By assisting healthcare organizations in keeping a population healthy, predictive modeling can play a vital role in accountable care organizations. Chief Medical Officer at Optum Analytics Jeremy Orr, MD, MPH, provides insight on ways predictive modeling can be used in an ACO and in population health management.
"Predictive modeling uses regression analysis to project forward and try to predict future health events," says Dr. Orr. When using predictive modeling, "the quality of data and the size of the data set are extremely important, and in healthcare we have an abundance of both of these now."
"Healthcare organizations have some great information at their disposal," says Dr. Orr. "The challenge is finding the points to act on."
When applying predictive modeling, Optum presents a patient-centered view to healthcare organizations. The information "allows the organizations to line up providers on a variety of metrics," says Dr. Orr. The information also helps providers determine which patients to connect with to bring them into a care program.
Optum is using predictive modeling in a number of ways with its provider customers. For example, it "is using predictive modeling to determine which patients are most likely to visit the emergency department in the next six months," says Dr. Orr. "Optum is also using a tool to predict patients' likely expenses over the next year."
"Predictive modeling shifts organizations from being reactive to being proactive," says Dr. Orr. For example, High Point, N.C.-based Cornerstone Health Care used predictive modeling to search for diabetes patients who had not received care for their condition in a year. Cornerstone reached out to those patients to assist them with making appointments. "That one proactive step made a drastic difference," he says. The results were outstanding. The patients who Cornerstone reached out to used a third less acute care services, and their control metrics improved by a third.
Dr. Orr also provided Milwaukee-based Aurora Healthcare's predictive modeling success story. Using the information supplied by Optum, Aurora Healthcare reached out to 139 of its congestive heart failure patients to ensure they were getting the care they needed. The purpose of reaching out was to keep patients healthy and "prevent unnecessary hospitalizations," says Dr. Orr. By reaching out, Aurora Healthcare was able to reduce heart failure readmissions from 14 percent to 2 percent and decrease all cause readmissions by 30 percent for this particular group of patients. In addition to better outcomes, "the patients were overwhelmed their health system would reach out to them proactively," says Dr. Orr.
Along with the benefits of using predictive modeling, there are some challenges. Dr. Orr says many times there are cultural challenges that have to be overcome for healthcare organizations to reap the positive outcomes associated with using predictive modeling. This is because "physicians have to start paying attention to the patients that aren't right
in front of them," he says. This is a very exciting time in healthcare that allows physicians to focus on quality, health outcomes and the patient experience, but it requires physicians getting "more comfortable in depending on information that is provided to them about a population."
More Articles on Predictive Modeling:
Natural Language Processing Key to Predictive Analytics
How Predictive Modeling Can Help Identify Clinical Trial Eligibility
4 Steps to Utilize A Hospital's Own Patient Data to Improve Self-Pay Collections