ASP AI Engine Coding Intelligence

The gap between a CPT and the right ICD specificity is where 2026 risk lives.

A surgical CPT tells you what was done. It does not tell you the laterality, the anatomic region, or the secondary diagnoses that risk adjustment pays on. ASP AI Engine closes that distance in layers, and it is honest about the one place a model still cannot.

SHORT ANSWER

ASP AI Engine does not guess an ICD code from a CPT. It reads the operative descriptor, gates it against an anatomic region check so the diagnosis cannot land in the wrong body area, then sweeps for the HCC-relevant secondary diagnoses the note documents. The limit that remains is descriptor coverage, and we tell you exactly where it is.

The three layers

From operative descriptor to defensible specificity

Each layer takes the output of the one above it and refuses to pass anything it cannot ground in the documentation. Peel them apart and the logic is boring on purpose.

Layer 01 01

Read the CPT long descriptor, not the code number

The 5-digit CPT is an index, not a meaning. The signal is in the AMA CPT 2026 long descriptor: the words that name the approach, the structure, and the extent of the procedure. ASP AI Engine anchors on that descriptor text and the operative note together, so "arthrodesis" is understood as a fusion, not matched on a number it never saw in the record.

AMA CPT 2026 long descriptors operative-note readback

Input28730 · arthrodesis, midtarsal or tarsometatarsal, multiple

→

Understood asmulti-joint fusion, midfoot, extent = multiple joints

Layer 02 02

Gate the diagnosis against an anatomic region check

This is the layer that stops the embarrassing errors. A deterministic code-family map ties every CPT to the body region it can legally describe, so a midfoot fusion can never be paired with a lumbar-spine diagnosis. When a model wants to attach a code outside the region the descriptor allows, the gate blocks it before it ever reaches a claim.

deterministic region map blocks wrong-region codes

Candidate ICDM48.06 lumbar stenosis ✕ region mismatch

→

Region-valid ICDM20.5x acquired deformity, foot · with laterality

Layer 03 03

Capture the secondary diagnoses risk adjustment pays on

The primary diagnosis rarely carries the risk weight. The documented comorbidities do. ASP AI Engine runs a secondary-diagnosis sweep that surfaces chart-supported conditions to the highest level of specificity the FY2026 guidelines require, and flags the ones that map to an HCC under the CMS-HCC V28 model that governs payment year 2026. Nothing is captured that the note does not document.

secondary-dx sweep CMS-HCC V28 flagging chart-documented only

In the note"long-standing DM, neuropathic foot, CKD stage 3b"

→

Captured + HCC-flaggedE11.42 · N18.32 · both HCC-relevant V28

One chart, three gates, a defensible line

The same operative note moves left to right. Each gate can only pass what the layer above it grounded in the record. The dashed return path is the honest part: when the descriptor library has no long descriptor to match, the code is held for a human, not invented.

The honest limit

Where descriptor coverage still caps a model

No amount of prompting fixes a descriptor a code book does not contain. On our own charts we measured exactly where the ceiling sits, because you should buy data, not adjectives. These are ASP AI Engine internal bench figures, not published rates.

1 / 175

Foot and ankle CPT codes that carry a long descriptor in our current code book

ASP AI Engine internal bench

Times "coalition" appears anywhere in the descriptor set we ship today

ASP AI Engine internal bench

97%

CPT accuracy across a 75-chart production run, where the descriptor existed to match

ASP AI Engine internal bench

75%

ICD accuracy on that same run, held down precisely by the missing descriptors

ASP AI Engine internal bench

The fix is not more rules. It is the AMA long-descriptor file, the one data purchase that lifts every layer above at once. Until it is loaded, ASP AI Engine holds the unmatched line for a human coder rather than guessing, which is the difference between an assist tool and a liability.

Why layer three pays for itself

The secondary diagnosis is the risk-adjusted dollar

A clean primary code keeps the claim from denying. The documented comorbidities, coded to FY2026 specificity and mapped to V28, are what the risk-adjusted contract actually reimburses. ASP AI Engine surfaces them from the note the surgeon already wrote.

E11.42 DM w/ diabetic polyneuropathy , HCC-relevant under V28, missed when only the procedure is coded

N18.32 CKD stage 3b , the FY2026 split of stage 3 rewards the specific sub-stage, not "CKD, stage 3"

I50.32 Chronic diastolic heart failure , specificity of the failure type is the HCC trigger

RULE ICD-013 · SECONDARY-DX

"Code every chart-documented comorbidity to its highest specificity. Flag the ones that carry a V28 HCC. Capture nothing the note does not support."

Payment year 2026 runs on 100% of the CMS-HCC V28 model. Under-specified secondaries do not just risk a denial, they quietly leave the risk score under water.

Grounded in the real 2026 rulebook

Every layer maps to a named guideline

AMA CPT 2026 Code Set

Long descriptors are the source text ASP AI Engine reads in Layer 1, not the numeric index.

EFFECTIVE JAN 1, 2026

ICD-10-CM FY2026 Update

CDC/NCHS and CMS annual code and guideline revision that sets the specificity ASP AI Engine codes to.

EFFECTIVE OCT 1, 2025

ICD-10-CM Official Guidelines FY2026

CMS documentation-specificity guidance: code to the highest level of detail the record supports.

SECTION I.B · SPECIFICITY

CMS-HCC Risk Adjustment Model V28

The model that governs payment year 2026 at 100%, and the map behind Layer 3's HCC flagging.

PY2026 · 100% V28

CMS CY2026 Physician Fee Schedule Final Rule

The payment context ASP AI Engine-coded professional claims are built to survive.

PFS · CY2026

CMS CY2026 OPPS / ASC Final Rule

Facility and ASC context for the surgical lines the descriptor layers resolve.

OPPS / ASC · CY2026

Bring us a stack of your own operative notes.

We will run them through the three layers, show you exactly where the specificity holds, where the descriptor gap holds a line for a coder, and what the secondary-diagnosis sweep would have recovered. No slideware, your charts.

Book a ASP AI Engine descriptor review →

ASP-RCM Solutions · Frisco, Texas

The gap between a CPT and the right ICD specificity is where 2026 risk lives.

From operative descriptor to defensible specificity

Read the CPT long descriptor, not the code number

Gate the diagnosis against an anatomic region check

Capture the secondary diagnoses risk adjustment pays on

One chart, three gates, a defensible line

Where descriptor coverage still caps a model

The secondary diagnosis is the risk-adjusted dollar

Every layer maps to a named guideline

Bring us a stack of your own operative notes.

More on this topic

The descriptor said lumbar. The code said cervical. ASP AI Engine caught it.

The FY2026 ICD-10 code calendar an AI coding engine must honor

The estimate is spoken, printed, and compliant before the patient leaves the desk.