Can we predict favorable health outcomes?

We can personalize our browsing experiences with Google Chrome and our Netflix queue — so why not our healthcare experience? A resource deck by Rock Health explores the opportunities and challenges applying predictive analytics to customize and improve the patient experience.

By definition, predictive analytics is a tool to learn from historical data to forecast future occurrences.

The goal in applying these high confidence algorithms to heathcare? To predict the best actions and times to intervene in patient care and ensure more instances of favorable outcomes.

Predictive analytics also have the potential to lower patient costs if interventions correct patient health practices before more serious procedures are needed.

As cited in the deck, the American Medical Infomatics Association predicts that healthcare data will grow from 500 petabytes, as assessed in 2012, to a staggering 25,000 petabytes by 2020. To set the frame of reference, a petabyte is one million gigabytes.

So what can we do with all this data?

According to the report, two areas where predictive analytics could be used include readmission prevention and chronic disease management.

Teresa Wang, one of the authors of the Rock Health Report, says predictive analytics do what practitioners do already on a much larger scale. Practitioners “look at patient symptoms and apply training and experience to diagnose and predict the best treatment.”

For Vinnie Ramesh Co-Founder and Chief Technology Officer for Wellframe, a company dedicated to helping health organizations manage risk, predictive analytics differ in that they combine clinical medicine with psychosocial, behavioral and/or biometric data to understand the complex relationships between biology and unique factors.

Currently, companies who claim to use predictive analytics are serving the provider space. Though, predictive analytics could be used to aid diagnoses and recommend treatment.

According to the report, challenges to realize the potential of predictive analytics include:

  • Infrastructure exploration and creation to collect real-time data
  • Normalizing the data for biases
  • Guarding patient anonymity and ensuring compliance with HIPAA
  • Determining a means to visualize the data
  • Rendering the collected data actionable
  • Creating a rapid feedback loop to adjust the predictive algorithms so they remain accurate

Authors of the report envision data collection occurring in one of two main ways. Patients could submit data into the predictive algorithm themselves. Or, patients could provide the information to doctors who then input it.

Right now, patient-reported data is the third largest source of data in the predictive analytics space.

Interest in applying predictive analytics to healthcare isn’t expected to cease anytime soon. Since 2011 investors have spent $1.9 billion funding predictive healthcare ventures, according to the Rock Health Funding Database.

Realizing the power of predictive analytics is only possible by marrying efficient use of collection technologies with constant surveillance and attention to maintain the accuracy of the algorithms.

View the complete Rock Health Report here.

Talk to founding sponsor MentorMate more about the potential to redefine data collection in your business.

Photo courtesy of robuart.