Automating Prescription Approval: Rules Engine Best Practices

Manual review of every prescription that comes through your online store is not scalable. As your DME business grows, you need an automated rules engine that can approve or reject prescriptions based on configurable criteria — while still routing edge cases to human reviewers. Here is how to set up and optimize your rules.

What Is a Rules Engine?

A prescription verification rules engine is a system that applies predefined criteria to each prescription submission and determines the outcome: approve, reject, or flag for manual review. Think of it as a decision tree that processes prescriptions in seconds without human intervention.

A well-configured rules engine handles 80-90% of prescriptions automatically, leaving only ambiguous or high-risk cases for your team to review manually.

Core Rules Categories

1. Document Quality Rules

These rules check whether the uploaded document is usable:

• File type — accept JPEG, PNG, PDF; reject unsupported formats
• Image resolution — reject blurry or extremely low-resolution images where text cannot be reliably extracted
• Document classification — reject uploads that are not prescriptions (insurance cards, receipts, selfies)
• Readability score — AI assigns a readability score; documents below the threshold are rejected with a request to re-upload a clearer copy

2. Prescriber Verification Rules

These rules validate the healthcare provider who wrote the prescription:

• NPI present — require an NPI number on the prescription (some merchants allow manual NPI lookup as a fallback)
• NPI active — verify the NPI is active in the NPPES registry
• NPI name match — the name on the prescription must match the NPPES record (with configurable fuzzy matching tolerance)
• Provider specialty — optionally require the prescriber's specialty to be relevant to the prescribed device
• State license — verify the prescriber is licensed in the patient's state (for telehealth prescriptions)

3. Prescription Content Rules

These rules check the prescription details:

• Date validity — prescription date must be within the configured validity period (e.g., 12 months)
• Patient name match — the patient name on the prescription must match the customer's order information
• Product match — the prescribed device or device category must match what the customer is purchasing
• Signature present — a prescriber signature must be detected on the document
• Required fields — all mandatory fields must be present and readable

4. Fraud Detection Rules

These rules flag potentially fraudulent submissions:

• Duplicate detection — flag prescriptions that match previously submitted documents
• Manipulation detection — flag documents showing signs of digital editing
• Velocity checks — flag when the same NPI appears on an unusually high number of prescriptions in a short period
• Known bad actors — check against lists of deactivated NPIs or previously flagged providers

Configuring Confidence Thresholds

RxCompliant's AI assigns a confidence score (0-100%) to each verification. Your rules engine uses this score to make decisions:

• Auto-approve threshold — prescriptions scoring above this level are automatically approved. Recommended starting point: 90%.
• Manual review range — prescriptions scoring between your reject and approve thresholds go to your review queue. Typical range: 60-89%.
• Auto-reject threshold — prescriptions scoring below this level are automatically rejected with a reason. Recommended starting point: 60%.

Start with conservative thresholds and adjust based on your review data. If you find that manually reviewed prescriptions in the 85-89% range are almost always approved, raise your auto-approve threshold down to 85%.

Setting Up Rules in RxCompliant

In your RxCompliant dashboard:

1. Navigate to Settings > Verification Rules
2. Enable or disable individual rules based on your requirements
3. Set confidence thresholds for auto-approve and auto-reject
4. Configure notification preferences for manual review items
5. Set up escalation rules (e.g., if a prescription is not reviewed within 24 hours, notify a supervisor)

Monitoring and Optimization

After configuring your rules, monitor their performance:

• Approval rate — what percentage of prescriptions are auto-approved? If it is too low, your rules may be too strict.
• Rejection rate — are legitimate prescriptions being incorrectly rejected? Review rejected submissions periodically.
• Manual review volume — if too many prescriptions require manual review, tighten your auto-approve or auto-reject thresholds.
• Processing time — track end-to-end verification time. Auto-processed prescriptions should complete in seconds; manual review items depend on your team's response time.
• False positives — track how often manually reviewed prescriptions are approved vs. rejected. High manual approval rates suggest your auto-approve threshold is too conservative.

Best Practices

1. Start strict, loosen gradually — it is better to manually review more prescriptions initially than to auto-approve fraudulent ones
2. Review your rules monthly — as your AI processes more prescriptions, patterns emerge that suggest rule adjustments
3. Document your rule configuration — for compliance audits, you need to explain why your rules are set the way they are
4. Test rule changes — before modifying rules, run them against your historical data to see how the changes would have affected past decisions
5. Keep a human in the loop — automated rules should handle the clear-cut cases, but human judgment is essential for edge cases

Ready to automate your prescription verification? Create a free RxCompliant account and start configuring your rules engine today.