By Pranam Ben, Founder and CEO, The Garage
We are standing at the cusp of what could be.
Predictive modeling is at a nascent stage when it comes to healthcare delivery. Although the industry has made enormous strides in the last several years -- such as better anticipating outcomes among high-risk patient populations -- healthcare providers’ ability to act on the intelligence in a disciplined, rational, risk-adjusted manner still needs to be further developed. That’s where applied innovation comes in.
Going forward, advances in healthcare information technology must be designed to help clinically integrated networks manage large, complex patient populations. The goal should be to progress toward a future state where applied machine learning and artificial intelligence will empower physicians, health systems, payers and patients to anticipate challenges, improve outcomes and reduce costs.
Research becoming reality
Considerable headway has been made in the last several years, as health systems have begun examining ways to apply predictive modeling techniques to practice.
In the age of value-based reimbursement, these are the types of clinical and financial forecasts that organizations must be able to generate and easily interpret on their own.
Much more is on the way, particularly concerning individualized care. Preventive medicine, for instance, could be improved with predictive models that help make diagnoses more accurate and treatment regimens more precise. Genetic predictive modeling research promises to take preventive medicine a step further by helping providers intervene for conditions that have not yet advanced to the point where symptoms have emerged. By tailoring healthcare in this way, providers and payers can help reduce the 20% of care costs wasted every year,  while also improving clinical outcomes and treatment adherence.
Improving health literacy
From a patient-facing perspective, the future of analytics may also include leveraging predictive modeling to improve health literacy. A huge number of data sets and analytical models already exist. These instruments could be combined with patient-contributed data to help predict and fill patients’ clinical knowledge gaps. That way, patients could more confidently manage their own health and make healthier choices on their own.
Likewise, payers could support this patient engagement by developing algorithms that deliver accurate coverage and cost estimates to patients before they undergo treatments. Comparing affordable, high-quality providers should not only help reduce spending, but also encourage greater efficiency among healthcare organizations that want to stay competitive in their markets.
The tools to overcome practical challenges
The challenge facing providers today is a practical one: Predictive modeling requires strong data infrastructure, user engagement, staffing and other resources. Yet I believe applied machine learning and artificial intelligence can pave the way for more robust, mainstream predictive models in healthcare.
As algorithms and software tools grow more sophisticated, technology can do the heavy lifting by sorting through vast amounts of data to derive the intelligent insights needed to drive better care decisions and outcomes. Other practical benefits should include:
When predictive modeling does become more ubiquitous across clinically integrated networks, we are sure to witness clinical and financial performance breakthroughs unlike we have ever seen before.
 “Shocking truth: 20% of healthcare expenditures wasted in US and other nations.” CNBC.com. January 13, 2017. https://www.cnbc.com/2017/01/12/shocking-truth-20-of-health-care-expenditures-wasted-in-us-and-other-nations.html