Graphical displays of data have been with us since the 1700s, according to visualization guru Edward Tufte. People are visual creatures, so if we want to really understand a set of data, it’s of great value to be able to see points of comparison, say in a timeline, graph or diagram. This is even more critical as the volume of data gets bigger and the nature of the data becomes more varied.
In healthcare, data is a powerful resource to help improve the overall health of populations, drive down the per capita cost of delivering healthcare, and improve the citizen experience—the three objectives of the Institute for Healthcare Improvement’s TripleAim.
Recently, CGI and a European hospital district collaborated on a clinical data study to visualize and understand patterns in healthcare visits. The focus was on psychiatric, trauma, injury and circulatory visits with strong cause-and-effect links to patient behavior and controllable lifestyle factors. We looked for correlations and trends that would spur ideas for improving health outcomes, while minimizing the cost of care. iam pleased to share some of the visualizations from the project.
These graphs show two groups of patients by age and gender. On the left are patients with visits to the psychiatric specialty, and on the right are patients with trauma or injury visits. The population characteristics that can be seen readily from these visualizations include:
- Psychiatric visits were much higher for children and teens, with the highest rates for males occurring at younger ages than for females.
- Males were more likely to receive care for traumas and injuries than females until the age of 60, but after age 70, females are more likely to receive such care.
Taking a deeper look at the issues of adults to determine possible actions, the team then examined data for psychiatric visits when combined with data for alcohol-related visits.
These graphs show adult patients by age and gender, comparing groups with alcohol-related visits without psychiatric visits to those having both alcohol and psychiatric visits timed closely together. Here, the patterns indicate:
- Patients with visits to the psychiatric specialty and alcohol-related issues requiring care were most often young adults.
- Alcohol-related care for patients not having psychiatric visits was concentrated in middle-aged males.
Since subgroups of patients with high costs of care were identified, an important goal is to find ways to reduce costs, while maintaining or improving patient health. From these insights, possible actions could include:
- Preventive measures aimed at the root causes of the combined issues in young men and women, which could have great potential to improve outcomes and keep costs under control.
- Programs targeting the root causes of alcohol abuse alone, which could have a positive impact on health outcomes for men in particular.
Data visualization is a powerful technique, and the ability to form hypotheses and produce simple, informative graphics is more important than the technical wizardry. The most successful pictures are those that lead to new insights and actions.
On a related note, when there is so much data, and an intense and winding journey to get to the compelling insights, we must remember that less is more when presenting the results. But in this case of course, less (i.e., presenting the actionable information) means more work to be done!
About this author
Data Scientist, CGI
Elina Jeskanen works as a data scientist producing data visualizations, advanced analytics and process modeling solutions across industries. She uses data science to raise awareness, bring new insights, introduce future possibilities and enable data-driven decision-making with true business value.