Sutter Health thinks it has found a way to use “predictive analytics” to cut down on costs and improve patient care — and it is willing to pay for it.

The non-for-profit health system is planning a partnership with a start-up called Qventus at two dozen of its California hospitals and 30 of its pharmacies in the coming months.

The project involves alerting pharmacists when patients need a change in medication — an example might be a transition from injectable Tylenol to to a cheaper oral alternative — or require extra support in how to take their meds. A second effort nudges providers to move patients more efficiently through surgery centers to prevent delays.

The ultimate goal for Sutter Health is to help its doctors, nurses and pharmacists focus on the patients with the most complex needs who need more help, and reduce expensive, unnecessary treatments.

Many big technology companies, such as Alphabet to IBM, are building analytics tools to help hospitals and clinics become more efficient. Many of these offerings claim to be “predictive” by leveraging advanced technologies like machine learning and/or artificial intelligence.

There’s a new appetite for these tools among health systems, given recent policy changes in favor of “value-based” care, which means hospitals are now being incentivized to improve the quality rather than quantity of patient care.

But Sutter’s chief health information officer Sameer Badlani says he often encounters more marketing than substance from technology players. Many founders are actually offering “diagnostics,” meaning they provide some insight into why something happened and/or occasionally point to a huge swatch of patients that are at risk.

“Telling me which 2,000 patients in a population might get a heart attack isn’t predictive analytics,” said Badlani. “Which patient in which bed will have a heart attack in the next 24 hours, a topic I did my earliest research on — now that’s truly predictive.”

Qventus takes a different approach.

The company was founded by former management consultant Mudit Garg, who spent years shadowing emergency room staff. His employees still spend days working alongside hospital workers to figure out how to be helpful, and not annoying with their notifications.

As he points out, many doctors are already overburdened with notifications, a phenomenon known as “alert fatigue.”

The company, which was formerly known as analyticsMD, has raised more than $15 million to date in financing from technology investors including Mayfield Fund and Norwest Venture Partners.

Its projects to date might be targeted in nature, but Garg says his company and others have an opportunity to shave “$100 billion of inefficiency” in the coming years. Qventus gets paid with a subscription model.

The key for those who stand out, he said, is to take the time to understand the pain that is felt in the “front lines of medicine” by those who are caring for patients.

“This was an important realization for us early on and a big reason for our success,” he said.