One segment of clients for a former employer included workers compensation insurers. For one particular client, the company aggregated and formatted specific data, and submitted it to an overseeing NGO-like entity on a quarterly basis on behalf of said client. This data (not covered under HIPAA requirements) is used for policy advancement and as some part of meeting requirements for certification as a responsible partner in the industry.
This task took multiple days every quarter for one Account Manager, and longer at year-end. It also required putting the data into an archaic machine-readable whitespace-delimited format for submission. This was something the company only did for this client, but automating it enabled offering it to others and the likelihood of gaining preferred vendor status with these as well.
I wrote a python script that took in two files – a CSV of client data from the company’s CRM-like system, and a very large (client-supplied) text file of claims data. The script then ran comparisons matching claims, comparing procedures, claimants, claimant metadata, and other data, then output the required properly formatted and delimited data to an ACSII file. It also included robust error and anomaly logging for checks and verification.
Stretching of my skillset, the script had to accommodate and verify against claimants with multiple claims and multiple (same) procedures, and be valid whether it be the same or a different claim. It also did some date and data formatting checks and adjustments.
Sixty-two pieces of data were declared in the script (including column delineations). Fifteen data points were output to the final ASCII file per line, with up to 3 procedures per line. If the number of procedures exceeded this, a new line was added.
I’m a bit proud to say the output file was validated and accepted on the first submission. This project was a good example of how my expectations of like a 12-24 line script ended up a robust and well commented 380 lines.