AI in Healthcare - Opportunities and Challenges

Add to your list(s) Download to your calendar using vCal

• Prof Radek Bukowski - University of Texas at Austin
• Wednesday 02 April 2025, 15:00-16:00
• https://cam-ac-uk.zoom.us/j/87922475493?pwd=jAe4pxUSrczydAzxzahBmpVv5jovXG.1.

If you have a question about this talk, please contact Andreas Bedorf.

AI offers significant opportunities in healthcare while facing substantial challenges. The opportunities and challenges are best appreciated from a time perspective. In the short term, the opportunities are likely in the individualization of testing and interventions. AI tools also show substantial promise in assisting in the interpretation of images, radiologic or histopathologic, potentially increasing the accuracy and rapidity of those tests. Recent advances show AI’s potential for improving healthcare efficiency by automating administrative tasks.

In the long-term, the opportunities are less clear but potentially very exciting. AI has the potential in the process of ever-increasing information synthesis to achieve an individualized medical decision-making system. In such a system, the vast and varied digital information created by each of us flows into an ever more individualized and nuanced model, the digital twin, which, by comparison with other digital twins, can be used to simulate different futures and make optimal medical decisions.

Although very desirable to patients and clinicians, such a system might be far away on the horizon, and in the meantime, AI faces many challenges in healthcare today. The main ones are related to various aspects of privacy concerns, the potential for propagating inequalities in healthcare, inadvertent disruption of the current workflow during the integration of AI tools, and concerns of developing over-reliance on AI tools in medical decision-making resulting in atrophy of clinicians’ own decision-making skills. The long-term challenge emerging with the broad use of AI is the data sources AI relies on. The data used by AI is “content that is accessible to the public without any proprietary restrictions or privacy concerns.” In consequence, the data used by AI is increasingly the data generated by itself and any bias in such a positive feedback mechanism will be exponentially propagated by the AI.

Considering those challenges, initial attempts toward the digital twin simulation-based medical decision-making are promising. Examples are: On the individual patient level, evidence synthesis informing individual decision-making. On the healthcare system level, system dynamic models informing healthcare policy decision-making.

Bio: Radek Bukowski is a doctor, academic physician, scientist and an inventor. He is the director of Computational Health & Medicine Initiatives at the Texas Advanced Computing Center, at University of Texas at Austin.

Prof Bukowski is most known for his works in the fields of computational medicine, preterm birth, maternal fetal, and neonatal mortality and morbidity and fetal growth abnormalities. His works have been published in New England Journal of Medicine and American Journal of Obstetrics and Gynecology. He is also the recipient of 2008 March of Dimes Award for his research in prematurity.

This talk is part of the CCAIM Seminar Series series.

This talk is included in these lists:

• CCAIM Seminar Series
• https://cam-ac-uk.zoom.us/j/87922475493?pwd=jAe4pxUSrczydAzxzahBmpVv5jovXG.1

Note that ex-directory lists are not shown.