Inside the Rule Engine: How Wound Care Coding Boundaries Get Enforced Consistently
product
Area thresholds, depth tiers, add-on units, and modifier logic — how a deterministic rule engine keeps wound care coding uniform across every visit.
AI drafts codes, but a rule engine is what keeps those drafts inside the lines. This is how the deterministic layer under the coding agent actually works and why it matters for consistency.
The problem with pure LLM coding
Large models can hallucinate a plausible CPT code that violates a boundary condition — billing 97597 when the treated surface area is under the threshold, or picking a debridement depth code without evidence of the tissue removed. Consistency across clinicians and days is not guaranteed.
What the rule engine enforces
- Surface area thresholds for 97597 base vs. 97598 add-on units, rounded per CMS convention.
- Depth hierarchy for 11042–11047 — subcutaneous, muscle/fascia, bone — with each requiring an explicit tissue statement in the note.
- Add-on unit math — additional 20 sq cm increments computed from the imaged measurement, not from clinician estimate.
- Modifier logic — modifier 25 attached only when the E&M meets the significant-and-separate test; modifier 59 for distinct procedural service where applicable.
- Bundling checks against NCCI edits before the line ever reaches the superbill.
- LCD alignment for skin substitutes, including the 2026 CMS rule thresholds.
How boundary cases are tested
Every rule ships with a test suite of edge scenarios — a wound measured at 19.9 sq cm vs. 20.1 sq cm, a debridement note that mentions bone without confirming exposure, an E&M that is really a pre-procedure assessment. These tests run on every model update so behavior does not drift.
Where determinism meets AI
The LLM proposes candidate codes with source spans. The rule engine accepts, rejects, or downgrades each one. The clinician sees the final list with the reason a proposed code was blocked — turning coding into a reviewable decision instead of a black box.
The result
Billable lines that pass into WISER claims and compliance are already reconciled with CMS thresholds, so denials for miscoded units or missing modifiers drop sharply.
See the rule engine on your workflow with WoundScribe AI.