This is the 3rd blog in our series “From data diffusion to insights with enterprise-wide data strategy”
An interview with Sameer Siraj, Strategic Product Management, Health Care Technology as a Service, Optum
In the third of our four-part series, Sameer Siraj describes how an enterprise data platform combined with machine learning and analytics supports contextualization. Contextualization identifies patient circumstances in order to modify the plan of care as needed. Sameer, vice president of Strategic Product Management, describes other near-term uses for big data in the payer organization.
Q1: The promise of an enterprise data platform is the ability to contextualize member data. What does that mean to a payer organization?
Today, data and the ability to analyze and apply it are central to how all organizations compete and collaborate. Payer analytics deployed in the past have solved small data problems in the organization. But they’ve fallen short of taking a whole view of the member. As we’ve discussed in previous blogs, value-based care and consumer demand for value are profit drivers. These drivers compel payers to make the data they manage work harder for the business. Doing so also helps payers adapt to industry-wide forces and remain relevant.
Contextualization is a process of identifying individual patient circumstances in order to modify the plan of care if necessary. It requires the ability to understand data from a variety and growing set of data sources. This provides a way to steer patients to the optimal care setting and provider at the right time. To do this, the payer needs access to provider claims, facility, pharmacy, behavioral and clinical data. Merging all of these data sets is necessary to create a 360-degree view of the individual and design the highest quality and lowest cost care plan.
Q2: Can you provide a scenario in which contextualization will help to optimize care?
Contextualization cannot happen in isolation. It requires an enterprise data platform that supports machine learning and other business intelligence tools. This enables the plan to be fully capable in delivering the next right action in care and proactively anticipate the needs of the patient.
Patient data collected in discrete systems may be useless. For example, let’s say patient Jane Doe is admitted to the hospital with a hip fracture. Her historical and real-time data is not available at point of care. With an enterprise data platform, the payer is empowered to create the insights necessary to optimize Jane’s care plan. With three-hour data sweeps, the payer will know when, where and why the patient was admitted. The payer will know she has diabetes. And that the wound sustained in a fall became infected while she was in the hospital. It will know whether she has a caregiver at home to help her during recovery or if a skilled nursing facility is required to optimize Jane’s recovery. The payer has all of these data points as well as the results of care on hundreds of similar patients by hundreds of providers in hundreds of care facilities. These insights enable the payer and provider to adjust care for a particular individual.
Q3: When building a business case, what are some other near-term applications for EDP and enterprise-wide analytics?
An EDP facilitates continuous learning algorithms that accelerate processes and deliver insights across a number of broad categories:
- Individual and population health care management
- Network design
- Master data management
- Financial management
- Health care data models
- Workforce optimization
For example, in operations, adjudicating pre-authorizations for medications and procedures can be time-consuming and arduous. But once the data is available, a complete picture of the individual’s health allows pre-authorizations to be adjudicated more quickly. On a context-enriched data platform with machine learning, the payer can quickly see:
- What previous treatments have been tried
- Whether the medication has severe side effects
- If the medication is very costly
They can also recommend threshold therapies
In terms of care management, the EDP can be used to understand emergency room utilization — even before it happens. It can recommend intervention before a condition emerges. A common starting point for artificial intelligence is population health. Using predictive models built on machine learning or cognitive systems, payers can use characteristics from current patients to:
- Predict patients at risk of chronic conditions
- Identify patients who are not adhering to care plans
- Identify providers whose patients are frequently readmitted
The ability to predict expensive medical events before they occur allows for earlier intervention and reductions in cost of care.
Q4: What is the potential of EDP to reduce costs?
In the past, there has been tremendous effort to collect and store patient data in EHRs and share it in health information exchanges. However, data have not been analyzed and broadly shared. Applying big-data analytics across the health ecosystem could reduce health care spending in the U.S. by $300 to $450 billion annually.
Opportunities for cost reduction are plentiful:
- Reduce test duplication
- Flag high-risk providers
- Accelerate claims processing
- Improve payment accuracy
- Reduce fraud, waste and abuse
- With the potential of genome sequencing, utilize information on individuals and groups of patients to determine susceptibility to disease and preventive measures
Sameer Siraj discusses the cost savings that can be achieved when computing data in the cloud. Listen now.
Q5: What is the connection between EDP and precision medicine?
Unlike conventional databases, EDP enables a health management approach called “precision medicine.” Precision medicine accounts for individual differences and environments and uncovers hidden correlations and patterns that can advance disease understanding and interception — and ensure delivery of appropriate therapies. Precision medicine replaces the traditional “one-size-fits-all” approach. While successful in many cases, the traditional approach fails in some. For example, in cancer patients, precision medicine aims to “genetically” understand the unique properties of each patient’s disease. Treatment can then target the right cancer drug at the molecular level.
In order to achieve these benefits with an EDP, the payer will need a health-focused cloud partner. This partner will include data scientists who can support and anticipate the discovery needs of the CIO and chief medical officer and other key stakeholders. They provide best practices in machine learning and artificial intelligence that uncork the tremendous value of data.
How might a CIO begin to construct a data strategy around a readmission use case?
Visit Optum.com/CIO to learn more about business use cases for the enterprise date platform and big-data analytics.
Catch up now on the entire series: From data diffusion to insights with enterprise-wide data strategy
About the Author:
Sameer Siraj is vice president of Strategic Product Management at Optum. Sameer leads product strategy and definition for the Health Care Technology as a Service portfolio. His team is responsible for developing digital domain products and services spanning infrastructure, applications and data for health care payers and providers.