Peter Chang
At UCI, “there’s the ability to connect with basic science researchers through clinicians who are world experts in treating some very specific diseases,” says Dr. Peter Chang of the Institute for Precision Health. UCI

In February, UCI launched the Institute for Precision Health, a campus-wide, interdisciplinary endeavor that merges UCI’s powerhouse health sciences, engineering, machine learning, artificial intelligence, clinical genomics and data science capabilities. The objective is to identify, create and deliver the most effective health and wellness strategy for each individual person and, in doing so, confront the linked challenges of health equity and the high cost of care.

IPH will bring a multifaceted, integrated approach to what many call the next great advancement in healthcare. The institute is an ecosystem for collaboration across disciplines. 

Besides leading the applied artificial intelligence research group for IPH, Dr. Peter Chang is co-director of the Center for Artificial Intelligence in Diagnostic Medicine and assistant professor-in-residence in radiological sciences in the School of Medicine. Dr. Chang’s unique perspective arises from experience both as a radiologist physician and full-stack software engineer with over a decade of experience building FDA-cleared tools used in hospitals around the world.

Chang came to academia after launching a successful start-up company and has the distinction of being one of the few medical doctors in the country teaching in a computer science department. At IPH, his job is to use AI and machine learning to design practical solutions to real-world clinical problems for cost-effective, value-based care. Here, Dr. Chang speaks about the promise of AI and what he sees for the Institute for Precision Health’s future. 

For the uninitiated, how would you describe why machine learning and AI are important to healthcare right now?

You have to understand that the newest form of AI – the deep-learning neural network family of algorithms – has completely revolutionized the way machine-learning algorithms learn and think. Traditional forms of AI would require a human to carefully go through a list of patterns, rules and assumptions and manually build in or program that human experience into a computer. Modern forms of AI, however, allow computers to extract patterns and make inferences without a priori human assumptions. For example, if I wanted to teach the algorithm how to play the game of chess, I could simply explain the rules of chess and allow two AIs play against each other.

This paradigm shift is a completely new way to approach the design of learning algorithms. And, interestingly, this strategy has allowed modern AI systems to learn new or interesting information that may be previously unrecognized by even human experts. With video games, oftentimes we may think that the AI is intentionally losing, only to realize at the very end that the computer has come back and beat the human by a small but consistent margin every single time. For healthcare, the implication of course is that an AI may be allowed to discover patterns without the biases of flawed human assumptions or explicit programming – that’s really where the power lies.

And that’s a core component of precision health – healthcare informed by AI and machine learning. Typically, with advancements, though, there can also be downsides. Is there a downside to precision health?

I don’t know if I’d characterize it as the downside, but I will say that there is a lot of hype, which means that the expectations are oftentimes overinflated, and the inability to eventually meet those expectations and perhaps turn people away from the technology is something I’m very aware of. I’m obviously an advocate for this technology, but there are a lot of things we don’t know. It’s really in its infancy in terms of development and especially so in the field of medicine. The room for improvement is tremendous. So, we should acknowledge that. And we should acknowledge that the progress and potential, while it’s absolutely there, may be slow to realize.

UCI launched its Institute for Precision Health in February 2022. What most excites you about the Institute and what do you hope to achieve?

The IPH team is a strong collaborative team with diverse backgrounds. I think that’s a key part to our unique approach to precision health and big data at UCI. At the same time, though the team is large, my role is very specific. In particular, my background is unique in that I both build modern AI algorithms on a daily basis and also practice as a board-certified radiologist. With this perspective, my hope is that I’m able to bridge the gaps between technical and clinical experts to help accelerate translations in AI research for healthcare.

How unusual is it be a medical doctor and professor with the AI background?

Currently in 2022, this combination remains extraordinarily rare. As an illustrative example, I’ve heard that I’m the only physician teaching a technical deep learning class in a computer science department anywhere in the country. The course incorporates a hands-on curriculum building new AI algorithms each week with healthcare imaging data using the same libraries and tools developed by experts at Google, Facebook and Uber.

Before UCI, my real-life experience in AI started with research in the precision health AI field which eventually resulted in a startup company in the radiology deep-learning space. As part of the company, I work actively with our data science and engineering to innovate and translate the latest AI technologies into medical imaging diagnosis. In this capacity my experience with AI and machine learning comes from building state-of-the-art algorithms with industry-standard tools as well as regulatory clearance through the FDA, European CE-Mark and other international agencies. All of this experience is complementary to what you would normally expect of a medical doctor, I guess.

What led you to UCI?

When I was looking for full-time faculty positions, I wanted one that would allow me to continue pursuing hybrid clinical and AI work. Interestingly, that type of position really didn’t exist three years ago. So, in large part what brought me to Irvine was as opportunity the UCI leadership saw for me in expanding AI and machine learning capacity across the healthcare community. More specifically, I was recruited here to build the AI center, an integral component of what is now the new Institute of Precision Health. While our team has grown tremendously, I was one of the first faculty in School of Medicine to dedicate my time and career to AI in healthcare. In this way, I’ve been invested in the Institute for Precision Health from the very beginning.

The process of finding your dream job must’ve been interesting.

If you had asked me four or five years ago where I most likely saw myself, I probably would’ve imagined myself working in industry. But, to put it simply, my single priority throughout has always been having access to resources that allow me to purse impactful work in healthcare AI. If that meant working at Google, I would’ve ended up at Google. But the reality is that here at UCI I was given a unique set of tools and resources that even the tech giants in industry could not match. As example, even Google – with all the Google resources and talent in engineering and other data science – does not have access to a hospital. By contrast, here at UCI I can take a tool built in the lab and turn it on the next day in a realistic clinical environment to see if it actually helps doctors do their day-to-day work. There’s the ability to connect with basic science researchers through clinicians who are world experts in treating some very specific diseases. And on top of all that, here at UCI I continue practicing medicine as a radiologist in the hospital one day a week.

When you’re not practicing medicine or teaching, what does your job look like?

By design, I try to immerse myself in the technical details of new AI technology as much as possible. Working with students in the lab and writing software code is the highlight of day. Most of the time, you’ll find me with my engineering and data science team, both building algorithms and figuring out how best to plug those algorithms into clinical practice. You could say that my team are the “boots on the ground” to bring applications to real-life practice. And, of course, I spend a significant amount of my day with clinicians to discuss potential AI solutions for their daily problems. Almost invariably a conversation will start with, “Peter, I have a problem … if you could solve this problem, it would make my life a hundred times better.”

How many are on your team?

We have about seven or eight full-time staff now at the Center for Artificial Intelligence, in addition to a large number of grant-sponsored students, trainees and postdocs. Additionally, I try to take on as many student volunteers as possible – trainees who are looking just to get exposed to the field. Even without formal funding, there is a large community of individuals who just want to learn. Because of this, I think that the center has become a popular place around UCI.

What do you hope to tackle first now that IPH has launched?

Prior to the launch of IPH, a smaller group at UCI had been focused on precision health and omics. In parallel, my team was focused on AI and machine learning applied to precision health problems. In that regard, combining our expert backgrounds, we have some early projects looking at AI predictive analytics across multiple diagnostic modalities including electronic health record (EHR), radiology and omics data. including DNA, RNA, proteomics. This cross-disciplinary work truly embodies IPH and would quite frankly be impossible unless you had experts like those on our team to help guide you along the way. A few key priority areas of research specifically include ALS, dementia, gastric cancer and COVID.

Any last words about the future of precision health?

AI and precision health are exciting new areas of research, but for now I’d urge everyone to stay grounded and be patient. There are a lot of unknowns and a lot to explore and understand, so a balanced perspective is needed to truly make strides translating these technologies in ways that ultimately will help researchers, clinicians and patients.

If you want to learn more about supporting this or other activities at UCI, please visit the Brilliant Future website at https://brilliantfuture.uci.edu. Publicly launched on October 4, 2019, the Brilliant Future campaign aims to raise awareness and support for UCI. By engaging 75,000 alumni and garnering $2 billion in philanthropic investment, UCI seeks to reach new heights of excellence in student success, health and wellness, research and more. UCI Health Affairs plays a vital role in the success of the campaign. Learn more by visiting https://brilliantfuture.uci.edu/uci-health-affairs/.

About UCI Institute for Precision Health: Founded in February 2022, the Institute for Precision Health (IPH) is a multifaceted, integrated ecosystem for collaboration that maximizes the collective knowledge of patient data sets and the power of computer algorithms, predictive modeling and AI. IPH marries UCI’s powerhouse health sciences, engineering, machine learning, artificial intelligence, clinical genomics and data science capabilities to deliver the most effective health and wellness strategy for each individual person and, in doing so, confronts the linked challenges of health equity and the high cost of care. IPH is part of UCI Health Affairs, and is co-directed by Tom Andriola, vice chancellor for information, technology and data, and Leslie Thompson, Donald Bren Professor of psychiatry & human behavior and neurobiology & behavior. IPH is a comprised of seven areas: SMART (statistics, machine learning-artificial intelligence), A2IR (applied artificial intelligence research), A3 (applied analytics and artificial intelligence), Precision Omics (fosters translation of genomic, proteomic,  and metabolomic research findings into clinical applications), Collaboratory for Health & Wellness (provides the ecosystem that fosters collaboration across disciplines through the integration of health-related data sources), Deployable Equity (engages community stakeholders and health-equity groups to create solutions that narrow the disparities gap in the health and wellbeing of underserved and at-risk populations.) and Education and Training (brings data-centric education to students and healthcare practitioners so they can practice at the top of their licenses).