UCI Precision Health through Artificial Intelligence Initiative Seminar
featuring Distinguished Speaker,
Dr. Gustavo Stolovitzky
UCI Precision Health through Artificial Intelligence Initiative presents two special lectures followed by a reception on Thursday, April 9 from 4 p.m. to 6:30 p.m. at the Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
Peter D. Chang, Assistant Professor-in-Residence,
Co-Director for the Center for AI in Diagnostic Medicine
Democratization of Deep Learning Algorithm Development and Pipelines
4 p.m. Presentation and Q&A
Gustavo Stolovitzky, IBM Exploratory Life Sciences Program Chair, IBM Research
Promoting data sharing and collaborative science through crowdsourcing: The DREAM Challenges
5 p.m. Presentation and Q&A
6-6:30 p.m. General Discussion (Reception to follow)
About Peter D. Chang
Dr. Peter D. Chang is an Assistant Professor-in-Residence for the Department of Radiological Sciences and Co-Director for the Center for Artificial Intelligence in Diagnostic Medicine (CADIM) at UC Irvine, a new multi-specialty initiative to develop and integrate artificial intelligence technology across the UCI’s healthcare system. Dr. Chang is also the co-founder of multiple AI startups including most recently Avicenna.ai, a company focused on deep learning for medical imaging.
His unique perspective arises from experience both as a radiologist physician and software engineer with expertise in developing deep learning algorithms, working closely with many industry partners including Nvidia, Amazon and Canon Medical. His work has led to five best conference paper awards, two patents/copyrights, and numerous abstracts, manuscripts, and top finishes in various competitions including the international 2016 MICCAI challenge. Dr. Chang serves as deputy AI editor for the American Journal of Neuroradiology. He is currently a member of national AI steering committees for the RSNA, ASNR, ASFNR, SIIM and ACR.
Abstract: Democratization of Deep Learning Algorithm Development and Pipelines
Despite the growing interest in deep learning technology applied to various domains in the biomedical sciences, there remain significant barriers to entry that preclude widespread adoption of these cutting-edge techniques. To a large extent these bottlenecks ultimately relate to practical challenges in implementation rather than inherent limitations of machine learning technology. In this session, we will explore several of these common barriers including physical hardware expenses, data availability and scarcity of AI experts. In addition, he will highlight several active ongoing initiatives and programs at UCI aimed to overcome many of these limitations, including work at the UCI Center of AI in Diagnostic Medicine (CAIDM) and Precision Health through Artificial Intelligence (PHAI) initiative.
About Gustavo Stolovitzky
IBM Exploratory Life Sciences Program Chair, IBM Research Dr. Gustavo Stolovitzky is the Chair of the Exploratory Life Sciences Program at IBM Research and an adjunct Professor at Columbia University. As the Founder of the DREAM Challenges, Dr. Stolovitzky has pioneered the use of crowdsourcing as a tool for scientific research in systems biology and of the Wisdom of Crowds as a robust methodology for predictive modeling. His current main interests are in the fields of crowdsourcing, ensemble learning, translational research and liquid biopsies. Dr. Stolovitzky, who received his Ph.D. in Mechanical Engineering from Yale University, has been distinguished with several awards including Yale University’s Henry Prentiss Becton Prize award, the HENAAC’s Pioneer Award for Great Minds in STEM, and the “Premio Raíces” awarded by the Argentinian Government. Dr. Stolovitzky has been elected Member of the IBM Academy of Technology, Fellow of the NY Academy of Sciences, Fellow of the World Technology Network, Fellow of the American Physical Society and Fellow of the American Association for the Advancement of Sciences, and IBM Fellow, the highest honor IBM bestows to its technical leaders.
Abstract: Promoting data sharing and collaborative science through crowdsourcing: The DREAM Challenges
Crowdsourcing is a powerful but underutilized strategy in scientific research. When data is shared in the context of scientific challenges, crowdsourcing can contribute to the solution of important biomedical problems by fostering collaboration between research groups, accelerating the pace of discovery and objectively assessing the performance of predictive models. The DREAM Challenges is a community driven effort that leverages the use of crowdsourcing to solve pressing problems in biomedicine. I will discuss a few of the recent DREAM Challenges focusing on examples of challenges where we fostered biomedical clinical data sharing, including ALS progression prediction and patient stratification challenges, the prostate cancer survival prediction challenge and the breast cancer screening challenge. Challenges, however, are only one component of an ecosystem that will in the end maximize the value of communities in AI: the ABCD ecosystem, which integrates Algorithms, Benchmarking, Crowdsourcing and Data.