Healthcare Analytics Through a Clinical Hub

RHIA,Global Healthcare Industry Ambassador, IBM Information Management

We frequently hear that we live in the Information Age, and healthcare is no exception. I’m often involved in discussions on what healthcare analytics can do to improve care and decrease costs.

Recently, Dr Scott Schumacher, Chief Scientist and Distinguished Engineer for IBM InfoSphere MDM, shared his perspective on MDM and analytics during a webinar, Applying Operational Analytics through a Clinical Hub. He articulated a few key points that help us understand the broader reach of MDM, and how MDM underpins many forms of analytics. Here’s a very simplistic summary of how the Clinical Hub works:

  • Uses HL7 messages (current and historical) and unstructured text (discharge summaries, consultation, op notes, etc) analytics to build the longitudinal history of a patient. These messages and content are integrated with healthcare specific natural language processing, and text analytics applied based on healthcare vocabularies including SNOMED, LOINC, RxNorm and others. 
  • De-identifies data through a separate instance of the IBM Initiate MDM product and creates a limited data set to support privacy and research requirements. Only the MDM instance holds the patient-identifiable data.
  • Applies a patient similarity service and framework against a longitudinal patient record using IBM Research assets. This service identifies “like” patients based upon the clinical data parameters that the user has input and multiple algorithms are applied to determine “closeness” of subjects to the clinical data parameters. 
  • Automatic user interface to support cohort selection

Clearly, this is very complex (compared to my simplistic overview) since multiple algorithms and research technologies are involved. The benefits of this approach are:

  • Built on the IBM Initiate MDM product, which has broad support in the healthcare industry. 
  • Quick time to value, as the time expended by information technology resources and researchers is markedly decreased in the cohort selection phase. This is a classic example of “time is money.”
  • Easy user interface, which contributes to faster cohort selection and shorter research timeframes
  • Better and broader results from the advanced analytics approach, which also decreases research costs
  • Research timeframes can be decreased, which means products can be tested and brought to market faster. 

While Dr Scott’s primary example for the webinar was cohort selection for research projects, this same approach can be used for quality measures, best practice analysis, and other types of research.

Watch the webinar replay: