Q&A: Rich in Data, Healthcare Presents Steep Integration Challenges
Healthcare's rich set of data holds the promise of improving the quality of care while reducing cost. First, though, it needs to be integrated and managed effectively.
- By Linda L. Briggs
- August 19, 2014
Donte London, a practice principal for the analytics and data management practice at HP, has extensive hands-on experience with IT in the health and life sciences industries. He has led large BI initiatives and worked on several projects for the U.S. military and commercial clients. London won an HP Innovator of the Year award in 2010 for his application of BI technology to facilitate chronic care management programs.
In this first part of a two-part interview with BI This Week, London discusses some of the technology challenges -- and opportunities -- facing healthcare in integrating large quantities of data.
TDWI: How much of a challenge is information integration in the healthcare sector, and what opportunities does it present?
Donte London: It's a huge challenge. It's historically been a big challenge back when organizations were dealing only with structured data inside the four walls of their organization. Today, the complexity has increased substantially to include unstructured, semi-structured, and structured data -- both within the organization and outside -- across the healthcare ecosystem.
We are now capturing patient data at multiple patient entry points, including the emergency room, during clinic visits and hospital stays, from worksite wellness programs, from social media, at call centers, and through mobile fitness applications.
That's a good thing, because all of that data can inform and enrich traditional research, segmentation, and practice. Integrating biometrics and other clinical data will allow us to use analytics to understand and measure wellness, apply patient and disease segmentation, track improvements, and use predictive algorithms to deliver personalized medicine. This rich set of new data holds the promise of improving the quality of care while reducing cost.
Certainly, there are challenges as well. Healthcare's unique vocabulary, taxonomies, and the numeric codes used to track activities across the care continuum make these types of data particularly challenging to use. On the other hand, the data is uniform, and a commonly understood lexicon across the entire industry makes it tailor-made for complex data integration.
Other industries are envious of the rich set of standardized data we have in healthcare, including:
- CPT (Current Procedural Terminology): codes for services rendered
- ICD-9/10 (International Classification of Diseases): codes for reporting the state of the patient
- SNOMED (Systematized Nomenclature of Medicine) taxonomy: comprehensive, multilingual, clinical terminology used as a standard for reporting of data
- Claims data
In addition, we have such things as revenue codes, discharge dispositions, and modifiers. All of this data helps in that it assigns meaningful quantitative value to otherwise disparate activities, but it also increases the challenge of integration.
Despite all of these ways of recording data, I believe the widespread utilization and understanding of this structured vocabulary will occur over time, thus accelerating the data integration process.
How are technologies such as the cloud, big data, security, and mobility changing healthcare?
They are all accelerating the ongoing process of moving the patient to the center of healthcare. Patients have been conditioned by experiences in online retail and banking, and have come to expect the same level of service from the healthcare industry. So far, most patients give low marks to healthcare, and their satisfaction decreases in proportion to the number of customer interactions.
These technologies will help to change that by capturing useful information about individual members across multiple interaction touch points, informing individual patient needs and preferences, defining new benefit plans, and delivering customized and consistent experiences for the patient. Technologies such as cloud and mobility will act as triggers, I predict, opening opportunities for better engagement between doctor and patient.
Some examples of that might include:
- Segmentation of the consumer population at the individual level to identify preferences in modes of communication and outreach.
- Providing relevant information in context for a patient's health profile. This includes facilitating collaboration across the health system to create a universe of "trusted" information.
- Increased demand for information and transparency, exchanging information between physicians and patients, and leveraging best practices for patient self-management. Mobile devices will play a bigger role in our healthcare experience, serving as advanced data collection mechanisms. They will gather and share research on outcomes and physician and hospital rankings. They will also be used with wireless sensor devices in disease management programs such as weight monitoring for congestive heart failure, diabetes, and heart rhythm monitoring.
- Secure methods and techniques such as encrypted thumb drives allow health consumers to safely transport their medical history electronically.
How can predictive analytics be used in healthcare?
Diagnostic, predictive, and prescriptive algorithms will all work together to drive a multi-channel interaction. Using analytics, that interaction can be personalized to each member, closing gaps in care, improving quality, and reducing cost. This will be accomplished by aligning treatment goals with opportunities to approach, touch, and persuade patients.
Analytics can also help us identify best-performing populations at the individual level as well as by state, provider network, and health plan. Predictive analytics can:
- Predict the most prevalent traits of individuals at high risk to contract a chronic disease
- Predict a patient's propensity to change, thus qualifying him or her for admission into a medical management program
- Identify candidate molecules with a high probability of being successfully developed into drugs that act safely and effectively on biological targets
How important is information integration in making analytics possible in healthcare?
Our ability to gather data of all types has greatly outstripped our capacity to use it. At the same time, scientific and societal advances are increasingly dependent on new insights and tools to exploit data effectively for timely delivery of relevant and accurate information and for knowledge discovery.
Information must be integrated to some degree in order for analytics to be useful. That doesn't mean that we must apply all data quality rules and master data rules to disparate data. However, it does mean that we must have some broader form of data integration to facilitate analytics. Technological advances have resulted in accelerating increases in size, diversity, and complexity of data in virtually all aspects of human endeavor. That makes the integration of all data unrealistic or impossible.
Information integration for analytics in healthcare is required to realize the full transformative potential of data in this increasingly digital and interconnected world. Information integration projects may support the diverse functionalities and processing needs for data, information, and knowledge from disparate and uncoordinated sources, or cope with the changing landscape of computing platforms at scales ranging from small mobile devices to potentially global-scale cloud and networked computing resources.
For example, we may simply need to know that different data records belong to the same person. Information integration should be able to recognize patterns across inconsistent data in order to advise users to take the next step in improving their health.
[Editor's note: In the second part of this interview, London discusses how changing consumer demographics, coupled with reform, have made the patient experience extremely complex. Fortunately, technology can help.]