Skip to main content
Skip to footer

How Analytics Algorithms Can Improve Patient Outcomes

Published on 4/16/2018

Dr. Earl Steinberg, the CEO at xG Health Solutions, was a recent guest on The Cerner Podcast, where he discussed how analytics algorithms can improve patient outcomes and provider performance. Launched by Geisinger in 2013, xG Health Solutions empowers health care organizations to optimize clinical and financial performance through care redesign and management, analytics and clinical content and decision support.

Big data analytics and patient outcomes

There’s a lot of potential for the field of big data analytics in health care. Can you characterize what many hospital and health care systems are facing in terms of the influx of data today? 

Dr. Steinberg: As with almost everything in health care, leaders are facing unique challenges and exciting opportunities due to the sheer amount of data and the different kinds of data that they now have access to. Perhaps the greatest challenge we face is the seamless retrieval of claims data from insurance companies and clinical data from a multitude of EHRs that independent but affiliated clinicians are using.

Getting this data to a point where we can use it productively is a tremendous challenge. Once we’ve received data, it undergoes a number of process to normalize it into a standard format before we can even analyze it. This is not an easy or simple task, and even today, most providers are not equipped to handle the influx of data that we face. However, this task is critically important to population health management. 

There are many opportunities. In the short term, access new types of data such as the social determinants of health, patient preferences and patient reported outcomes can provide us with critical information, but we need access to this as unstructured data within the EHR. Over the next 3-5 years, we’ll see a greater impact from artificial intelligence, deep learning and the massive amount of data that we are compiling due to recent advancements in genomics. 

For the organizations that can interpret and leverage data successfully, the implications for patient outcomes are huge. How can an analysis of population health data alongside clinical workflows lead to change? 

Outcomes data is increasingly necessary to manage the populations we’re trying to provide care for. To influence behavior, we have to have actionable insights stemming from data that is integrated within the clinician’s workflow. High-level, descriptive analyses are incredibly valuable, but they won’t change behavior without this ability to pull insights. We can measure outcomes all we want, but merely providing dashboards within the workflow and portals with access to reports is not enough to drive change in behavior.

Payment reform and value-based care

What’s the relevance of the improved patient outcomes and delivery system financial performance brought on by data insights and the industrywide shift to value-based care?

The shift to value-based care cannot occur successfully without a sophisticated analysis of the data we are pulling within the clinical setting on a daily basis. When we use the term “value-based care,” we need to understand that payers are basing some of the reimbursements or payments to providers on quality, performance or cost, but there is substantial financial risk here. You cannot manage this model effectively without access to data and an ability to pull insights from this data. 

We also need to consider how value-based care is going to reduce waste, which opens capacity within health care. This encourages competition in the industry, so patients will be able to take their money to the most affordable and cost-effective providers.

Health care organizations are facing mounting pressures in the face of both the shift toward payment reform and value-based care. Can you shed some light on how data-driven insights can impact hospital revenue and the business of care? 

With effective population health management, one of the easiest things to reduce is hospitalizations and ER [emergency room] visits for ambulatory care sensitive chronic conditions – common conditions such as asthma, heart failure, emphysema and diabetes. For providers, hospitalizations for these common conditions are low-margin visits where health systems are unable to make very much. By preventing these under a value-based payment system, the health system can open up capacity to fill beds and slots in the ER with patients who need these services much more, and who tend to be much higher margin. 

By developing algorithms that can accurately predict hospital visits, we can improve the experience for both patients and providers. For instance, the point of population health management is to provide proactive health measures for individuals to reduce hospitalizations. Data analytics can helps use identify at-risk populations for seasonal maladies or chronic conditions. We also hope to develop more cost-effective provider networks, but this requires data on both the quality and efficiency of care. 

Data analytics and population health 

Can you discuss the future-state of data analytics and the implications for population health management? 

It’s no secret that artificial intelligence will transform health care industry in the next decade. We’re seeing new forms of data analyses take hold in health care that are quite different from the statistical approaches that were once prevalent. I share this enthusiasm in the field of AI, but we are a ways off before these are common in the clinical setting. Physicians have been reluctant to embrace the results of AI analyses because those who are producing these findings are sometimes unable to explain how we reached these insights. We want the comfort of understanding why.

We will certainly have more data liquidity in the coming years as data flows more easily from system to system, which is still a critical pain point in care. The need for interoperable technologies is not a new conversation. There has been progress made, but we still have work to do. There are other types of data, such as the social determinants of health, that are immensely important in patient care, that we should see more of in the next three to five years.

Do you want more health IT insights? Check out Cerner's Perspectives, a thought leadership publication.