Overjet for Insurance: How Payers Are Using Dental AI to Deny Your Claims

The Other Side of the Overjet Story

Overjet has built a strong reputation in the dental AI space. Walk into any dental conference and you'll hear practices talk about how Overjet's radiographic AI surfaces caries and bone loss findings that inform treatment planning. Case acceptance goes up. Disease gets caught earlier. The clinical story is compelling.

But Overjet isn't only a practice tool. Payers are using Overjet on the other side of the transaction — specifically to analyze the X-rays you submit with claims and identify inconsistencies between what the radiograph shows and what you documented clinically. When the AI finds a discrepancy, the claim gets flagged, routed for additional review, or denied outright.

Delta Dental, the country's largest dental insurance carrier system, has been a prominent Overjet partner for the payer-side application. Other regional and national carriers have built or licensed similar AI adjudication capabilities. The technology is advancing rapidly, and practices that don't understand how this scoring works are flying blind into an environment where their own X-rays are being used to evaluate the validity of their claims.

This article explains the mechanism, what it means for your claims, how to adapt your documentation practices, and what to do when AI-assisted denial hits a claim you believe is valid.

---

How Payer-Side Dental AI Adjudication Works

The Basic Mechanism

When you submit a claim with radiographic attachments — which is required for most restorative, periodontal, and many diagnostic claims — those X-ray images are increasingly being processed by AI analysis before or alongside human adjudication.

The AI does several things:

Radiographic finding identification: The AI scores the X-ray for the presence of key findings — interproximal caries by depth (enamel vs. dentin vs. pulpal approximation), calculus deposits, crestal bone loss, furcation involvement, root morphology, and restoration integrity. It produces a probability score for each finding.

Documentation consistency check: The AI compares its radiographic findings against the clinical documentation in the claim. If you're billing for a three-surface composite on tooth 14 but the AI scores only a shallow interproximal shadow — not a clear dentin-level lesion — the claim gets flagged. If you're billing for four quads of scaling and root planing but the radiograph shows minimal calculus and the probe depths in your documentation are primarily 3-4mm, that inconsistency becomes a denial trigger.

Historical comparison: Some payer AI implementations compare current X-rays against the patient's prior radiographs in the payer's records. If bone loss has remained stable for 3 years and a new periodontal claim appears, the absence of progression shown in the radiographic record is flagged.

Treatment frequency analysis: AI can flag frequency anomalies — a practice that submits crown claims at 3x the regional average for patients in certain age bands, or D4341 billing rates significantly above comparable providers.

Auto-Adjudication vs. Enhanced Review

Not every AI flag produces an automatic denial. Payers use a tiered response:

• Auto-adjudicated (approved): Claims where AI scoring is consistent with clinical documentation and within normal treatment patterns — these often process faster with AI than without it, which is a benefit to practices with clean documentation.
• Enhanced review queue: Claims where AI flags a potential inconsistency but confidence isn't high enough for auto-denial — a human adjudicator reviews these with the AI scoring as context. The adjudicator is now primed to look for the discrepancy, which affects their judgment.
• Auto-denied: Claims where AI confidence in a documentation-radiograph inconsistency exceeds the payer's threshold — these go straight to denial without human review.

The auto-denial threshold varies by carrier and is not publicly disclosed. This opacity is a legitimate industry problem — practices are being denied by algorithm without knowing the scoring criteria.

---

What Delta Dental's Overjet Deployment Means in Practice

Delta Dental has been explicit in investor and partner communications about its AI adjudication capabilities. The system is designed to:

1. Reduce fraudulent claims — the stated goal, and a legitimate one
2. Improve adjudication consistency — AI scores the same X-ray the same way every time, unlike variable human adjudicators
3. Accelerate clean claim processing — AI can process straightforward claims faster than human review queues

For most practices with clean documentation, this system should theoretically be neutral or positive — accurate documentation of genuine findings should score well. The problem is in the gray zones:

Clinical judgment gray zones. A finding that justifies treatment in one experienced clinician's judgment may not produce a definitive AI score. Early-stage interproximal caries in enamel is a legitimate treatment decision in some clinical philosophies (conservative restoration) and a watch-and-wait decision in others. AI adjudication that defaults to "no lesion confirmed" when the finding is ambiguous will deny legitimate clinical decisions.

Documentation quality gaps. Practices that don't document the radiographic findings supporting treatment in their clinical notes — practices that code the treatment without documenting the indication in the record — are now more exposed than they were when human adjudicators reviewed claims more charitably. An AI doesn't fill in documentation gaps the way an experienced claims examiner sometimes did.

Radiographic quality issues. Bitewing X-rays taken at suboptimal angulation, with cone cutting, or with processing artifacts will score differently than clean X-rays of the same pathology. If your imaging technique or equipment quality produces ambiguous radiographic evidence, AI adjudication will flag more of your claims.

---

The Documentation Changes That Matter Most

For Restorative Claims

The AI is looking for radiographic evidence of caries that justifies restoration. Your documentation should include:

• Specific tooth and surface identification for the carious lesion — not just the treatment planned, but the finding that justifies it
• Depth characterization: "interproximal caries approaching DEJ on mesial surface of tooth 14" is materially more defensible than "cavity on 14"
• Radiograph reference: Note which radiograph (bitewing date, full series date) contains the diagnostic finding
• Prior observation if applicable: "Lesion observed on 6/2025 BWX, now demonstrating progression" gives the AI historical context

For composite restorations where the AI may not see a definitive radiographic lesion — because the lesion is clinically visible but radiographically subtle — add a specific clinical examination finding: "Caries detected on clinical examination with explorer, confirmed by tactile resistance and cavitation. Radiographic shadow present at mesial contact."

For Periodontal Claims

Periodontal treatment claims are among the highest-frequency AI denial triggers because the radiographic evidence of disease severity (bone loss) has to align with the clinical documentation of severity (probe depths, bleeding, mobility).

Key documentation practices:

• Probe depths must be documented in the chart before billing D4341/D4342. This sounds obvious, but practices that do full-mouth perio treatment without updated documented probe depths create an automatic documentation gap that AI adjudication flags.
• Bone loss characterization: Note the percentage of bone loss radiographically: "Generalized mild-to-moderate horizontal bone loss, crestal bone 2-4mm apical to CEJ on posterior teeth bilaterally."
• Furcation involvement where present: Furcation involvement is a key AI-scored finding that supports periodontal disease severity. Document it explicitly.
• Calculus location: "Subgingival calculus evident radiographically on posterior teeth" connects the radiographic finding to the scaling need.

For Implant and Major Restorative Claims

Implant claims increasingly require prior authorization, and AI is being applied at the prior auth stage as well. Bone density, ridge dimension, and opposing dentition context in the radiograph are all being scored.

• Include a periapical radiograph that clearly shows available bone volume
• Document the reason for tooth loss and the time elapsed since extraction
• If there is a bone graft, document the graft material, site, and the timeline for osseointegration assessment

---

When AI-Assisted Denial Hits a Valid Claim

The denial is not the end. AI adjudication can be wrong — it's making probabilistic inferences from radiographic data, and those inferences can miss clinical context that a human reviewer would incorporate.

Step 1: Request the Denial Reason Code

Every AI-assisted denial should come with a denial reason code. For Overjet-based adjudication at Delta Dental affiliates, look for codes referencing "radiographic insufficiency," "documentation inconsistency," or "clinical necessity not established." These codes tell you what the AI flagged.

Step 2: Prepare the Appeal Package

For clinical necessity appeals on AI-flagged claims, your appeal package needs to do one thing: provide the clinical context the AI couldn't access.

• A narrative letter from the treating clinician explaining the clinical basis for treatment
• The relevant radiograph (clean, properly processed, good diagnostic quality)
• The complete clinical notes from the date of service and the preceding examination that identified the condition
• ADA clinical guidelines reference if your treatment decision is at the conservative end of the spectrum

Step 3: Request Human Review

For AI auto-denials, you have the right to request human adjudicator review. Request it explicitly in your appeal: "We request human clinical review of this claim and do not consent to AI-only adjudication."

This request is not always honored, and the regulatory framework governing AI adjudication in dental is still developing — but making the request creates a record and escalates the claim to a level where human judgment enters the process.

Step 4: Track AI-Denial Patterns

If you're seeing repeated AI-based denials on the same procedure codes or the same payer, that's a pattern problem, not a claim-by-claim problem. Document the pattern and bring it to your state dental association or to your state insurance commissioner if the denial rate is statistically anomalous.

---

Axlow for Policy Verification Before Submission

One of the most efficient ways to reduce AI-triggered denials is to verify payer policy for specific procedures before treatment — knowing exactly what clinical criteria and documentation the payer requires for each code you're billing.

Axlow (axlow.com) provides dental payer policy verification — helping practices confirm coverage criteria, documentation requirements, and prior authorization rules by payer and procedure code before the claim is submitted. If you know Delta Dental's Overjet-scored criteria for D4341 before you document, you can structure your clinical notes to meet the stated criteria. That's not gaming the system — it's understanding what you're documenting against.

The gap between a denied claim and an approved claim for the same clinical scenario is almost always a documentation gap, not a treatment validity gap. Knowing the criteria before treatment is the most efficient way to close that gap.

---

The Broader AI Adjudication Landscape

Overjet is the most visible payer AI partnership, but it is not the only one. Other tools operating in the payer adjudication space include:

• Videa Health: Has carrier partnerships beyond practice-side use; AI caries and bone loss detection is being licensed for payer review
• Verifiable Health: Focused on dental claim verification and fraud detection, used by several carriers
• SKYGEN (formerly DentaQuest's tech platform): Integrated AI adjudication tools used in government dental programs
• Internal carrier tools: Several large carriers have built proprietary radiographic review AI that doesn't publicly identify the underlying model

The trajectory is clear: radiographic AI adjudication is becoming standard infrastructure on the payer side. Practices that adapt their documentation practices now will see this as a neutral or positive development — cleaner documentation, faster processing of well-documented claims. Practices that don't adapt will see increasing denial rates on claims that have historically been approved.

The tool isn't going away. The documentation standards it enforces are now part of the claims landscape. See the Avized vendor profile for Overjet for the practice-side perspective, and use Axlow to verify your documentation is meeting payer criteria before claims go out the door.

---

A Note on the Regulatory Gap

The use of AI in insurance adjudication is operating in a regulatory environment that hasn't fully caught up with the technology. Most states have AI-in-healthcare transparency legislation in progress, but dental insurance is regulated at the state level and the patchwork of AI disclosure requirements varies significantly.

• Several state dental associations are actively lobbying for disclosure requirements on AI-adjudicated denials
• The ADA has published policy positions on transparency in AI-based claim processing
• Class action litigation regarding AI-only health insurance denials (primarily medical, but with dental implications) is creating case law that may affect dental AI adjudication standards

Stay current through your state dental association. The regulatory landscape around this specific issue is moving, and practices that document their AI-related denial experiences create the evidence base that state associations need to advocate effectively.