Behind the Dashboard: The Process of Data Discovery
“My data is good, I know it”- Are you sure?
Conceptually everyone wants to know things, people want to know more about things that interest them or apply to them. How many times have you asked the question Why? If the pandemic taught anyone anything it was that wanting to know more isn’t necessarily a bad thing and is a key piece of decision making- making decisions for the right reasons, based on the best available information. However, information and statistics are daunting. Armchair statisticians believe that the act of putting together information and gaining meaningful insights are a product of simply having numbers at your disposal and some basic math principles that you mastered along the way. The reality is that moving from data concept to operations is so much more.
For people who work in statistics day in and day out, the nuance is the norm, and getting to the actual, right answer is a much more complex process. This doesn’t stop consumers or executives from wanting to know more. Confidence in an excellent and skilled report writer is no substitute for bad data entry. It is only when you take a deep dive into the information that the true quality of the data behind the report writing is revealed.
What does data discovery do that I am not already doing?
In healthcare the process of data discovery is a key piece in revealing workflow issues. Many healthcare organizations believe they understand issues with workflows based on known departmental eb and flow, as well as knowledge of specific employee habits and personnel turnover. It is not often that these organizations are taking the time to look at the data that will give them key insights on cost, quality, and access as a conduit to workflow issues.
What am I looking for/How do I apply this?
Outliers, automated in a visualization provide a keyway to remain compliant with regulatory issues and proactively identify and correct issues of non-compliance. Taking the time to examine the data, closely and careful yields better results in the end. Key features to examine when performing data discovery include:
Illogical dates/times
Example: Check In occurs after a patient is roomed
Missing data values or date/time stamps, when necessary, precursor events exist
Example: A patient has an ‘in room time’ and no ‘out of room time’
Identification of duplicate entries where one value in one cell is different
Example: A singular case appears or is counted twice due to different providers appearing in the provider field- this could be due to a co-signature on a note, but only one performing provider should appear for the encounter.
Trivial or Obsolete Data/Fields
Example: Extra columns were added to a data set that were never used in the past and are not applicable and have no future use to the dashboard build OR the column is always blank/null
Data discovery can be done by either a skilled data steward or through the process of enabling AI programs. Regardless of the path that is chosen, it remains a vital element of extracting the best information for decision making from a dashboard.
We understand you have core projects on your hands, let us help take the data discovery process off yours. Talk to someone today.