UMass Memorial Health is working with Google Cloud to use the tech giant’s data and artificial intelligence capabilities to better connect patients with difficult-to-access drugs for conditions like diabetes and heart disease, Healthcare Dive has learned.
The partnership is a bid to improve population health, according to Google and UMass executives.
It should also help lower costs for UMass, a large integrated health system in central Massachusetts, as hospitals continue to look for ways to tamp down rising expenses.
UMass is first focusing on cardiometabolic conditions because they’re common in the system’s patient population, according to Michael Hyder, the executive director of UMass’ Center for Digital Health Solutions.
The conditions are also targeted by a number of novel drugs that have recently come to market or been approved by regulators, like GLP-1 receptor agonists for diabetes and obesity or SGLT2 inhibitors for kidney disease.
These medications can be hard to access, given they’re typically prescribed by a specialist and have sky-high list prices. They can also have serious side effects that need management.
As a result, they’re frequently underprescribed despite significant demand: Hyder estimates only 30% to 40% of UMass’ patients that could benefit from the medications are actually being prescribed the drugs.
UMass already uses algorithms to analyze its data to find patients who might benefit from more intensive interventions. But the program’s algorithm is simple, only using four or five different variables, Hyder said.
Medical workers can tell whether a patient has a heart condition, but not other circumstances like food or housing insecurity that might affect their ability to adhere to a treatment plan.
“We’re good at identifying patients who need colonoscopies and mammograms and preventive immunizations and such,” Hyder said in an interview. “Where we and others are trying to get to next is how do we attack population health?”
In the new partnership, UMass will use Google’s data and AI capabilities to identify a list of patients who might benefit from the drugs before looping in their primary care physicians. If after speaking with their primary care doctor a patient opts to take the drugs, UMass’ team of nurses and pharmacists will meet with them virtually to discuss intake, logistics of medication usage and cost.
After a patient is put on a drug, that team of providers manages their care, including specialists to virtually consult with the primary care physician.
Currently, UMass has 500 patients in the program. After the system stands up Google’s predictive model — expected in three to four months — enrollment could swell to “tens of thousands of patients,” with much of that demand driven by obesity, according to Hyder.
UMass wants to expand the program to more clinical use cases down the line, including in rheumatology or neurology, to combat conditions like arthritis and dementia as the U.S. population ages, Hyder said.
UMass stressed weaving in Google’s analytics will give its medical workers more ability to personalize care, improving outcomes. It should also help the system financially through levers like lowering administrative burden and reducing utilization of the emergency room, according to Hyder.
Despite signs of improvement, many nonprofit hospitals are reporting they’re still swamped by rising costs that outpace revenue. That hasn’t been the case for UMass, though margins continue to be slim. The system brought in operating revenue of $3.8 billion in its 2023 fiscal year, up almost 8% from 2022. Its expenses of $3.7 billion rose 3% from the year prior.
UMass says its partnership with Google could also lead to the creation of new therapeutic protocols by boosting ongoing research and development efforts.
The program plans to refine the models and publish outcomes from the program later on. As the models become more accurate, they could potentially be used to help clinicians make prescribing decisions, Hyder said.
It’s the latest hospital partnership for Google. The tech giant continues to ink deals with health systems to leverage its data analytics and cloud capabilities to free up their patient data, which is often locked up in silos.
Google’s other health system partners include for-profit giants HCA Healthcare and Community Health Systems and academic medical system Mayo Clinic.
Technology like generative artificial intelligence can then help clinicians make sense of that information, said Aashima Gupta, head of healthcare for Google Cloud.
Technology companies have raced to develop and peddle generative AI to providers over the past year despite concerns from some industry watchers about the technology’s unproven efficacy in real-world settings. Google launched new healthcare-specific generative AI models in December.
