See Wikibon and SiliconAngle's full video and article coverage from IBM Information On Demand 2012

For those of us hungry to learn more about how IBM intends to help transform the healthcare industry, a few days in Las Vegas at the end of October was time well spent. Twelve thousand participants descended on The Mandalay Bay Resort for the annual Information on Demand (IOD) event to hear IBM, its partners, customers, and other esteemed guests deliver the Big Data message. Searching for the healthcare thread throughout IOD was easy, as just about every main tent speaker included a healthcare reference, and indeed several of the presentations were dedicated to introducing new healthcare solutions and customer case studies.

The Move to Big Data in Healthcare

Tackling healthcare Big Data is both a challenge and an opportunity. In addition to access to insurance claims information, lab reports, operational and PACS systems data, access to electronic health records (EHRs) is becoming a highly valued instrument of change. EHRs contain a variety of structured and unstructured data rich with valuable information to support clinical decision support systems, quality measures, CPOE research efforts, as well as information coveted by big pharma to support clinical trials.

The biggest drivers for tackling healthcare data are the promise of improving patient outcomes and, in the process, decreasing costs. With the Affordable Care Act (ACA), aka Obamacare, incrementally rolling out carrots, including billions of incentive dollars via “meaningful use”, for providers to update and migrate to “modern” EHR solutions, and sticks, such as denial of Medicare claims for “avoidable” hospital readmissions starting in October 2012, most providers are scrambling to make adjustments to how they deliver or manage care as well as how they track their efforts.

For leading edge institutions such as The Cleveland Clinic, Kaiser Permanente and the Mayo Clinic that have made significant financial investments over the last several years, the move to EHRs and a focus on quality of care vs. fee for service has reaped benefits for their care givers and patients. However, the vast majority of providers are still struggling to make the transition or flat out resisting it.

IBM Healthcare Innovations

In the midst of this transition to quality and improved outcomes, IBM is making big bets that the healthcare industry will follow other industries, such as banking and insurance, that have embraced Big Data challenges as key to driving new business and revenue models.

IBM’s Watson is helping to usher in a new era of cognitive computing. In development for almost a decade, Watson leverages more than 40 IBM sub-systems including natural language processing (NLP) and content analytics, the predictive analytics capabilities of its SPSS Data Modeler, IBM’s UIMA architecture, pattern recognition, and several healthcare specific taxonomies and ontologies.

On display at IOD were several Watson-Ready solutions for healthcare that share elements of the underlying Watson technology. Earlier this year, IBM Content and Predictive Analytics (ICPA) for Healthcare was successfully implemented at Seton Health in Austin, Texas to address the problem of predicting which patients would have the highest probability of being readmitted to the hospital for congestive heart failure (CHF).

Predictors included more than 130 structured and unstructured data sources such as age, lab tests, medical history, and most importantly as Seton discovered, anecdotal information about the patients including where they lived, if they had family members near by or if they were able to drive themselves to doctor’s appointments. Interestingly, it turns out smoking was not a predictor for CHF re-admissions, but living in an assisted-living facility was a major predictor. Why? People in assisted-living care often have memory loss, forget to take meds or follow up on doctor’s appointments and care plans.

One of the IOD keynotes was given by Craig Rhinehart, who leads business strategy and market development for IBM’s Enterprise Content Management (ECM) within the IBM Software Group. Lately, Craig’s focus has been almost entirely on the healthcare industry. At IOD, Craig introduced Patient Care and Insights which uses similarity analytics capabilities developed by IBM Research.

According to Rhinehart, “A provider could examine thousands of patient attributes at once. That includes not only clinical attributes but also demographic, social, and financial ones. By assessing similarities of attributes in broad patient populations, providers can better anticipate disease onset, compare treatment effectiveness, and develop more targeted healthcare plans.”

During IOD, Craig and I spent 30 minutes or so on the Cube discussing healthcare with Wikibon and Cube co-founder Dave Vellante.

Importance of Big Data in Healthcare

“Context Accumulation” was the major theme in keynote speaker Jeff Jonas’ presentation. Since IBM acquired Jonas’ data analytics company SRD a few years ago, he has been “working on a way for an organization to make sense of a piece of data the moment it arrives, so that it {the company} can figure out what the smart thing is to do while it is still happening. Sense, and respond. It’s like knowing when to blink or duck.

Can an enterprise respond intelligently? Not today. If you visualize this, it’s almost no different than taking puzzle pieces out of a box and seeing how they fit together. I call it context accumulation. You’re only as smart as the puzzle pieces you have. The puzzle pieces incrementally accumulate. And so does the context.”

In the context of healthcare, the more information you have about a patient or a set of patients at the point of care or in a research environment, the better informed the care provider and the patient will be.

In the context of big pharma companies analyzing millions of individual EHRs going back perhaps 20 or more years, data scientists have discovered that unlocking insights derived from those longitudinal or historical patient records is invaluable in determining the efficacy of new drugs and determining which formulas have a higher probability of success. Indeed, EHR data can be more valuable than data gleaned from traditional drug trials which have limited capabilities to correlate illnesses or derive accurate data from an artificial environment compared to real-world data.

Today there is more data available to make more accurate decisions. Consequently Big Pharma is actively, but quietly, acquiring “de-identified” EHRs creating a highly profitable additional revenue source for care providers. PACeR Health in New York State is just one of many examples.

Context accumulation supports a data gathering model that provides a more holistic view of the patient with a variety of data sources well beyond just EHRs including, medical journals, lab reports, print streams (e.g. explanation of benefits (EOBs), claims data, bills), epidemiological trends, expert opinions and papers – any data source that may be useful.

Getting Analytics-Ready

In order for patients, providers, pharma, and payers to take advantage of available data, it needs to be in a format that is accessible to analytics solutions. Sadly, much of our healthcare data is either still on paper and microfiche or is being converted to non-searchable PDF formats or compressed in TIFF files where only the metadata is readable.

The PDF and TIFF file conversion “mania” solves the short term problem of getting EHR records “online” to meet a very low hurdle to qualify for meaningful use reimbursements and simultaneously renders it useless for most research purposes, for clinical decision support systems, and big pharma analytics.

Through its Healthcare Infrastructure Services, IBM enables providers and payers to meet the requirements of analytics readiness. A roundtable discussion led by Premier Healthcare CTO, Denise Hatzidakis, focused on the “data integration platform” and “data model” Premier is developing with IBM. Premier is a provider-owned entity that processes financial, operational, clinical, labor, and patient data for 200 of its owner organizations that represent 40% of the hospitals in the U.S. Premier shares best practices and benchmarking data with its owners to reduce waste, prevent infections at facilities, help improve quality of care, and decrease costs. Data sources are both structured and unstructured, internal and from external sources.

Premier owners have a variety of EHR solutions. According to Hatzidakis, one provider has 28 versions of the same EHR solution, so rationalizing and stabilizing data as well as bringing provider and payer data together is key. Premier has developed 300 “data cleansers” to help provide an aggregate view of data. IBM’s FileNet and Production Imaging Edition (PIE) solutions are integral to this process, helping convert paper records to accessible digital formats and intelligently capturing text and data held in EHRs.

Data Fracking

If data is the oil of the 21st century – only better because it can actually increase in value with use – then the “fracking” analogy works. Providers are sitting on a mother-load of data, but most lack the means to access it. Whether paper, PDF, or TIFF file format, PIE has the ability to convert paper to image and perform a full text index of the content and, if desired, “intelligently” capture all or portions of the text using IBM’s Content Analytics solution. PIE’s OCR and ICR capabilities can also “crack” PDF and TIFF file formats to expose the content to full text indexing and intelligent capture capabilities.

Healthcare Following Financial Industry?

Due to regulations going into affect in the last decade including SEC 17a-4 and changes in rules governing electronic discovery (see FRCP) or ediscovery, the financial industry has seen an exponential spike in requests to produce information or ‘evidence” contained in unstructured data included in emails, documents, images, social media messaging, and other forms of communications such as voice mail, along with paper. Savvy lawyers and the courts realized several years ago that online data was a more fruitful method of uncovering “smoking guns” or producing evidence germane to a court case, and that point-of-view has only been magnified given the recent financial crisis.

Initially, most financial organizations took a reactive posture to this increased scrutiny. Recently however, forward-thinking financial institutions have been much more focused on information governance programs and solutions that not only support ediscovery and litigation activities such as defensible disposal, but also those solutions that enable data to be shared across the enterprise and be used for various purposes including compliance, customer service, cross-selling, fraud detection, logistics, marketing, operations, and new business or product development. IBM’s Information Lifecycle Governance (ILG) solutions group, also prominently on display at IOD, was formed to address these very issues.

By most accounts, healthcare IT is roughly 10 years or so behind the financial industry. Now the healthcare industry is beginning to exhibit many of the same trends and market dynamics that were evident with the financial industry a decade ago, precipitating a transformation in the management and access of data – especially unstructured data which represents 80% of all electronic data.

Several apparent or emerging Healthcare IT trends include:

• Policy changes driving reactive solutions acquisition,
• Spike in ediscovery requests from government agencies and lawsuits,
• Proliferation of personal data (PCI, PII, PHI) in the wild across enterprise,
• Escalation of data storage costs and footprints,
• Poor data integration, and interoperability getting worse,
• New business cases emerging for data use

Integrating Patient, Provider, and Payer Views

For all the right reasons, whether policy driven or an undeniably good business practice, the integration of healthcare data is critical regardless of the source. Patients are demanding more information about the conditions and diseases they suffer from, the treatment modalities that caregivers are providing, and available options for health coverage. At the same time we are all concerned about privacy and the potential malicious use of our personal healthcare information (PHI).

Coincidentally, either due to policy changes or for practical purposes, more care providers are transitioning from fee-for-service to quality-care programs, while payers are struggling to stay relevant and adopt their business models. A catalyst for this transformation is data or access to relevant data.

To that end, IBM and several of its partners are integral to enabling the healthcare data eco-system. I spoke with Colin Shearer, who is responsible for IBM’s new Analytic Answers product, which leverages the cloud to address the needs of mid-size organizations including health insurance companies looking for insights into payment information and other applications. Here is a link to Colin’s Cube interview with John Furrier and Dave Vellante.

Partner Crawford Technologies exhibited its prowess for enabling print streams to become “analytics-ready”, offering an additional view of the patient for providers and payers who have IBM’s CMOD repository installed. Think about all of the billing, claims, drug reaction, and EOB information that can be mined for additional context. Unlike providers who possess deep patient information by “transaction” through continuity of care documents or lab reports within their hospital or provider network, payers view the patient across the spectrum of their reimbursable healthcare experience.

Partner DATASKILL made it to the Cube via its CEO, Nigel Hook, who discussed how they are implemented IBM solutions, including PIE, in major hospitals on the West Coast. Hook also made the point that unstructured data is critical to driving value for analytics.

I also spoke with David Mancusi, CEO of Fairfax Data Systems, an IBM FileNet solutions partner on the East Coast. Fairfax has experience implementing IBM’s Datacap Taskmaster solution which extracts information from document images for use in enterprise content management (ECM) and line-of-business systems. Taskmaster operates as a universal capture portal that transforms all forms of documents entering an organization including paper, email, fax, or other electronic formats.

Conclusion

Big Data is transforming organizations across many industries, including financial entities and leading-edge healthcare providers and payers. It is evident that the healthcare industry is experiencing market dynamics similar to those felt by the financial industry over the last decade due an increase in regulatory scrutiny and a desire to improve quality, decrease costs, and create new revenue streams. By design, the most visible systems where based on IBM’s Watson, and Watson-ready solutions are being applied are finance and healthcare.

IBM FileNet and PIE are playing a significant role in both industries, allowing data to be “analytics-ready” and available to care-givers or other knowledge workers, and IBM is making a strong case for why its vision for healthcare will continue to be relevant for decades to come.

Action Item: Healthcare CIOs, CMOs (Chief Medical Officers) and business executives need to have a strategic information management and data governance plan in place to take advantage of Big Data opportunities and not fall prey to tactical, reactive practices that sap resources and result in the adoption of practices and vendor solutions that quickly lead to dead-ends and under-optimized systems.

Footnotes: