Data visualization tools make it easier to glean intelligence from volumes of information to improve health programs, clinical healthcare delivery, and public health policy. But they have failed to incorporate the science of human visual perception into the technology, resulting in tools that deliver great “eye candy” but poor human comprehension of the data.
For example, today’s visualizations tend to utilize color combinations that contribute to user confusion, or to add distracting animations or graphics that reduce clarity while impeding the discovery of insights in the data being presented.
These tools also tend to visualize what we already know about the data, rather than the unknown. The ideal is software that lets researchers find things that they don’t know, but should. Helping people find outliers, expose hidden trends or clusters, and dive deep into fast changing data sets is where visualization provides real value.
As healthcare meets the “Internet of things,” the ability to discover anomalies in real-time streaming data from thousands of medical devices, sensors and monitors will be of huge value. Or as Electronic Health Record databases become ubiquitous, for example, effective visualization of the data could unveil previously unseen adverse treatment patterns.
A new generation of visual data discovery technologies are emerging that incorporate human psycho-visual principles to produce visualizations that are easily understood. What’s also new is that these technologies can now visualize data “in flight,” before it gets warehoused, to help users discover unexpected patterns, outliers and relationships in real-time data.
Pre-attentive processing: visualizing the way people see
New healthcare visualization interfaces are incorporating the principle of pre-attentive processing, which are the visual properties that people process almost subconsciously within milliseconds without the need for focused attention. It utilizes the innate pattern-sensing capabilities that everyone possesses.
Different shapes in 2D or 3D, colors differing in hue, saturation and brightness, graphic patterns or placement in a two-dimensional space are all things that humans perceive automatically and unconsciously. For example, the automated and effortless perception of information is what visualization tools such as treemaps make use of. Treemaps are great for analytical problems that require people to find patterns, clusters, gaps, or outliers in very large data sets.
The University of Maryland’s Human-Computer Interaction Lab, for instance, developed a Gene Ontology analytic application using treemaps to gain greater insight in genomic processes and their biological functions. The application gives users a 19-level hierarchical acyclic graph that catalogs approximately 14,000 genes according to their biological functions. Users can quickly view and spot patterns in the gene ontology while showing expression level data with color and size coding. The application is helping to accelerate research with a more visually intuitive way to explore gene ontology for understanding microarray gene chip experimental results.
As volumes of data for things like patient care, insurance claims and medications grow exponentially, healthcare facilities need real-time analytical capabilities to improve the quality and effectiveness of their services. The ability to analyze real time data “in flight” is becoming increasingly important in healthcare informatics for applications like epidemic monitoring, geographic epidemiology and gene expression microarray analysis. For example, hospitals are increasingly using data visualization to monitor end-to-end care delivery across a variety of settings. Applications like clinical decision support systems connect real-time clinical observations like physiological data, such as that streaming from a bedside monitoring device in a hospital room, with health knowledge to improve patient outcomes.
Real-time visualization can also be put to work for operational purposes. It can help healthcare administrators identify the drivers of critical healthcare variables impacting cost, track patient care compliance metrics, understand staff utilization, or detect fraudulent insurance claims.
Combining real-time data “in flight” with pre-attentive processing processes can dramatically reshape how healthcare informatics can improve clinical, operational and medical research decisions. Displaying healthcare data in visual formats requires an in depth understanding of the human psycho-visual system to produce visualizations that are easily understood.
Ben Plummer is the CMO for Datawatch Corporation.