Artificial intelligence has accelerated healthcare’s response to the COVID-19 pandemic, enabling faster understanding of at-risk populations and better management of healthcare organizations, AI experts said at the annual – and, this year, virtual - TCU Annual Health Care Forum, presented by the TCU Health Care MBA and the TCU and UNTHSC School of Medecine. Panelists at the forum, held Wednesday, were Steve Miff, president and CEO of the Parkland Center for Innovation in Dallas, and Sai Bezawada, IBM’s public sector leader in Texas. Moderators were Dr. Stuart Flynn, dean of the medical school, and Daniel Pulling, dean of the Neeley School of Business at TCU.
“Everywhere you look, (AI is) upon us,” Pullin said during the forum. “Over the last nine months, as we have all endured the COVID-19 pandemic, I think we have seen AI on full display.”
Q: AI can really be a broad term. Give us a sense for how you define and describe AI? What does it mean in your world right now?
Bezawadai: “Data is being created while we’re sleeping. AI helps us understand and get to outcoes. During the pandemic, we were able to accelerate how we solve these problems. AI was a critical set of capabilities.”
Miff: “My parents were both physicians, and they used to do home visits. They got to know the whole family. They had all those different insights in addition to knowledge of healthcare, but it was not scalable. (With advances in technology) we’ve been able to scale that data, (but) we’ve lost some of that personal component. What I believe AI can do is bring all this information together, bring back that whole person health, but do it at scale.”
Q: Help us understand inflection points as to future of AI, relative to COVID. Where was AI going before COVID? What have we learned about it that could alter course?
Miff: ”Before the pandemic, we saw increased targeted applications: identify risks upstream, predictive models that help identify patients vulnerable to sepsis 8-12 hours before admission, help identify at-risk populations. Those have been well-progressed applications before the pandemic. What we’ve seen also starting to accelerate is progress across areas of the ecosystem. AI itself cannot be that effective. Moving data around, making sense of it, next-generation data visualization, we’ve seen quite a bit of progress there. With COVID, it accelerated partnerships for better and broader access to data.”
Steve Miff
Q: Look in rearview mirror. When this all started, how did IBM approach this? What was the company looking to as their possible role during this pandemic?
Bezawada: “Our thoughts were how do we take the technology assets we have, and how do we put this in the hands of our research community, how do we put this in the hands of our frontlines, our doctors? We moved quickly. Two themes: business continuity, and enabling frontline researchers and colleagues in medical profession.”
Q: On Parkland’s partnership with local health authorities, how did that come together? How were you able to bring creativity of multiple people to provide such precise information in such a timely way?
Miff: “If you wait for a crisis, you’re probably too late. The partnership was already in place. There was trust there. We were able to reach out and formalize a data-sharing exchange with the county. Since then, on a daily basis, we receive 100 % of the cases and information for COVID-19 at the individual level. That also brings trust and a lot of responsibility (that required robust security). Just getting data in is a necessary first step. We really focused on where we can use that. We had four focus areas. 1) Ability to visualize on a real-time basis and provide to city, county, healthcare environments so they can track and understand progression, with privacy to the community. 2) We developed a proximity index to really understand at the individual level the exposure. 3) We developed a vulnerability index to allow us to understand who was at greater risk. And 4) We developed a capacity forecasting model for Parkland and other healthcare organizations so we could anticipate needs for beds, ventilators.”
Sai Bezawada
Q: To Bezawada, on IBM hearing in March from a neonatal unit in Florida that needed help in researching COVID treatments. IBM has also helped researchers, physicians and scientists accelerate COVID drug discovery. What did the hospital ask for? What was out of your wheelhouse?
Bezawada: “My parents also are physicians. From a medical standpoint, I had a vested interest. From a clinical side, we looked at the use of ventilators. How long do you keep a patient on it, (what are the) optimal settings. Not in our wheelhouse: (Personal protective equipment) shortage was big. It was right at the top. Non-traditional suppliers were stepping up to provide this. Hospitals did not have that as part of their supply chain. We launched rapid supplier connections, so this would build that trust. This would help find suppliers somewhere. We had the technology, but we really got out of our comfort zone so we could collaborate with that.”
Q: What happens the next time? We will have the tools, talent, will to tackle the next pandemic or healthcare scare. Will we do a better job the next time?
Miff: “I am so excited and pleased with how much progress we’ve made as an industry and how many new capabilities we’ve infused. Those partnerships we’ve developed will sustain and will continue. All those we’ll be able to leverage. There was a lot of concern about use of mobility data. That’s something we’ll use, in a way that protects individual privacy. (Data identified locations for testing sites and at-risk populations.) Some of those things will continue. We’ve already been able to pivot and use that for flu vaccinations.”
Q: A lot of this involves tracking cell phone movement. How do you accrue data?
Miff: Parkland uses data from “organizations that track that, not at individual level, but a summary. We’re not tracking the individual, but communities. You’re able to look at a block level, and understand that is a huge hotspot. As a group, what are the places they tend to visit the most? Home Depot, church. Where do we need to put access points to make it convenient, because they’re already going to those places? Where do you tailor the communication? Who are they? You can actually tailor the messages. That’s the evolution of being able to do this.”
Q: What are the supply chain implications of vaccine approval, getting vaccine to right folks?
Bezawada: “We’ve been testing a blockchain solution embedded with AI with pharmacy, transportation providers, airlines. It’s humanly not possible to keep track of every vaccine. I see a role for AI specifically in storage and making sure there is a level of comfort in which we can prove through data that a set batch hasn’t been touched.”