Visualization of Legacy Healthcare Content Using Tableau
By George Florentine, VP, Technology
At the time of this writing (spring 2020), we are all seeing a lot of information about COVID-19 data, much of which has been presented using Tableau. As we look at ways to contribute to that conversation, we have developed and recently released a Tableau web connector that provides access to data stored in our Digital Hub for Healthcare product.
Let’s look at a few examples of the power that a platform that combines Tableau visualization with Digital Hub curated content can provide to Health Information Management (HIM) professionals.
We’ll start our visualization example by building a data model in Tableau, based on Digital Hub content. The Digital Hub content is a lightweight data model that has enough structure to be useful but isn’t overly complicated. Part of the art of visualization is building the appropriate model to allow the consumers of the model the opportunity to gain insight without being overwhelmed with complexity.
Image 1: Digital Hub Tableau Logical Data Model
Once our HIM user, Holly, has this model defined, she can start to build reports that can be used to support her care system’s COVID-19 response team. Each of these reports can be created in a few minutes and can be shared by posting to a web portal or perhaps used to create a PowerPoint presentation for use in education to colleagues within the care systems, patient advocacy groups or other providers in other care systems. One of her initiatives is to evaluate historical data to see what orders were fulfilled, to understand where there might be capacity problems as her care system has transitioned to a much greater emphasis on telehealth.
Image 2: Historical Order Types
Not surprisingly, complete blood count (CBC), comprehensive metabolic panel (CMP), and vital signs are at the top of the list. But two other high frequency occurrence orders stand out – Neuro Checks and Aerosol Treatments.
With a greater emphasis on telehealth and telemedicine, how will her team execute Neuro Checks effectively? Perhaps the team will procure and distribute wearable devices for monitoring activity, heart rate, etc., or proactively schedule more virtual check-ins to provide continuity of care while elective admissions are being restricted.
Another worrisome data point is the high use of aerosol treatments. This may indicate that the patient population is at higher risk for COVID-19 because of existing comorbidities such as asthma, COPD, lung disease, etc. With this visualization of historical data, Holly may recommend an increase in capacity in the pulmonary lab in preparation for a surge in COVID-19 related pulmonary issues.
In many provider systems, we see that a relatively low number of providers generate a large percentage of orders. This may be because they’re well known and have a large patient population, or it may because they’re in specialty areas that require frequent orders to manage acute or chronic disease areas.
To help her team in the transition to a more telehealth-centered model, Holly examines legacy orders by providers:
Image 3: Medical Orders by Providers
We noticed that the top 5 providers are responsible for a disproportionate number of orders. We might then decide to look at the data a little differently and show the number of orders grouped by order type. The understanding of this will allow us to focus on the high frequency-of-order providers and understand what kind of medical orders they’re requesting.
Image 4: Order by Provider and Category
In the report above, we can hover over a column and see each different order type. If the results were too large, we could use the filters on the right to restrict to a subset of providers and order types.
Now that we have a sense of some of the trends in our data, let’s put together a dashboard we might use to socialize our discoveries with our leadership team.
Image 5: Dashboard of Orders and Providers
Switching focus a bit, Holly receives a request from her logistics coordinator, who is looking to gain insights from legacy data on the frequency of visits to each location by a patient, to prepare appropriate disposables for COVID-19 testing – masks, thermometers, nasal swabs, test kits, etc.
Image 6: patient Visits by Facility
Holly might then be requested to look at patient visits to facilities for a set of high-risk patients:
Image 7: Facility Visits by Patient
If we wish to see a simpler, high-level view of patient visits to facilities over time, we can use the same data to represent a simple visualization:
Image 8: Patient Visits by Location, Facility, and Admission Date
Tableau Visualization for Digital Hub has the power and flexibility to enable Holly to create ad-hoc reports in response to business needs that may change over time. Holly can modify the Tableau data model based on the type of legacy content stored in Digital Hub and can use the cross-data source relation features in Tableau to join legacy data with current EHR data, data from regional HIE, and publicly available research data, etc.
By doing this type of analysis across legacy, EHR, and research datasets, Holly and her care team are ready to provide excellent care to their patients.
As we’ve learned in this current health care crisis, execution in unusual and stressful situations is predicated on doing crisis management before the crisis hits. This type of data visualization can be a vital component in preparing a care team to function when health emergencies occur, regardless of the cause.