Datapalooza17: 4 ways healthcare organizations are using predictive analytics to improve clinical care

It’s no secret that healthcare providers are integrating predictive analytics into their clinical workflow. Using analytics effectively is another story entirely.

On the surface, healthcare analytics has the unmistakable “cool” factor of any new, promising technology with the potential to improve care. But the real-world complexities of integrating predictive modeling into a complex healthcare system can rub away the shine of predictive analytics rather quickly.

“That’s where I kind of get stuck and I think that’s where a lot of people get stuck—how to implement analytics to improve care,” said Ravi Parikh, M.D., an internal medicine resident physician at Brigham and Women’s Hospital in Boston.

Parikh, one of the authors of an NEJM article published this week on how precision medicine needs a parallel initiative that focuses on care delivery, moderated a discussion at Health Datapalooza on Thursday in which several organizations outlined ways in which predictive analytics was making a tangible impact on patient care.

RELATED: Precision medicine will falter without a parallel focus on delivery, researchers say

Making analytics meaningful for physicians is critical to that success, but, as Parikh noted, that discovery is often filled with missteps. He pointed to a new readmission risk predictor tool recently rolled out through Brigham and Women’s EHR. Most of the physicians he works with have already deleted the tool from their dashboard because they view it as an annoyance.

RELATED: How hospitals put predictive analytics into action

Here are four organizations that found success with predictive modeling:

  • NYU Langone Medical Center: Yin Aphinyanaphongs, M.D., Ph.D., director of predictive analytics for the Center for Healthcare Innovation and Delivery Science at NYU Langone Medical Center, led a team of data scientists to develop an algorithm using a broad range of clinical factors that identified patients who were likely to stay less than two midnights. The tool is being used by the system to help physicians know when to place a patient on observation and avoid a claim denial from the Centers for Medicare & Medicaid Services.
  • Atrius Health: By combining claims data and clinical data, Atrius Health built a predictive model that focused on improving care for high-cost, high-need patients. By identifying key clinical factors, the nonprofit network of primary care and specialty care providers integrated a model into the EHR system in which a color-coded banner alerted clinicians to patients that were at risk for hospitalization within the next six months.
    Craig Monsen, M.D., medical director of analytics and reporting at Atrius Health, said the tool wasn’t particularly useful for physicians, but it did serve as a mechanism for medical secretaries fielding calls to quickly triage high-risk patients.
  • Tabula Rasa HealthCare: By integrating pharmacogenetic information into patient records, Tabula Rasa HealthCare worked with providers to improve medication therapy management and identify certain drugs that could negatively interact with other drugs or a patient’s genetic makeup.
  • Utah Health Information Network: As the designated healthcare information exchange in Utah, UHIN holds vast amounts of clinical and claims data that can be used to improve the health of communities throughout the state. Using predictive modeling, the organization developed alerts that notified providers when a patient was at risk for readmission or needed more targeted interventions.
    In one instance a health payer saw a huge spike in admissions among its members. Predictive analytics showed that the increase was linked to a change in the medication formulary, UHIN CMIO Mark Hoffman, M.D., said. The insurer quickly put that medication back on their formulary, and admissions leveled off.