6th Health Data Analytics Summit discuss the latest AI and machine learning trends in healthcare

The 6th Health Data Analytics Summit brought in leaders from academia, the medical community, and the tech industry to discuss the latest AI and machine learning trends in healthcare. The event was organized by the Health Care Engineering Systems Center (HCESC) and sponsored by Jump ARCHES, OSF HealthCare, University of Illinois Urbana-Champaign, and University of Illinois College of Medicine Peoria collaboration.

This year’s focus was on investigating how generative models (GM) can be leveraged to synthesize increasingly realistic and useful healthcare data, including images such as MRI scans, and text such as electronic health records (EHR).

 The summit kicked off with a keynote talk by renowned physicist-turned-computer-scientist Alan Yuille, from Johns Hopkins University. Prof. Yuille discussed how generative AI can be used to enable the identification of tumors in liver and pancreas CT scans with little to no annotated images. The next keynote was given by Rajesh Ranganath, a rising star from the Courant Institute at NYU. Prof. Ranganath discussed challenges associated with training an AI model on one dataset and deploying it on another dataset with distributional differences, and presented a framework to prevent models from overfitting to "nuisance" parameters such as the type of instrument used for a CT scan. A third keynote was given by UIUC's own Jimeng Sun. Prof. Sun gave a comprehensive historical overview of the developments in AI for healthcare and presented several exciting tools developed by his group, including a generative AI approach for generating synthetic clinical records, which can be used for training other AI diagnosis tools.

In addition to the keynotes, summit presenters discussed various deep generative model frameworks, a wide range of their applications, and their promise of accelerating healthcare-related predictive model development.

“In the practical session we discussed GM implementation strategies, roadmaps and where the transformations in healthcare delivery should start in order to control safety and efficacy,” said Summit organizer George Heintz, Assistant Director Health Data Analytics in HCESC.

Presenters came from seven universities, one agency (FDA), five industry partners, and ten on-campus stakeholders.

“Lately, we have all been hearing about how generative AI is changing the world,” said Summit co-chair Ilan Shomorony, an assistant professor of electrical and computer engineering. “From AlphaGo to ChatGPT, we are witnessing the development of AI algorithms that can exceed human capabilities in an increasing number of tasks. Generative AI can now be used to generate different types of very realistic data, including text and images, and the healthcare industry stands to benefit a great deal from this.”

Shomorony continued to say generative AI can help in disease diagnosis and drug development. It can also generate personalized treatment plans, predict patient outcomes, and improve medical imaging analysis. However, there are still some concerns about responsibly using AI in the field, especially when it comes to data privacy.

“As such, this summit is extremely timely and important, bringing together different perspectives and areas of expertise to discuss how we can harness the great potential of generative AI while also mitigating potential risks and ensuring ethical use in healthcare,” he said.

“This is an exciting moment as we are witnessing a transformative paradigm shift from discriminative AI to generative AI, and the integration of both approaches,” said Summit co-chair Yuxiong Wang, an assistant professor of computer science. “We believe that our summit has made a tangible contribution to this ongoing effort. With three enlightening keynote speeches, twelve insightful plenary talks, a fruitful and stimulating panel discussion, and an engaging live demonstration, the summit covered a broad range of topics from models and evaluation, to systems and clinical workflows, and to medical education, business, policy, and potential risks.” Wang continued that “the discussions and ideas that were shared in the summit will lead to the establishment of new collaborations and initiatives, paving the way for a promising future in generative AI for healthcare and beyond.”