Using Data to Unlock the Patient Experience


Breaking down big data.

The topic of big data has rocketed to stardom, often popping up as a business publication's lead article, headlining a conference and even commandeering casual conversation. And there's good reason for it; the collection and analysis of patient demographics, preferences and product feedback helps companies develop and communicate the value of their products and services to their end users.

Just like Nike wants to know when their ideal consumer runs, on what surface and for how long so the footwear giant can design the perfect running sneaker, pharmaceutical marketers are hot on the trail of differential patient types, contingent on a patient's preferences for a particular medication profile - or lack thereof. The overall objective is to gain an insider's view of the patient journey to achieve better outcomes and overcome barriers to therapeutic success. So how do pharma leaders access the insights they need? The solution lies in the elusive world of big data. But before jumping on the big data bandwagon, an understanding of what it entails is crucial.

What is big data anyway?
In the healthcare arena, big data is an umbrella term that encompasses volumes of data related to patient health and well-being, disease diagnoses and progress, tests performed, lab results, medications prescribed or administered, side effects and more. The data sets that comprise big data are large, complex and unruly. Think of the data sets as a pack of wild horses-they're difficult to corral because of their size, speed and utter indifference to playing by the rules. But bring in a well-regarded horse whisperer, and the horses line up and behave accordingly.

Big data aimed at improving the patient experience is gathered from many sources and growing by the nanosecond. Sources range from electronic medical records, market research and clinical trials to personal blogs, online communities and social media posts. The challenge is to define what big data means for certain healthcare stakeholders.In the race to make sense of big data, the challenge is to define what big data means for certain healthcare stakeholders so the right data is gathered and analyzed to gain the right insight.  To make it more manageable conceptually, big data can be divided into two buckets: structured data and unstructured, or qualitative, data.

Structured data.
Generally speaking, structured data is the easiest data type to capture and categorize in a database. Structured data comprises information entered into pre-defined fields inside an EMR, such as diagnosis codes and patient demographics, or pharmacy claims data, such as type and timing of medications filled.

Pharma manufacturers recognize an increasing need to understand the patient journey, treatment patterns and reasons for therapeutic drop-off. In fact, medication non-adherence is a major, costly issue, one that is thought to raise hospital readmission rates. Many factors can result in medication non-adherence or therapeutic drop-off, such as language barriers that obstruct understanding or transportation issues that prevent medication pick-up. Structured data, however, may not reveal the bigger picture such as why a patient didn't adhere to his medication; this critical information resides in unstructured fields, waiting to be unlocked.

Unstructured data.
Unstructured clinical data is cryptic, disorganized and often text-heavy, and can be collected from a variety of sources including:

  • Open-text fields in EMRs
  • Emails, images and videos
  • Patient support groups
  • Social-media conversations and personal blog posts

Like wild horses, unstructured data is practically begging to be tamed. It just requires the right approach. Once the patterns within are uncovered and understood, unstructured data has the potential to improve both patient outcomes and product performance.

When a physician enters an EMR progress note about why a patient is non-adherent or a medication is stopped, for example, the text cannot be manipulated to fit into neat little checkboxes and pull-down text fields. The text is based on a dialogue between patient and physician, and is often highly individualized. These data sets are trickier to unlock, but the payoff can be literally off the charts.

Discovering the value within.
Nowadays, there's a much larger emphasis on leveraging data to drive better patient outcomes. But even as patient-centered data has grown in importance and become more vital to the success of product performance, it's the manufacturer's ability to understand and harness data that will set it apart from others.

Pharma partners are working furiously to develop systems to identify patterns that improve the value and usability of unstructured data. Think of these industry partners as the data whisperers: they know how to tame the data by extracting relevant pieces from the crowded field of sources and break them down into digestible pieces.

To get a better handle on the real-world performance of their medications and the patients who take them, manufacturers are gathering insights from regular interactions between patient support services providers and patients, providers and payers. The most innovative companies are combining these insights with data collected from internal and external sources to create targeted strategies that elevate patient outcomes and improve product performance.

Data-mining tools such as Natural Language Processing enable deep analysis of text-based documents, and open the door to a deeper understanding of the patient experience. Pharma is working to extract meaning from a variety of unstructured data sources, including social media spaces where patients discuss diagnoses and medication side effects, ask questions and provide general support to one another. There's also been an uptick in integrating EMRs with pharmacy and medical claims, aiming to follow patients longitudinally and gather data along the way.

Self-reported patient data is rapidly emerging from patient-support groups, social media conversations, blog posts and mobile apps. These data sets are of great value to pharma as they might reveal a drug's effectiveness outside the comfort and familiarity of the clinical trial setting. To gather insight into the numerous factors affecting a drug's post-launch success, companies must crack open the real-time, real-world feedback reported by the patients themselves. Take a female patient with Crohn's disease, for example. Perhaps she tweeted about affordability challenges. Or maybe she opened a discussion on a disease-specific community forum like myCrohnsandColitisTeam about problems with medication side effects. These pieces of patient-reported information are crucial to filling the digital gap in pharma's patient-support puzzle.

The patient insights uncovered by mining unstructured data are priceless, including revealing cause and effect relationships. When Patient A takes Medication B, for example, then Side Effect C happens. By collecting enough of these data types, analytics can predict how the next patient might respond to a medication or when it will take effect.

It's not one piece of data that's valuable but rather, a blend of multiple data sources that provide a more complete picture of patient needs. With the right analytics tools at their disposal, manufacturers can uncover data patterns that correspond with meaningful conclusions, conclusions that would have otherwise been missed.


Sign in or register to leave a comment