At HIMSS23, UnitedHealthcare Senior Analytics Leader Talks AI—in the Payer Context

April 20, 2023
On Wednesday, April 19 at HIMSS23, Craig Kurtzweil shared with his audience how he and his colleagues at UnitedHealthcare are leveraging AI for its predictive capabilities—at scale

On Wednesday, April 19, at HIMSS23, which is taking place at Chicago’s McCormick Place Convention Center this week, Craig Kurtzweil, chief data & analytics officer at UnitedHealth Group, shared with attendees a payer perspective on the leveraging of artificial intelligence (AI) for population health management.

The description of his session, entitled “UnitedHealthcare Tests AI at Scale to Help Support Patient Care,” noted that “The future is here with Artificial Intelligence. It is embedded into most businesses but rarely directly into patient health due to challenges such as the ‘Proof-of-Concept to Production’ gap, operationalizing solutions in real-time, and realizing the full potential/value of AI. UnitedHealth Group and Teradata will present best practices and lessons learned to achieve speed, scale, and precision when using AI to improve quality of care, quality of life, and reduce costs while considering SDOH, equity of care, and the importance of mental health post-pandemic.”

To give attendees a sense of scale, Kurtzweil noted that most provider analytics initiatives might focus on populations in the hundreds of thousands, but that “At UnitedHealthcare, we have over 50 million covered lives, so if we use AI, it has to be big.” Indeed, the numerous runs that he and his colleagues are executing on “are are now running data multiple times a day,” and involve data and information on millions of plan members.”

Per that, Kurtzweil emphasized that he and his colleagues will not involve themselves in AI or other analytics efforts that are not at a high level of scale and are not truly meaningful; they won’t simply “run data for the sake of running data. “The data has to be actionable,” he said. “We need to make sure that whatever we get involved in leads to changes in a member's life. Our ability to do real-time risk assessment multiple times a day.”

What’s more, Kurtzweil said, “What we’re trying to do is not only to determine where a member is or has been, but where they’re headed—what pathway a member is on. Are they on a bad path in terms of their health? It becomes very important to us to prioritize and to focus on the highest-risk patients. And whatever we do has to focus on action.”

Kurtzweil emphasized the importance of strategy around artificial intelligence work, noting that the five key words for action are around the ability to “identify, assess, uncover, predict, and recommend.”

And then he added that, when it comes to leveraging AI to perform analytics that will help providers care for patients, “Solutions involving patients and providers can’t be general; they need to be at the right time and involve the right care, at the right scale.”

Ultimately, he said, “What we’re trying to do is to make sure that AI provides analytics that can lead to differentiated care, with the ability to do that on a real-time basis for over 40 million people.”

Among the areas that he and his colleagues have focused on so far, Kurtzweil said, are diabetes, heart failure, and end-state renal failure. Encouragingly, he reported, while “Claims data doesn’t give us the most accurate data” on renal failure patients, and “it’s also old,” their AI-driven analytics, even in its initial run, provided a level of 84-percent accuracy when it came to identify the organization’s renal failure patients who were at the highest risk.

In that regard, Kurtzweil told his audience, “We can now identify which states of renal failure members are in. And the ability to predict their next stages, is advancing forward; we can predict within the range of weeks or months when members will move on to another state of renal failure; the predictive ability has gotten much, much better.” Renal failure is also an excellent area in which to leverage AI, he added, given that “Costs for renal failure are very high, and the impact on patients is very high.”

Other disease states and areas of particular interest to him and his colleagues, Kurtzweil said, are diabetes and long COVID, the actual occurrence in the population is actually underreported for both diseases. In all these cases, a part of the longer-term potential, he said, is that “We can start to predict and prevent hospitalizations,” and can suggest subtly different treatment choices, as appropriate.

Meanwhile, Kurtzweil shared with his audience a graphic showing “headwinds” and “tailwinds,” meaning areas that will be challenging going forward, and those that have some advantages built into them. The “headwinds” he cited include “long COVID, the preventive care gap, behavioral health issues, and generational shifts”; with regard to generational shifts, analytics is determining that millennials are using in-person bricks-and-mortar-facilities-delivered care at a far lower rate than are Baby Boomers, while they are using ED care, urgent care, and telehealth at the highest rate of demographic groups. As for “tailwinds,” he cited the consumer enthusiasm for telehealth, improved personal hygiene (per infection, following the COVID-19 pandemic), and infectious disease protocols developed during the pandemic.

Another area in which AI will be very helpful, Kurzweil said, is in predicting deepening anxiety and depression among men, who analysis finds are far more reluctant to seek behavioral care early on than are women.

Most of all, he concluded, the true potential of AI is in predicting future trends, across populations and disease areas. Its potential for health plan leaders to collaborate with providers to ensure improved patient/member outcomes remains the most exciting element in the journey ahead.

Sponsored Recommendations

State of the Market: Transforming Healthcare; Strategies for Building a Resilient and Adaptive Workforce

The U.S. healthcare system is facing critical challenges, including workforce shortages, high turnover, and regulatory pressures. This guide highlights the vital role of technology...

How AI-Native Locating Intelligence Revolutionizes the RTLS market

Discover how leveraging an RTLS solution with artificial intelligence as the location engine can increase efficiency, improve safety, and elevate care without the compromises ...

Harnessing the True Power of Cultural, Clinical and Operational Data

Optimize healthcare performance by combining clinical, operational, and cultural insights. A deeper understanding of team factors improves care and resource management.

How Digital Co-Pilots for patients help navigate care journeys to lower costs, increase profits, and improve patient outcomes

Discover how digital care journey platforms act as 'co-pilots' for patients, improving outcomes and reducing costs, while boosting profitability and patient satisfaction in this...