Five Ways to Turn Big Data into Relevant Data

Here is an expose from Dr. David Lee Scher. You can find the original entry here.

‘Big Data’ may be defined as data that is so large or complex that it cannot be processed by the usual applications. Current common uses of Big Data in healthcare include the collection of genomic data, insurance claims data, medication prescription data, mandated hospital reporting data (admissions, diagnoses, readmissions, other).  The problem with amassing such data is that much of it is trapped in silos across the healthcare continuum and not available (even for studies obtaining consent and protect privacy) for analysis.  Robust analytical tools are not a part of most IT platforms utilized at points of care.  Organizations which can profoundly influence treatment guidelines incorporated into evidence-based medicine, healthcare policy and patient advocacy are not availing themselves of relevant data. I will discuss some ways in which Big Data can be transformed into little or relevant data.

  1. Longitudinal follow-up. As a clinician I appreciate the enduring contributions of  Sir William Osler, considered to be the father of modern medicine.  One fundamental concept he championed was the study of the natural history of a disease and the longitudinal follow-up of treated patients. The recording and storage of data longitudinally is called a registry.   Registries can serve as potential sources of relevant data. If data set query is large enough, the right questions are asked of the mined data and adequate and creative analytics are applied, it’s not difficult to transform Big Data from registries into relevant data.
  2. Obtaining data across disparate digital platforms. Following a patient longitudinally often necessitates obtaining data points acquired from providers who utilize different EHR platforms. This presents problems for clinicians, epidemiologists, and patients and their caregivers. This cross-logistical data is critical for registries. A patient participating in a clinical trial or registry unexpectedly seen in an emergency setting away from home or even by a specialist nearby with an EHR different than that of the patient’s registry clinician/investigator presents both logistical and technical challenges of incorporation of pertinent data into the patient’s registry profile. A platform which can ‘scrape’ data of a specific patient across EHR silos assures the collection of all relevant data.
  3. Customizable Interfaces and analytics. Not all clinicians desire to view all or the same data collected on patients involved in a registry or population health IT tool.  They might want to view the same data in a different visual context. This type of customization is a welcomed reprieve from the confines of traditional EHRs and registry tools. Customization of this type transforms Big Data into relevant data for clinicians. Looking at data in different ways can stimulate the viewer to approach a patient, group of patients or a treatment plan in a different manner.
  4. Applying best practice guidelines and evidence-based medicine results. A data collection platform which furnishes evidence-based practice guidelines for treatment of a specific disease and provides the ability to compare real-time individual or population health group data to outcomes of optimally treated patients can be very powerful.  It thus provides a direct connection of big data to relevant data, allowing for mid-treatment corrections during care to meet best practice outcome metrics. In addition it can serve as a mechanism to correct the significant geographic variations in healthcare.
  5. Incorporation of patient-reported outcomes measures. Patient reported outcomes measures (PROMS) have been described with relevance to healthcare economics for over eight years. PROM is an integral part of the value-based payment system which will dominate how healthcare will be paid for.  These outcomes are determined from the collection of specific clinical data sets. Population health management IT tools must incorporate these measures to meet payment requirements. This goal can be accomplished in a piecemeal fashion with multiple IT vendors or ideally with one.

Much has been said about the power and prospects of Big Data. But this data, like the food we eat needs to be distributed, filtered, and relevant parts applied to good use. Healthcare may lag behind in the use of digital and mobile technologies, but the amount of Big Data garnered by all its stakeholders is incredible. The future of improvements in healthcare (whether they be in cost savings, clinical outcomes, efficiency or other areas) lies in the way in which Big Data can be made relevant. It’s like the old adage, “All politics is local.”

As a disclosure I am an advisor to Pulse Infoframe which has capabilities to accomplish all of the above.