Predictive Modeling Meets Patient Behavior

In January, Bill Fox, the senior director of commercial healthcare for LexisNexis, discussed troubling stats regarding the U.S. healthcare system and how predictive analytics can help to fix it.

Chronic ailments like heart disease and diabetes affect a growing number of citizens each year. Nearly 27 percent of the population suffers from 2 or more chronic diseases. Each year around 15 million more citizens will join this group. This section of the population amounts to nearly 75 percent of healthcare expenditures. In 2009, that number was $1.9 trillion.

Another issue is the age of our population. Nearly two-thirds of U.S. citizens over the age of 65 have multiple chronic diseases and roughly 10,000 Americans will turn 65 each day until the year 2030. Healthcare is expected to cost the country 25 percent of GDP in the near future.

Fox alluded to the retail, finance and banking industry’s use of predictive analytics to assist with decision-making for operations. These verticals turn to predictive models when estimating market reactions, and determining what products their customers are most likely to buy.

The healthcare industry is now beginning to harness the power of predictive modeling. Most examples are seen in clinical data, but the models typically fail to understand the patients. Assuming all data could be collected regarding comparative effectiveness and the course of a disease, cost-savings and beneficial results may not be realized until there is a better understanding of how to impact patient behavior.

Since most health insurers only use data regarding a patient’s condition to manage association with wellness and disease management programs, they miss behavioral data predicting that patient’s willingness to participate in those programs.

Even if the patient were to participate, insurers fail to collect data regarding their engagement in the program or if it suited them. One suggestion is to combine public data regarding an individual along with information about their condition. The idea is to focus on patients, not just diseases. The data can be used to create predictive models that will better assist doctors to make meaningful change in the behavior of patients with chronic diseases.

Through advanced analytics, Fox believes a kind of “intelligent case management” can be achieved through recruiting, retention and compliance. Using big data to create better predictive models, effectiveness can be studied by aggregating and comparing clinical, claims and public data, which can breed a better set of modeling.  ;

Predicting a program’s effectiveness in regards to patient impact could provide a new means to affect behavior. A theoretical “impactability score” would identify patients most likely to benefit from a given wellness program. It would also discern the likelihood of compliance and outreach required to facilitate participation. Items impacting adherence and best practices affecting patient behavior would also be taken into account.

The effectiveness of disease and wellness management programs increases greatly when they successfully change the behavior of patients with multiple chronic diseases. However, they have been fairly ineffective in achieving this goal. Changing the analytics model from understanding diseases to understanding patients may realize a better outcome in motivating behavioral changes.

Research in comparative effectiveness has shown improvement in outcomes by implementing basic steps like strict drug regimens, weight loss and exercise. Mostly simple and sometimes inexpensive, these treatments have a strong impact on patient outcomes. However, they rely on a patient’s behavior and adherence to the medical plan. So, improving compliance has the ability to lower costs and produce better outcomes.

Fox reminds readers the burden healthcare has on the U.S. economy. He believes creation of impact-based models, which incorporate more patient data, are a necessity to reduce the growing cost in the system.

Related Stories

Pervasive Technologist Sheds Light on Predictive Analytics

NetApp Partnership Targets Healthcare

Missing Opportunities in Healthcare Data