 (Purple Room), I will present unpublished preliminary data on the AI-augmented chest pain pathway at the CSI congress in Frankfurt! A real application of deep neural network (the AI of our century) #ECG assisting clinical decision-making at the first point of cardiovascular patient contact in Heart Center Aalst.

86

4 Kommentare