Panning for gold: Insights in the healthcare data ocean

The healthcare industry has mass quantities of data. The challenge is turning it into something useful. In this guest post, Brian Robertson, CEO of a company that integrates and simplifies complex big data from multiple sources, will explain how artificial intelligence (AI) can help hospitals do that.


During the various Gold Rushes of the 1800s, prospectors would sit for hours by a stream, day after day, year after year, patiently scooping up silt in their pans and sifting through it in the hopes of finding that one gold nugget that would change their lives forever.

Today, the new Gold Rush revolves around healthcare data. Used correctly, it has the chance to change millions of lives, not just one. But there’s a slight problem.

The technological maturity of electronic health records (EHRs), data warehouses and other sophisticated systems of record have enabled health systems and payors to absorb unthinkably massive quantities of data.

An annual research report by EMC and IDC predicts the digital universe will contain 44 trillion gigabytes of data next year (roughly one byte of data for every star in the universe, with gigabytes to spare). Nearly a third of that data will be collected and stored by the healthcare industry, according to a Ponemon Institute study.

Alarmingly, this flood of data will never crest. Dr. John Halamka, CIO of Beth Israel Deaconess Medical Center, predicted that every patient will add 4MB of data to his or her EHR storage every year – and this was before the steady ascension of wearables, apps and other consumer devices.

The challenge is about 80% of this healthcare data is unstructured. Because those “dark” data elements are difficult to identify and apply to business or clinical challenges, they have little inherent value. That makes it more like prospectors standing alongside a raging river, scooping up buckets of silt while looking for microscopic bits of gold. Clearly, an impossible task for any human.

But not for AI.

AI discovers the nuggets

The analysis and optimization of administrative and financial transactions, health records or other complex and repetitive tasks can quickly subsume even the most innovative enterprises. High-performance machines and algorithms can examine complex continuously growing data elements far faster, and capture insights more comprehensively than traditional or homegrown analytics tools.

AI has carved established inroads in multiple industries, including health care. Much of that innovation has been put to work solving clinical challenges, but more health system leaders are considering its value on the financial and consumer experience segments of their enterprise.

There are several reasons for this. First, clinical applicability is heavily regulated and strenuously tested for consistency, reliability and safety. True, administrative and financial tasks bear similar risks, but those risks aren’t nearly as amplified. As such, AI can be more readily implemented in this area, and the results are more immediate.

Second, AI is ideally suited to tasks that are both repetitive and complex, a common attribute on the non-clinical side of health systems. When an AI solution completes a task, the outcome is evaluated, and lessons learned are applied to make the next task more efficient. This mimics human learning, only at a speed and scale far beyond what’s possible for even the smartest individuals.

Third, the consumerization of patient populations has placed enormous demands on healthcare infrastructure. Financial and administrative processes, such as billing, appointment scheduling and communication preferences, to name a few, are particularly acute stress points.

Take self-pay as an example. Revenue cycle departments were built to primarily interact with commercial and government payors. Today, all health systems are adjusting to the patient-as-payor, where personalization and experience bears more weight than the business rules of an insurer, and touchpoints extend far beyond the revenue cycle department.

Thanks to high-deductible health plans (HDHPs) and other out-of-pocket obligations, 30-plus% of a large health system’s revenue could come from patient payments. Just as healthcare organizations are being asked to take on more risk by traditional payors, they’re also assuming more financial risk of non-payment by patients. As a result, health systems are beginning to see diminishing or even negative cash flow and financial margins.

Moreover, patient obligations aren’t a claim sitting in an electronic queue – most often, it’s a bill sitting on the kitchen table, along with a car payment, mortgage, utilities and the kids’ tuition. Who is this consumer and where are they in their life? Do they need financial assistance? Do they have questions about their bill? Is the bill’s size forcing them to consider delaying or forgoing care the next time?

Bottom line, healthcare payment is rapidly transitioning from a primarily B2B transaction between providers and payors to one that now includes an ever-increasing consumer element. These consumers have heightened expectations for the business side of health care, thanks to their purchasing experiences in every other aspect of their lives.

AI offers the possibility to uncover insights from across the enterprise, enabling health system leaders to develop consistent and meaningful strategies that optimize the patient experience while also improving the health system’s bottom line. This is because machine learning distills complexity, finds patterns within billions of data points and gives organizations data-driven insight into the best opportunities for improving the quality and cost of healthcare.

Launching an AI initiative

Health systems need to transition quickly to value-based care while delivering consumer-centric experiences. A lever for accelerating change already exists within every healthcare organization – and that lever is the organization’s own data. However, analytics involves a different set of skills, a different organizational mindset and a different suite of technologies. The meaningful use of data requires more than light-lift SQL queries, dashboarding or marginal enhancements to flowcharts.

That said, breaking ground on a net-new AI investment isn’t necessarily the heavy lift many fear it would be.

Like all healthcare technology implementations, a successful go-live begins with a solid strategy. For example, AI doesn’t need to solve all problems right out of the gate. Instead, consider a “crawl, walk, run” strategy that begins with automating specific tasks.

Iterate around your most essential pain points; AI is best done in an agile, experimental environment, rather than one that is broad and formless. Look at your available data, then pick a business process that has potential for optimization. There’s little risk in leveraging data around specific pain points. With each incremental iteration, you can move onto more ambitious initiatives. Next, create a culture and overall framework for rapid innovation. Set up a feedback loop that allows you to run experiments and gain results that provide value to end-users.

Striking it rich

Acquiring the infrastructure, technology and brain trust needed to uncover insights from incomprehensibly large and continuously growing data sets is the industry’s next great challenge. Millions of lives and billions of dollars of revenue and cost efficiency is at stake.

Yet unlike the prospectors of old who invested all that time and effort without any guarantee of success, the Gold Rush of AI is available today to any organization that makes a commitment to pursue it. It can take the unique complexities of the financial, administrative and consumer experiences sides of healthcare, and deliver results that are immediate and transformative.

Brian Robertson is the CEO of VisiQuate, a company that integrates and simplifies complex big data from multiple sources and presents the information as meaningful insights, actionable workflows and prediction pathways in the healthcare industry.


Source: Healthcare Business & Technology,