Dr. med. ChatGPT? Experts from UAS Technikum Wien contribute to the discussion on artificial intelligence in healthcare
27 August, 2024
Lars Mehnen and Bernhard Knapp are contributors to the book “Health Care and Artificial Intelligence: Understanding Ethical Aspects – Shaping Developments”. The book examines the role of AI in medicine and sheds light on ethical and practical challenges.
Two experts from UAS Technikum Wien are involved in a new textbook on the topic of “Health Care and Artificial Intelligence”. Lars Mehnen and Bernhard Knapp (both lecturers and researchers at UAS Technikum Wien) have made a significant contribution to understanding the ethical and practical aspects of AI in healthcare through their research. The book, edited by Andreas Klein, Sebastian Dennerlein and Helmut Ritschl, was recently published by Narr Francke Attempto Verlag and deals with a variety of current issues relating to the use of AI in medical diagnostics and therapy.
The role of UAS Technikum Wien in current AI research
In their chapter, Mehnen and Knapp focus in particular on the use of GPT models in medical diagnostics and examine the precision of these AI systems in the detection of common and rare diseases. In their study, the researchers worked with two doctors to investigate the diagnostic accuracy of ChatGPT in common and rare diseases. They used 50 clinical case vignettes to evaluate the performance of ChatGPT 3.5 and 4. The results showed that ChatGPT is very accurate for common diseases, but for rare diseases it needs several suggestions to find the correct diagnosis.
We asked Lars Mehnen for an interview about AI in healthcare.
Senior Lecturer/Researcher
How was the diagnostic accuracy of ChatGPT demonstrated in your study?
Lars Mehnen: In our study, ChatGPT showed high accuracy for common diseases, with the correct diagnosis usually being among the two most likely suggestions. For rarer diseases, however, the model required several attempts. These results make it clear that AI models already work very well in certain areas of medicine, but they also have their limits.
In your opinion, what tasks should ChatGPT currently be used for in the medical field?
ChatGPT can certainly serve as a valuable tool in the medical field, especially in the preliminary stage of diagnosis. It can help physicians to consider a wide range of possible diagnoses and thus support them in the decision-making process. However, it should be emphasized that ChatGPT should never make the final diagnosis. The final assessment must always be in the hands of the doctor, who compares the results of the AI with their own expertise and the clinical context.
What developments do you expect to see in the field of AI in medicine over the next few years?
In the coming years, I expect AI to find its way into more and more areas of medicine, particularly in individualized medicine and the analysis of large amounts of data. While AI is already being used in diagnostics, for example in image analysis in radiology, I see great potential in genetic analysis and the development of personalized therapies. Here, AI could help to identify genetic anomalies faster and more precisely, which could lead to customized treatment strategies. However, we also face ethical challenges that need to be clarified, such as the question of how much responsibility can be transferred to AI in medical decision-making processes. Overall, however, I am optimistic that AI-supported systems will be able to provide valuable support and significantly improve the quality of patient care.
Further Links:
Book: Health Care and Artificial Intelligence
ChatGPT: A diagnostic sidekick for doctors? Caution advised for non-professionals
Book: Health Care und Künstliche Intelligenz – Ethische Aspekte verstehen – Entwicklungen gestalten (in German)
Publisher: Andreas Klein, Sebastian Dennerlein and Helmut Ritschl.
Publication date: 17.06.2024, ISBN: 9783825262570