Strive Health improves key KPIs with homegrown machine learning



Strive Health, a value-based kidney care provider, noticed many of its health IT vendors, like the provider itself, operate extensively in the value-based care space and collaborate with accountable care organizations and health systems.

THE CHALLENGE

These systems’ models need to be manifested in Strive’s technology and workflows, and Strive was having a hard time finding a technology vendor that could provide that unique use case the provider required.

That spurred the need for Strive staff to build their own technology versus buying it. They needed a tool that helped with everything from identifying which patients should be a part of which care pathways while leveraging data science and machine learning modeling to understand what patient care characteristics best align with parts of Strive’s care model.

Staff wanted a technology that was customizable and could be integrated directly into the tools already being used by its KidneyHeroes – nurse practitioners, nurses, social workers, care coordinators and dietitians trained to provide highly specialized care and understand the intricacies of kidney disease. Further, the technology had to be able to be flexible as staff continue to learn what works.

“It is best to have that ownership to be able to upgrade and update any components of either the underlying data model or the user interface as quickly as we can,” said Matthew Temba, chief technology officer at Strive Health. “We never want our business to be able to innovate faster than our technology, which is why having ownership control is key.”

PROPOSAL

The risk stratification logic that many other value-based care vendors use looks at a much broader population, and so they are typically identifying all patients with late-stage chronic kidney disease (CKD).

“Our proposal and what our technology sought to focus on is identifying the most high-risk patients within that group who we can have the most impact on,” Temba explained. “We wanted to use our technology to develop individualized care plans for people with CKD, guiding them on a care path that addresses the issue early – preventing progression to dialysis or transplant.

“The first big piece for resolving the VBC challenge was around patient identification and figuring out who the patients with late-stage CKD and end-stage renal disease (ESRD) are who would benefit the most from our intervention,” he continued. “Our risk stratification logic identifies those who need a more extensive care team.”

Once identified, staff then could better manage these patients’ health, giving them more healthy days at home, reducing their healthcare utilization management by keeping patients out of the hospital, and generating shared savings by focusing on this preventative care through a value-based care model.

“In addition to being more precise about the risk stratification, the other piece is making sure our technology is integrated into our operations and presenting data, analysis and care plans to our care team within their existing workflow,” Temba noted. “This way, our care team knows that a patient who is complex in risk is going to need more care naturally, and they can be prepared with the right frame of mind when they approach that patient.

“Care teams then can use their clinical judgment in concert with the risk stratification to make the right judgment calls on the patient,” he added.

MEETING THE CHALLENGE

The technology Strive Health created is called CareMultiplier. It aggregates and standardizes data from a multitude of sources and leverages a suite of predictive analytic models that support deep insights into kidney patient data profiles.

“The model has always been about building technology for our internal care teams and our KidneyHeroes who are our primary users,” Temba explained. “This way, all of our patient-facing staff – from nurse practitioners and medical assistants to care coordinators and social workers – are using the customized workflows we’ve built.

“The user interface they engage with on a day-to-day basis to manage documentation and manage their care activities is all part of our technology as well as the data modeling and risk profiles we create are all embedded into that user environment,” he continued.

The model also is integrated with a few other systems, such as Salesforce, AWS (which is used as the cloud service provider), and AdvancedMD, an electronic health record system. When nurse practitioners deliver care and need to create a claim, they do that documentation in AdvancedMD.

“To accomplish these tasks, our KidneyHeroes are the ones making a direct impact,” Temba said. “As an IT team, we enable them to make the best judgment for each patient. It also has a lot to do with streamlining their workflow and giving them information and insights and allowing it to populate so that when they’re having a conversation with a patient, all of that information is readily available and easy to use.

“By embracing these technologies, healthcare organizations can improve efficiency and reclaim time in their daily operations to give back to patient care,” he added.

RESULTS

At the beginning of 2024, Strive Health rolled out a new assessment workflow that made a huge impact on initial documentation time, cutting it almost in half. That created more time for the KidneyHeroes to spend with patients and not just collecting assessment information.

“We’ve got a lot more work ahead of us this year on how to make their time even more valuable, which is a KPI we continue to track against,” Temba reported.

“We also can look at the effectiveness of our risk stratification model to showcase hard results,” he continued. “That is one area we’re always tracking and always improving. Our machine learning environment allows us to tune our models as we get new patients, making that another area where we are always focused on doing new development.”

Because of the insights and data analysis CareMultiplier delivers to care teams, Strive has achieved a 36% reduction in 30-day hospital readmissions, a 49% reduction in hospitalizations among high-risk patients, and a 67% increase in optimal starts – a planned start of renal replacement therapy by receiving a preemptive kidney transplant, initiating home dialysis or initiating in-center hemodialysis.

“One example is our work with Oak Street Health,” Temba recalled. “Throughout our relationship with them, they have trusted us to do more and more transitional care management (TCM). That is a testament to what they saw in our workflows. They saw our team as an organization that has been doing TCM for many years and this solidified their trust in us to use our TCM program on our shared patients.”

ADVICE FOR OTHERS

“Other healthcare provider organizations should have a long-sighted worldview, though it takes a lot of investment capital for a business to have that luxury,” Temba advised. “There are multiple external factors at play and you have to have that organizational commitment to do it. Custom building a technology to suit the operational and clinical model of a startup organization is one that is filled with uncertainty.

“Organizations must think about how to build reusable technology that is extremely flexible,” he concluded. “Finding that right balance, and more important, ensuring the relationship with the operational teams and the clinical teams is really strong so there are tight feedback loops, is critical to success.”

Follow Bill’s HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.

WATCH NOW: Seattle Children’s Chief AI Officer talks better outcomes through the technology



Source link

Leave a Comment