Overview
Healthcare labs were left holding enormous COVID-era billing backlogs when insurers began denying claims en masse. At Lifescan Health, I led the engineering effort that turned what the RCM team estimated as a six-month manual project into a weeks-long automated run — and protected $20 million in potential revenue in the process.
The Problem
Insurers issued retroactive denials on tens of thousands of COVID test claims. Each appeal required:
- Pulling the original claim data from multiple billing platforms
- Cross-referencing patient records and test results
- Generating a compliant appeal document
- Submitting it through the insurer’s portal or clearinghouse
The RCM department had no automated tooling for this. At the rate staff could work manually, clearing the 66,000-claim backlog would take six months minimum — by which time many appeals would have aged out and the revenue would be unrecoverable.
What I Built
- A Python-based extraction layer that pulled claim data from external billing platforms and Databricks, normalizing across inconsistent schemas from multiple payers
- A PySpark pipeline to validate, enrich, and deduplicate appeal records at scale
- A document generation module that produced payer-compliant appeal packages (CMS-1500 format and custom payer formats)
- A submission engine handling batch uploads and portal automation for major payers
- A tracking dashboard so the RCM team could monitor appeal status, approval rates, and remaining backlog in real time
Results
- 66,000 appeals processed in under a month
- $20M in potential revenue protected from statute-of-limitations expiry
- RCM team staffing need reduced by 50% during and after the campaign
- Processing time for individual appeals dropped from 5 days → 2 hours
Stack
Python · PySpark · Databricks · SQL · CMS-1500 · REST APIs · Healthcare RCM