Home  |   Subscribe  |   Resources  |   Reprints  |   Writers' Guidelines

June 2017

Industry Perspectives: Blame Analytics for Slow HIT Progress
By Gurjeet Singh
For The Record
Vol. 29 No. 6 P. 32

EMRs have taken a lot of flak in the provider community. They are accused of being cumbersome, disrupting workflows, and failing to deliver enough usable insights to significantly improve clinical outcomes. But it is really the failure of health care analytics, which has been layered onto these EMRs, that is preventing the industry from unearthing game-changing best practices.

The trouble with most analytics platforms is that they are constrained by what clinicians already know, or suspect, are both the problem areas and the pods of greatness within their organizations. Queries drive current analytics, but what if clinicians aren't asking the right questions because crucial care patterns are buried within enormous data sets? After all, it's tough to find something you aren't looking for.

With more data come more potential questions, decreasing the likelihood that one generating value for patients, providers, or payers will emerge. Even when the right question is asked, it serves only to confirm a previously held belief.

Some experts believe that predictive analytics solves this problem, but it, too, is hypothesis driven—just in a different way. In predictive analytics, the choice of variables and algorithms are, in effect, guesses as to what will produce the best outcome.

Ultimately, both data and predictive analytics are flawed.

A New Analytics Framework
What's needed is an approach that unearths new trends to deliver a host of meaningful insights to clinicians, a technology solution that combines the best qualities of human intelligence (artificial intelligence) with the finest computing capabilities that exceed human ability (machine learning). When these technologies are operationalized systematically across an enterprise, it's termed applied artificial intelligence (AI).

Applied AI has begun driving care improvement, reducing costs, and improving clinical and financial decision making across health care enterprises and the entire health care continuum. Applied AI is not a concept, but rather a series of intelligent applications targeting discrete health care problems—from clinical variation to population health.

These applications have a collection of capabilities that make them intelligent—of which all need to be present. Let's examine those capabilities.

Discovery: Intelligent applications must support both unsupervised and semisupervised discoveries. These capabilities are quite rare but serve as the foundation for industry efforts to move past hypothesis-driven inquiry. In practical terms, this means that an intelligent application considers all data and all possibilities within those data to detect the patterns, groups, or anomalies that elude traditional approaches.

Using their own record systems, including EMRs, financial data, patient-generated data, and socioeconomic data, health care organizations can automatically discover patient groups that share unique combinations of characteristics. These groups can then be used to tailor and personalize diagnostics and care paths, for example.

Alternatively, health care organizations may also discover unique patterns or outliers within their claims data to aid in member retention and preventing fraud and waste. Unique to AI, this type of holistic discovery improves prediction and makes operational insights possible.

Predictions: Intelligent applications must predict with high accuracy. Holistic discovery enables even better predictive models through the unbiased creation of groups or the identification of patterns. Superior prediction gives health care organizations foresight into future needs, costs, disease burdens, and patient risks.

For example, intelligent applications can determine the patient groups projected to have the highest escalation of costs over time, as well as other outcomes such as the conditions likely to appear for each group and an individual's predicted change in utilization. Predictions, which can be made across multiple targets, are multifaceted, considering all factors—whether they're health care-related or not—occurring outside of the health care system.

Justification: For users to be confident to act upon its recommendations, an intelligent solution must justify its predictions, discoveries, and actions transparently. For example, a health care app may reveal differentiating characteristics of patient risk trajectories, what factors make them high or low risk, and descriptions of individual factors that lead to variation in cost and quality. Justification is key because without a thorough understanding of the "why" behind predictions, organizations are unable to adopt AI into day-to-day decision making.

Action: An intelligent system that is not effectively operationalized will become less intelligent over time. Actionable information that guides and augments human decision making is what makes AI a part of daily operations. For these systems to deliver optimal value, they need humans in the loop providing feedback and governance.

Whether it be a recommended care path or a detailed risk profile, intelligent applications allow organizations to collaborate on the best actions tailored for each patient population or physician. Across the care continuum and within health systems and health plans, this allows organizations to better assess individuals and prescribe the best course of care.

Learning: Intelligent applications "learn" to improve predictions over time. As more data are analyzed, the technology learns from these complex data points. Whether it be claims, medical records, or socioeconomic data, AI taps into this information to generate more accurate, personalized predictions. Further, intelligent apps learn the impact of actions over time to support improved decision making.

Applied AI in Action
A large hospital system decided it wanted to reduce clinical variation across its enterprise to improve outcomes for all patients. It implemented machine intelligence, including unsupervised machine learning techniques that run algorithms using the system's own data—not benchmarks—to uncover actionable insights. The technology correlates and analyzes EMR and financial data including treatments prescribed, procedures performed, drugs administered, length of stay, and costs per patient. The goal was to discover and refine clinical pathways that are optimized to drive higher quality of care and lower costs.

The machine intelligence solution identified a group of orthopedic surgeons who consistently had better outcomes with knee replacement patients. Specifically, these patients had shorter hospital stays and shorter time to ambulation than other total knee surgery replacements across the system. The solution also explained why this was occurring: The physicians producing better results prescribed a unique, not widely used medication at an earlier postsurgical time than their peers. The medication reduced patients' pain, which allowed them to get out of bed and walk around sooner, improving outcomes and reducing costs.

Prior to these findings, most clinicians were unaware of the success being produced by the medication variance. Now the hospital system has operationalized these best practices throughout each of its hospitals, decreasing costs for knee replacement by more than 5% and increasing patient satisfaction.

How to Make the Most of Applied AI
As more health care organizations begin to see the value of applied AI, they may worry that more robust technology means an increase in technical headcount. But an important component of a successful applied AI strategy is that it leverages the unique capabilities of both machines and humans.

New data scientists likely would not have the subject matter expertise needed to recognize and deploy any meaningful insights that may surface through AI. Meanwhile, domain experts, who are best suited to learn from the data, usually do not have an interface to read the data. Typically, subject matter experts interact with data using rudimentary applications such as Microsoft's PowerPoint or Excel.

The final component to a successful applied AI strategy is to wrap the results of machine learning and AI into customizable, business-facing applications. Through such technology, precise insights such as the optimal way to perform surgical procedures can be uncovered.

It's critical that the results of machine learning and machine intelligence actually make it to clinicians, instead of ending up siloed somewhere in the IT department. The successor to health care analytics must not only be more powerful and more precise but also more user-friendly.

Even when the next generation of EMRs are deployed, it won't eliminate the limitations of analytics. Health care can soldier on with insights that only marginally move the needle to improve outcomes and lower costs or it can combine AI with powerful machine learning to turn enormous datasets into business insights that really matter. Then, it can deliver those insights via easy-to-use business applications to the best clinician minds to operationalize this machine intelligence approach across the enterprise.

— Gurjeet Singh is chairman of the Board of Directors at Ayasdi.