Medical claims web app and ingestion pipeline

Problem

The client's undocumented claims codebase could create but not properly update claims, its validation logic could not be driven outside the browser, and the ingestion pipeline was hardwired to a local SQLite file while the production database was to be hosted remotely.

Solution

The team stood up the stack, traced and fixed the update-claim flow, located the real validation endpoint, and replaced a disabled-middleware workaround with a CSRF token endpoint whose token travels in the login header and cookie. In the Python pipeline it reworked encounter numbering, added a duplicate-claim guard, cached the claims table at start-up, specified a directory watcher to replace polling, and moved persistence onto SQLAlchemy behind a single database class so the target database is a connection-string change.

Key features

• Claim create, review, edit and state-change flow with CSV claim reports
• CSRF token endpoint enabling secure external claim submission
• Encounter numbering grouped by patient, service code and date
• Duplicate-claim rejection on ingest
• SQLAlchemy database abstraction with bulk inserts and rollback handling
• Directory-watcher design for incoming claim files
• Review of health-service-code rules including Mod10 checksum validation

Tech stack

• AWS
• Bitbucket
• Express
• MySQL
• Node.js
• PostgreSQL
• Python
• React
• SQLAlchemy
• SQLite
• csurf
• pandas
• watchdog