Decision Traceability and Data Posture

A pre-determination that can't explain itself isn't much use when a payer, a provider, or a patient asks why. Accord Dental is built so that every determination is reconstructable: the decision, the data it rested on, and the rules it applied are all recorded and traceable. This document describes how that works — and is precise about what exists today versus what's on the roadmap.

Sourced data: nothing is invented

The foundational principle of Accord's data posture is simple: every stored value is traceable to a source. No data is fabricated to fill a gap. Each figure in a cost estimate, each coverage conclusion, each extracted clinical fact traces back to the document it came from or the rule that produced it. This "no invented data" discipline is what makes the rest of traceability meaningful — you can follow any output back to a real origin.

The decision trace

When Accord assembles a determination, it builds an explicit decision trace as it reasons. Rather than emitting only a final answer, the engine records the path: which eligibility facts applied, which coverage rules fired, which conditions were triggered. The result is that a determination is not a black box — the reasoning that produced it is captured alongside the outcome, so a reviewer can see how the engine arrived at the number, not just what the number was.

Readiness signals

Accord computes a set of readiness flags for each case — structured checks on whether a pre-determination is complete and submission-ready. These flags make the state of a case explicit: what's satisfied, what's missing, and whether the case is ready to submit. Instead of a binary "done or not," the readiness signals show the specific gaps, which is both operationally useful and part of the record of how a case was evaluated.

The evidence graph

Traceability extends to the clinical evidence through a graph that links conclusions to proof. Evidence nodes represent individual extracted facts; evidence edges capture how they relate — which evidence confirms, contradicts, or corroborates a conclusion. This is what lets a determination cite its basis: a conclusion isn't just asserted, it's connected to the specific evidence supporting (or challenging) it. That's the backbone of the product's promise that every determination is explained, cited, and defensible.

Pipeline telemetry vs. audit trail

An honest distinction matters here. Accord records pipeline telemetry — events describing how a case moved through processing. This is operational instrumentation: useful for understanding and debugging the pipeline. It is not the same thing as a formal, tamper-evident audit trail of every access and change, and it should not be described as one. A production-grade access-and-change audit log is a distinct capability; where it isn't yet in place, it belongs on the roadmap rather than in a claim.

What reconstructability gives you

Taken together — sourced data, an explicit decision trace, readiness signals, and the evidence graph — these make a determination reconstructable: you can rebuild why Accord reached a given result from its recorded inputs and reasoning. For a clinical-financial decision that a payer might question or a patient might dispute, that reconstructability is the difference between "the system said so" and "here is exactly why, traced to the rule and the evidence."

Roadmap, stated plainly

Some capabilities associated with a mature data posture — a formal append-only audit trail, third-party security attestation such as SOC 2, and high-availability/point-in-time recovery configurations — are areas of ongoing work rather than current guarantees. This document describes the traceability that exists today and names the rest as roadmap, because in a healthcare context, precision about what is and isn't in place is itself part of good posture.