Three ways to use predictive analytics to identify patients best suited to RPM | Geneia

Three ways to use predictive analytics to identify patients best suited to RPM

May 15, 2018
Mark A. Caron, CHCIO, FACHE, CEO

Banner image of a remote monitored senior checking his vitals on a tablet

Without a doubt, numerous studies have shown remote patient monitoring works. A 2017 one that reviewed 19 previous studies, for example, found remote patient monitoring is an effective intervention to reduce heart failure hospitalizations and mortality. 

A Geneia study had similar positive outcomes. Specifically, the Geneia @HomeSM program:

  • Reduced hospitalizations: Hospital admissions declined 76 percent for monitored members and 31 percent for the control group.
  • Lowered costs: Per member per month (PMPM) medical spend was 50 percent less for monitored members.
  • Slowed disease progression: As measured by patient risk scores, the risk score of monitored members increased 29 percent compared to an increase of 58 percent for the control group.
  • Increased medication adherence: Pharmacy costs increased 18 percent for monitored members indicating a higher level of adherence to the care plan.
  • Improved member experience: Program participants reported an overall satisfaction rate of 96 percent.

Using Predictive Analytics to Identify the Patients Best Suited to Remote Patient Monitoring

What we also know is remote patient monitoring is not an effective care management intervention for all chronically-ill patients, not even all of those diagnosed with heart failure. Instead, a combination of clinical assessment and predictive analytics helps us determine the patients best suited to remote patient monitoring. 

At Geneia, we use predictive analytics in three important ways to identify the patients most likely to benefit from remote patient monitoring. 

  1. To create a base cohort of members most likely to improve their health and associated costs with remote patient monitoring, we recommend predictive analytic models that use the following inputs:
    • Propensity to engage in a remote patient monitoring program
    • Readiness to change as measured by patterns of activity such as enrolling in a Weight Watchers program, a recent hospitalization or a new medical diagnosis
    • Ability to self-manage their health
    • Date of last physician visit
    • Date of last emergency department visit and/or hospitalization
    • Date of last hospital readmission, if applicable
    • Medication adherence as a proxy for assessing program compliance
    • Number of different medications tried within the same class
    • Availability of a spouse or family caretaker
    • Physician office readiness to engage with remote patient monitoring biometric data

  2. We also apply predictive models for highest risk and costliest members and the probability of a near-term hospital admission to the base cohort to narrow the list to members best-suited to remote patient monitoring. 

  3. And once the list has been generated, we use predictive analytics to determine the most effective path for outreach to identified members.

To learn more about how we use predictive analytics in our remote patient monitoring program and to view our entire list of selection best practices, visit:

Banner image of clinical worker helping elderly woman interpret her vitals

Related Blogs