Thursday, March 3, 2016

Using BPM to Assure Information Quality

I know, there is no such thing as 100% quality, but we need to challenge ourselves to get there. To catch potential data quality issues, you’ll need to create a set of validations, added to a process that will identify and fix them.  This set of validations can be automated within a workflow, identifying rules and actions to perform given certain conditions. Finding and fixing issues is an ongoing task: it requires a balance of vigilance and curiosity, as well as caring. Usually issues come about because there is spotty accountability somewhere in the flow of information from the source to the downstream systems.

Goal

Close any data integrity gaps by applying validation checks and fixes.

How does this happen?

It can happen very gradually and surreptitiously. Within a company, unless there’s a strong information quality department, there will inevitably be data inconsistencies because each department has different priorities and validation requirements. All it takes is one form in one application with lax data input requirements. It could also be a lack of validation check during data input. Lastly, when software is upgraded or data is merged from one system to another, we assume wrongly that one source data is fully vetted.

Examples of how this happens

Downtime

Let’s say your system has an account lookup feature, but that feature is down, so you have to enter in the account information manually. The feature is fixed in a few hours, but by then you’ve entered 100 accounts. Does this data get validated later? If it doesn’t, a downstream application could have quality issues.

Patient information merges and updates

Lets’ say you work at a hospital. There’s a patient referral with the same medical record name (MRN) as an existing patient with the same birth date. The referral is entered in at the existing patient. This error is caught later and the patient information is fixed, but did the patient already get treated?

Towards Quality with Process Automation

By inserting workflows into the process, specific types of data inconsistencies can be identified, investigated and resolved. Below are some general design components for building a quality validation workflow:
·         Figure out how to funnel all data/content through the validation workflow. Using by doc type or input sources, the information can be collected and filtered as appropriate.

·         Create the rules to route issues into buckets. Here are some typical queues:
o   Routing: this queue has validation checks which compare metadata values against source of record values.
o   Issues queues: these correspond to the common issues that get identified.
o   Routing issues: this queue holds any doc that doesn’t match the issues queues.


·         These buckets can be evaluated during their initial manual fixes for potential automated solution, and identifying upstream, root cause, data issues.

No comments: