Unlocking the Power of Predictive Analytics
By Thomas J. Van Gilder, MD, JD, MPH, chief medical officer and vice president of informatics and analytics
As healthcare professionals, we have to contend with the weight imposed by the idea of big data — often to the detriment of practical clinical action. For several years now, thanks to healthcare technology and innovation, clinicians have gained access to unprecedented quantities of patient information in real time. Unfortunately, the volume of healthcare data has been burdened by assumptions about the importance — namely, that this abundance of data will naturally drive better population health outcomes, as well as lower costs.
Bringing Data and Analytics Together
However, data without an analytic component cannot deliver value to the busy practitioner; the massive healthcare information technology (IT) industry sells its products on the idea that the vast amounts of data can be aggregated, processed and presented in a way that necessarily drives practical action at the point of care. A more specific interpretation of healthcare data lies in predictive analytics — data modeling that can offer doctors and hospitals a better means to recognize individuals that are at risk of being readmitted (for example) — as well as those who could benefit from lifestyle modifications or preventive intervention.
Predictive analytics can also help physicians recognize instances in which relatively expensive, often-overused interventions (like antibiotics) could be more detrimental than helpful; or instances in which an individual might require a more aggressive course of action.
Transitioning from Episodic to Proactive Care
Big data analytics present an opportunity for population health management strategies that are more fully engaged with prevention than the healthcare industry is currently today. These predictive models could help determine a care plan that is forward-looking rather than reactive, which can help head off future costs for institutions and health partners; the same kinds of models can also be applied more broadly, to groups of individuals, which is where the population health potential could truly come to fruition. By some estimates, effectively leveraging big data analytics could result in more than $300 billion in healthcare cost savings a year in the US, particularly reducing inefficient spending in research and development, clinical operations and in public health expenditures.
However, on the ground, we’ve seen a big difference between the potential to drive action and the actual decisions that are made in real-time clinical practice, which rely upon a physician’s existing body of knowledge and experience, as well as the data models presented by analytics; physician knowledge and analytic models can work synergistically in diagnosis and action, but this is not always the case. What’s more, the process involves multiple stakeholders, oftentimes not able to refer to a single, comprehensive view of a patient’s medical profile.
Balancing Big Data and Clinical Practice in an Actionable Way
As the industry stands right now, we’re still trying to figure out how to balance big data and clinical practice in an actionable way. Physicians will require training programs that support the latest innovations in healthcare, including how to holistically integrate analytical models into their workflows. Modernization of physician training is critical to improving population health outcomes.
Further, predictive analytics that recommend intervention or lifestyle modifications won’t do any good without informed patients who understand that they themselves are an active part of the care team. Patients need to feel empowered to take control of their own health, as well as equipped to understand how their behavior and habits outside of the hospital are every bit as important to their health as the professional interventions within the clinic walls.
Unlocking the Power of Predictive Analytics
In our work toward building and maintaining healthy populations, we’ve recognized that predictive analytics can hold huge potential to elicit clinical decisions that are data-driven and proactive, but the industry must fully recognize that data models are only helpful if utilized to elicit a basis for intelligent risk assessment that leads to practical action or intervention.
How are you leveraging predictive analytics to improve population health?