The recent presidential election provided a fine example of how the analytics community can go wrong—surveys and polls can lead to erroneous conclusions if we don’t consider the impact of human behavior. Many presidential polls failed to account that people may provide false information to avoid looking different from social norms, or in the face of authority figures. The healthcare data analytics community can discern a key lesson from presidential election polling: Human behavior is not always predictable, and if the analytical community fails to apply scientific rigor, then results will have inaccurate conclusions.
To avoid errors in process, the healthcare data analytics community must rely on solid methodology that includes understanding the phenomenon or concept as we begin to build models and interpret results. Like the pollsters in the presidential election, it’s essential for the healthcare data analytics community to understand the impact of human behavior.
An important area to apply our understanding of human behavior can be seen in the effort to engage our patients and the larger community to adopt healthy behaviors. For instance, in the healthcare industry, patient engagement is an important strategy to improve patient care. Organizations are implementing technology and collecting data through electronic health records and patient portals, and they now need to focus on developing patient engagement strategies and using data from these (and possibly other) sources to help patients become active and engaged managers of their health.
What Makes Patients Tick?
Truly engaging and activating patients requires a shift in mindset. Instead of lamenting the sense that patients are failing to follow instructions, healthcare providers need to have a better understanding of what makes their patients tick, how to support health behavior change, and how to develop automated, scalable population-based interventions while customizing interactions with patients.
For example, medication nonadherence rate is another place to start when looking for opportunities to account for human behavior in healthcare analytics. Despite efforts, there is still ample opportunity to improve medication nonadherence, which is associated with poor health outcomes, progression of disease, and a cost of billions per year.
Patient-related behavioral factors have been studied and they include perceived side effects, a lack of understanding on the necessity of the medication, and a belief that the medicine was too expensive. Clearly human factors are greatly influencing medication nonadherence. So how can we use data analytics to help understand when patients have these concerns or fail to recall or even actually lie to us about taking their medications? How do we make room for the reality of human behavior?
During the election, the pollsters gave us a great lesson in what happens when you ignore human behavior. With a rigorous approach that includes data that explains demographics, social determinants of health, and an understanding of human behaviors, providers will be better equipped to engage in partnerships with patients to successfully influence the way they eat, sleep, exercise, manage stress, and take prescribed medications.
