In my last post, I wrote about the variables that determined the accuracy of predictive models. Accuracy, however, is only half of the equation. The data also must be usable; that’s today’s topic.
Timeliness is a critical aspect of usability in predictive analytics. For a provider to deploy predictive modeling in their organization, their own data must run through the models in a timely manner. Otherwise, they risk predicting outcomes too late to do anything about them. Data should be extracted automatically and continuously. Some clinical data elements can and should be incorporated in as little as two days from occurrence. Claims data, on the other hand, is 2–6 months old by the time it is typically reported.
The information coming out of the analytics engine must also be actionable. Predictive models should predict events that have significant impact on the quality and cost of care. Importantly, providers must also be able to positively impact these events. Let’s say, for example, that a model does a great job at predicting which patients will develop congestive heart failure (CHF). Fantastic, but it is not clear that providers can play a role in preventing this outcome. So while it may be nice to know that a patient is at risk of developing CHF, it’s not necessarily information on which providers can take action.
On the other hand, for patients that already have CHF, providers can take concrete actions to help prevent hospitalization. Similarly, other chronic conditions such as diabetes, chronic obstructive pulmonary disorder (COPD) and asthma also have well-proven approaches for admissions prevention—and are great candidates for predictive modeling.
In addition, once outcomes are predicted, providers need to be able to easily see what actions to take, for which patients. They should be able to easily stratify patients based on their level of risk, and then prioritize resources accordingly. This risk stratification should also link to actionable lists that highlight each patient’s clinical status and specific care gaps.
For more on usability within predictive analytics, download “Predictive analytics: Poised to drive population health.”
–Jeremy Orr, MD, MPH, Chief Medical Officer, Optum Analytics