At Bloomington, Minn.-based HealthPartners, AI adoption comes down to one priority: trust.
“We believe fundamentally that it will be detrimental to our ability to serve our members and our patients if we release AI competencies just for the sake of getting caught up in the race,” said Maggie Helms, senior vice president and chief data, AI and digital officer of HealthPartners.
As the integrated payer-provider with the eight hospitals expands its AI efforts, it is prioritizing data governance, clinician alignment and measurable value over speed. In this Becker’s Q&A, Ms. Helms discusses where AI is delivering results — and why patient-facing agents aren’t live yet.
The discussion has been edited for clarity and brevity.
Question: You stepped into the position of chief data, AI and digital officer in January. What are your plans for the new role?
Maggie Helms: One of the things that’s really exciting, leading both our digital and platform engineering practices and our data practice, it gives me the ability to bring together two really important tech competencies that need to harmonize now more than ever.
Previously, I found myself spending a lot of energy trying to navigate needing clean, well-governed data to get value out of some of the amazing engineering work we’re doing.
The more we can integrate teams that bring both data engineering excellence with software engineering excellence, the more quickly and meaningfully we’ll innovate.
We have to connect anything we do to the people we serve and to the folks who are working on the front lines.
Q: What strategies have been most effective around data governance?
MH: We have a data lake house strategy. We are calling it our source of truth, our system of reference.
When you have a big company like ours that has multiple points of view … simple facts like what’s a specialty, or what might we call a given condition or symptom, you’ll get different answers from different folks, and that can erode trust.
A key strategy of ours is to have a source of truth… so that we can use technology to help facilitate the agreement that we need across our complex industry.
Consumer trust is probably the biggest imperative that we have given this AI revolution that we’re experiencing, and in order to establish that trust, we need to agree on the core facts.
Q: Do you have patient-facing AI agents live today?
MH: Not yet. Our approach is to test everything with our subject matter experts to make sure that it passes the test before we release it to our members and our patients.
If we have an agent that is disagreeing with a human expert, that human expert for us is everything. We’re human-centered in our AI development.
We believe fundamentally that it will be detrimental to our ability to serve our members and our patients if we release AI competencies just for the sake of getting caught up in the race.
So we are working on it, but we do not have member and patient-facing chatbots on healthpartners.com yet for that reason.
Q: Where are you seeing AI deliver value today?
MH: For clinicians facing AI, we leaned in early to an ambient digital scribe solution. We’re leveraging artificial intelligence heavily in our clinical practice with previsit planning, post-discharge summarization, as well as a series of other back office activities.
In our health plan, we’re working on workflow automation and efficiencies… We have experimented with what we call an agent assist.
We have some AI in member and patient-facing settings, but it’s light, it’s not agentic: recommending next best action, changing the makeup of our website, informing what type of clinical reminder is best timed for what moment for a given consumer.
Q: Are you seeing measurable ROI from these efforts?
MH: I would say that we have achieved measurable value. I will stop short of saying that we’ve achieved a measurable return on investment.
We’ve seen a huge reduction in unnecessary and preventable emergency department use… improvement in some of our key quality outcomes… increases in consumer engagement and satisfaction scores.
But thus far, we have not seen a measurable change in our operating margin as a result of these initiatives.
Q: Can you share more about how AI is helping reduce ED utilization?
MH: We have a few amazing innovations: a nice hybrid of innovative human practices, complemented and paired nicely by artificial intelligence and data analytics.
We have a wraparound solution that changes everything from how we triage to how we communicate via email, text message and our website.
We’ve been able to see a substantial reduction, everything from a 4% reduction in some to a 16%-plus reduction in others.
We feel really comfortable that this is a very real and sustained change of behavior.
Q: What are you prioritizing in prior authorization and scheduling?
MH: Both our care group and our health plan are leveraging AI and automation to improve their prior authorization workflow process.
Artificial intelligence will not be making decisions about what sorts of authorizations are ultimately approved or denied. It will be able to make suggestions.
For scheduling, predicting demand for shift-based workers helps us to be more prepared and meet the need in a timely fashion.
The post Why HealthPartners won’t get ‘caught up in the race’ on AI appeared first on Becker's Hospital Review | Healthcare News & Analysis.
Source: Read Original Article
