Predictive AI Helps Providers Plan Patient Care

Health systems like West Tennessee Healthcare are using AI platforms that can map out a patient’s predicted stay, allowing clinicians to plan discharges and even coordinate with payers on insurance coverage.

Predictive AI is starting to show its value in healthcare by helping hospitals plan patient care, from length of stay and discharge to insurance coverage.

At West Tennessee Healthcare, executives say they’ve saved more than $5 million over the past year by using an AI platform from Xsolis to review patient data, enabling them to predict when a patient will be discharged and communicate with payers on authorizations and any denials.

“Prior to this everything was a manual chart review, so the case managers were reviewing everything manually, looking through every chart, digging for a lot of information and documentation,” says Debbie Ashworth, executive director for care management for the 90-hospital network. “Now they don't have to do as much. The AI has really improved workflow processes [and] timeliness of those reviews.”

Health systems and hospitals across the country are embracing AI at a rapid pace, but the opportunities for clear and demonstrative ROI have so far been few and far between. Using the technology to sort through data from the medical record and other sources and give clinicians insight into improving patient care is one of those use cases.

The key to this particular use case is that AI isn’t affecting clinical decision-making. It’s performing administrative tasks that had previously been done by nurses, doctors and other care team members, reducing time spent in front of a computer.

At West Tennessee, the platform creates a Care Level Score for each patient, which gives clinicians an expected length of stay and discharge time based on patients with the same healthcare concerns. That score is used by utilization review, case management, and physician advisor teams to map out care management.

The platform’s link with payers takes that one step further. Ashworth says two payers are now connected to the platform, giving health system clinicians and staff the opportunity to coordinate patient care and insurance coverage more closely, reducing time-wasting delays and lengthy appeals.

 “That has really improved our relationship with the payer and also the timeliness of getting the responses back,” she says, noting this cuts down on the phone calls, e-mails and faxes that often pass between provider and payer while the patient waits for care services. “The conversations we have with them are better [because] you’re both getting information that you can share.”

Ashworth says the biggest challenge to the technology has been change management. Staff and clinicians need a little time, she says, to get used to using the technology—and learning what to do with the time saved.

She also says the toll is “not 100% but very accurate,” and case managers always review the care plans and have picked up on errors.

“Just having that [data] right in front of us, to know that this patient should discharge tomorrow or … this discharge is going to be two or three days from now, [enables care teams] to prioritize a lot better and get to the appropriate level of payer,” Ashworth says. “Now they're able just to focus on what they need to do to get the patients discharged or get their insurance approved.”