Cedars-Sinai is creating a digital health strategy that uses artificial intelligence tools and virtual care to alleviate capacity issues in its in-person care locations.
Earlier this month, the Los Angeles-based health system launched Cedars-Sinai Connect, a virtual care app developed with digital health company K Health. A couple of weeks into the project, Cedars-Sinai CIO Craig Kwiatkowski told Becker's, "we're really seeing that pay off in the patient experience."
While Mr. Kwiatkowski is excited by the potential of AI to liberate clinicians from administrative work, he cautioned that AI is not a "silver bullet" to solve the staffing and burnout crisis.
To learn more about the health system's AI and digital health strategy, Becker's caught up with Mr. Kwaitkowski:
Question: How did Cedars-Sinai Connect come about? How does it work?
Craig Kwiatkowski: So essentially, we call it Cedars-Sinai Connect. It's a virtual healthcare option with physicians available 24/7, for urgent care issues, appointments and primary care needs. It's intended to alleviate and help with capacity and access pressures in our brick-and-mortar physical locations, primarily primary care and urgent care. It has the patient experience emphasis on meeting patients where they are. We're beginning to see that it's a modality that makes things easier for physicians. That was the genesis of that endeavor.
K Health developed the artificial intelligence technology and then we co-built the app experience that is out today. The workflow essentially starts with an AI-enabled question-and-answer chat that gathers relevant health details and symptoms from the patient and helps assess their medical concerns.
So patients, as they're answering, get information about their potential diagnosis via that AI-enabled symptom checker. They can then choose whether they want to have a real-time video visit with a physician or to schedule a visit in the future. If they choose a video visit, that intake information is gathered during the chat session and assimilated and presented to the physician. So the physician has all the necessary and relevant information to initiate a more seamless conversation with the patient.
In terms of the sort of feedback we've gotten so far, it's still early, we're a couple of weeks into the general availability launch, but the response has been quite positive. I think one of the ways in which it has been positive is because we gather all that information from the patient upfront. We're really seeing that pay off in the patient experience. Patients appreciate that the physicians have the relevant info and are prepared to discuss the patient's concerns with them right at the start of the visit. We've also heard feedback from patients that the AI-enabled workflow is easy to navigate. From a physician standpoint, they are appreciative of the efficiency of the solution that really leverages the AI front end to complete clinical intake and data entry assimilation, which is something that helps them focus more on patient care and less on the usual documentation burden. We're excited to continue to build the program and further refine the digital tools.
Q: How do you see these AI-enabled tools contributing to combating physician burnout and the healthcare staffing crisis?
CK: It depends on how broadly we define AI. We've used AI for years, mainly in all the rules-based machine learning categories. The way in which we've deployed the technology has a significant impact on quality and safety and hospital operations, but they've never really solved the clinician burnout, problem or challenge. These new generative AI tools, the large language models, really have enormous potential. I don't see any silver bullets necessarily, but it does seem that we're on the cusp of being able to build solutions that can begin to bend the burnout curve.
Provider organizations like ours have been moving from paper to digital for two decades now. Building the EHR foundation has been transformational in many ways, but it has often been at the expense of burdensome workflows for our frontline caregivers, with a heavy emphasis on documentation and coding. The good news is now that we have all this information digitally captured and available at our caregiver's fingertips. The bad news is it's a lot of information and it's a bit overwhelming. We're finally at a point where that digital foundation we've been building for so long will begin to pay off as we're able to embed these AI tools within existing workflows to reduce some of that burden. These tools have great potential to alleviate some of those pressures, reduce workload, decrease some of the daily staff frictions and allow clinicians and caregivers to focus more on patient care. We're still in the early stages; there's a long way to go. It's essential that we implement these tools thoughtfully to make sure new things are enhancing, assisting and mentoring; they are not here to replace human interaction or human judgment.
Q: How do you ensure that health equity is prioritized when building AI models?
CK: So one of the core pillars of our AI strategy, as an organization, is centered around sound and ethical use of AI and these tools. We really firmly believe that establishing ethical and unbiased standards for these technologies is crucial to get the sound use and to their adoption. It's our obligation to ensure responsible, evidence-based, equitable deployment. Neglecting any of those things could worsen existing disparities.
We created a framework for ethical deployment and use of AI. So, that framework and those policies are really designed to ensure that AI benefits patients, physicians and the community. It advocates for appropriate oversight for safe use, effective and equitable use.
The framework identifies who might be impacted, how the models are trained, what data is used, retraining capabilities and assessment bias. We look to ensure that we have explainability and reducibility of the results. We have an obligation to ensure that we continue to monitor these tools and algorithms proactively for ethics and health equity. There's more and more analysis and research being published about the potential model drift in the way in which these tools can evolve as they're released into the wild.
Q: What projects or technology are you working with that you think could change healthcare?
CK: I feel fairly bullish on ambient voice technology. I think about pairing that with a large language model AI tools as having the potential, I don't want to over-dramatize it, but it could revolutionize the way in which care is delivered. In certain patient interactions, we're working on rolling out a couple of pilots very soon. We're hopeful that those outcomes will be positive.
I think these tools, in particular, have the potential to tangibly and substantially reduce burdens on physicians and other providers. They can hopefully provide a more natural conversational experience so that clinicians can focus more on patient care and less on documentation. Hopefully, this allows the patient to interact more and be more engaged in their care rather than sitting on the other side of the computer screen.