Better information. Stronger safeguards. More useful dental intelligence.
DentAI is BOCA’s proposed dental intelligence layer, using machine learning to interpret radiographs, intraoral images, clinical documentation, and treatment history. Its purpose is to help verify evidence, detect patterns that may indicate fraud, waste, or abuse, and turn information gathered through care into useful clinical and administrative support.
Within BOCA Prime, DentAI would help assess whether submitted treatment is consistent with the available evidence, identify discrepancies that warrant review, and recognize patterns across patient and provider histories. That intelligence could help protect shared funds and reduce administrative work while preserving clinical judgment and human review.
Its potential extends beyond coverage. Helping practitioners translate descriptions of care into procedure codes, compare findings over time, and understand treatment outcomes could make DentAI useful to practices, patients, researchers, and other dental software providers.
01 / WHY BOCA NEEDS DENTAL INTELLIGENCE
Reliable rules need reliable information about care.
Knowing who is participating and defining how coverage works are essential foundations for BOCA Prime. Applying those rules also requires understanding the care being submitted: what was done, whether the documentation supports it, and how it fits with the patient’s history.
That information arrives in different forms. A practitioner describes treatment in clinical notes. A procedure code identifies the service being billed. Radiographs and intraoral photographs provide evidence, while earlier records add context. Making those pieces usable—and checking whether they tell a consistent story—can require substantial administrative and professional attention.
DentAI is designed to help with that work. It could translate descriptions into structured information, identify missing or conflicting documentation, and bring relevant findings to the attention of the people who need them. Catching a documentation problem early could help a practice correct it before it creates repeated requests, delays, or unnecessary review.
Protecting shared funds requires the same attention to evidence. Images associated with different patient identities, treatment submissions that conflict with the available records, or billing patterns that repeatedly differ from comparable cases can raise questions worth investigating. DentAI could help identify those patterns across more encounters than an individual reviewer could readily examine.
An unusual finding is a reason to look closer. Case complexity, legitimate differences in care, and incomplete records can all affect what the system sees. DentAI’s role would be to surface the evidence and explain the discrepancy so that an appropriate review can distinguish an error or justified exception from potential fraud, waste, or abuse.
The hypothesis behind DentAI is that specialized dental intelligence can make this work more consistent and less burdensome—helping practices document care, helping reviewers focus their attention, and giving patients and the system better information on which to rely. If successful, that could mean less time spent by front-office teams assembling submissions and responding to requests, fewer rounds of clarification, and faster coverage determinations and resolution of disputed or incomplete submissions.
02 / HOW DentAI WOULD WORK WITHIN PRIME
From submitted information to evidence the system can use.
DentAI would help connect the description of care, its supporting records, and the relevant history. Its proposed responsibilities fall into three areas.
01
Understand and organize / Making submissions easier to use
A practitioner might describe a procedure in a clinical note while images and other supporting details sit elsewhere in the patient’s record. DentAI could help extract the relevant information, suggest a corresponding procedure code, and organize those pieces for review by the practice before submission.
If a description omits a detail needed to distinguish between procedures, the system could ask for that specific clarification. The aim is to help practices submit complete, consistent information with less manual assembly and fewer follow-up requests.
02
Evaluate evidence / Checking whether the pieces agree
DentAI would compare submitted treatment information with the available supporting evidence. If a procedure identifies one tooth but the attached image shows another area, or the description appears inconsistent with the radiograph or intraoral photograph, it could identify the discrepancy and explain what needs clarification.
Working with Rootprint, DentAI could also help detect images showing the same dental anatomy under different patient identities. That finding could prompt a check for a record-assignment error or an attempt to participate through multiple identities.
Each check would produce a specific question or finding for correction or review. The evidence available for a particular procedure would determine what could reasonably be assessed.
03
Recognize patterns / Looking beyond a single submission
Some concerns become apparent only across repeated encounters. DentAI could compare treatment, utilization, and billing patterns over time, using similar procedures and patient circumstances to establish useful points of comparison.
For example, an oral surgeon might routinely bill substantially longer sedation times for impacted wisdom-tooth procedures than providers treating comparable cases. DentAI could flag that recurring difference and bring together the relevant records for review. The review would consider whether case complexity or other documented circumstances explain the pattern.
The same approach could help identify repeated treatment on the same tooth, unusually frequent use of a procedure, or patterns of complications that deserve closer examination. Those findings could support both protection of shared funds and efforts to understand clinical quality.
How the responsibilities fit together
Rootprint
Continuing patient and provider identities
DentAI
Analyzes the information
Prime Core
Applies the defined coverage rules
Practitioners and reviewers
Clinical judgment and human review
Rootprint helps connect records to continuing patient and provider identities. DentAI analyzes the information associated with those identities and surfaces relevant findings. Prime Core applies the defined coverage rules to determine the system’s contribution.
Practitioners remain responsible for diagnosis and treatment decisions. Where a discrepancy requires judgment, an authorized reviewer would examine the evidence and context. A DentAI flag would initiate a closer look, rather than serve as an automatic finding of misconduct or a coverage denial.
The intended result is a clearer path for routine submissions and more focused attention where a question genuinely needs to be resolved.
03 / MAKING EVERYDAY WORK EASIER
Describe the care. Spend less time translating it for the system.
The practitioner should be able to describe care naturally, while the software helps translate that into the structured information the system needs.
One proposed application for DentAI is coding and documentation assistance: helping translate a spoken or written description of care into suggested procedure codes and supporting details for the practitioner to review and correct. The aim is to reduce code searching, repeated data entry, and the work of preparing consistent submissions.
Further possibilities include helping practitioners review relevant treatment history and compare findings across available images over time. These applications would require additional design and testing with dental teams to establish where they are useful and how they fit into everyday practice.
Whether delivered through existing practice software or a separate interface, the test would be practical: does the tool reduce work and provide useful information without creating another administrative task?
04 / LEARNING FROM CARE OVER TIME
What happens after treatment should help inform what happens next.
A treatment record describes care at a particular moment. Understanding its quality and durability requires following what happens afterward. How long does a restoration remain serviceable? Where do complications develop? When does a treated tooth need another intervention—and what might explain it?
DentAI could help connect those later observations to earlier care, allowing practitioners and researchers to examine outcomes across time rather than evaluating each encounter in isolation.
Following outcomes / Learning what holds up over time
With appropriately authorized records connected through Rootprint, DentAI could help practitioners examine what happens after treatment. For example, a practice could look at restorations that remained serviceable for years alongside those that needed earlier repair, then explore whether materials, techniques, or patient circumstances help explain the difference.
Those comparisons would need context. A practitioner treating more difficult cases may see more complications, and a later intervention does not by itself mean that earlier care was inappropriate. DentAI would need to make gaps in the evidence visible and allow practitioners to correct records or contribute clinical context.
The purpose is practical feedback: helping dental teams recognize durable results, investigate recurring problems, and make better-informed decisions about future care.
Research participation / A separate choice, with a shared benefit
Learning across participating practices could also support research into treatment durability, complications, and outcomes. Participation in that research would be a separate, informed opt-in, with clear information about what records would be used, for what purpose, and who could access them. Joining BOCA would not automatically enroll a patient in research.
Where voluntary participation helps create commercially valuable research or tools, Cadmus would explore ways for contributing patients and practices to share in that value. The principle is that people should have a say in how they contribute and an opportunity to benefit from what their contribution makes possible.
A Learning Health Network / Bringing knowledge back to care
Together, these capabilities could support a Learning Health Network for dentistry: experience from care helps generate research questions, research tests those questions, and findings return to practitioners to inform future care.
The broader hypothesis is that BOCA could do more than coordinate and fund individual episodes of care. With the right evidence and safeguards, it could help dentistry learn from what happens afterward—and return that knowledge to the people providing and receiving care.
05 / DENTAI BEYOND BOCA
Intelligence developed for BOCA could serve dentistry more broadly.
The work DentAI is designed to support—interpreting dental information, checking evidence, and learning from outcomes—is relevant beyond BOCA’s coverage model. Capabilities developed and validated within Prime could become tools that practices, dental software providers, researchers, and other organizations choose to use.
For practices / Assistance that fits everyday work
Practices could use DentAI for coding and documentation assistance, reviewing relevant history, or examining treatment outcomes, even when the care is provided outside BOCA.
A further possibility is treatment-planning support. Given a proposed plan and relevant patient information, DentAI could help a practitioner compare available approaches, drawing on clinical evidence and appropriately comparable treatment histories. It could bring alternatives into view and explain the evidence, tradeoffs, and uncertainties associated with each. The practitioner would assess that information alongside the examination, clinical judgment, and the patient’s needs and preferences.
These capabilities could be offered through practice subscriptions or integrations with existing dental software. More advanced clinical assistance could be developed and validated separately as a later paid offering. Their value would depend on making everyday work easier and supporting better-informed care decisions.
Professional development / Turning feedback into improvement
Over time, feedback from a practitioner’s work could help guide their professional development. DentAI could help identify areas of consistent strength and recurring challenges that merit closer examination, then connect those findings to relevant continuing education.
For example, a pattern of earlier-than-expected restoration repairs might prompt a review of the cases and, where appropriate, education focused on material selection or technique. The practitioner would be able to examine the supporting evidence and contribute context before drawing conclusions.
The opportunity is to make education more responsive to the work a clinician actually performs, helping practitioners choose learning that addresses their needs and builds on their strengths.
For patients / Another way to understand quality
A further possibility is helping patients understand evidence about treatment outcomes alongside pricing information and patient reviews. A practitioner might eventually choose to share validated measures showing how particular treatments have performed over time.
Such measures could offer patients another basis for choosing care, provided they account for case complexity, follow-up, and the amount of evidence available. A meaningful measure would need to explain what it assesses and its limitations, with a process for correcting errors before publication.
This remains a research possibility. The aim would be to give patients useful context while representing practitioners’ work fairly.
For software providers and researchers / Specialized capabilities they can use
Dental software companies could integrate particular DentAI capabilities into tools their customers already use, such as documentation systems, procedure-coding workflows, or clinical record viewers. This could make specialized assistance available within familiar software, reducing the need for practices to adopt another standalone application.
Researchers could use analytical tools to investigate treatment durability, complications, and patterns across appropriately authorized records. Organizations conducting evidence review or quality analysis could use similar capabilities to organize supporting information and identify cases that need professional attention.
Research applications would follow the separate opt-in and shared-value principles described above. A software integration or commercial agreement would not itself authorize access to patient records.
The commercial opportunity / Several paths to useful products
For Cadmus, these applications could support practice subscriptions, software licensing, integration services, and research or quality-analysis tools. Individual capabilities could be developed and evaluated separately, allowing useful products to emerge without requiring the entire DentAI vision to launch at once.
The commercial case would rest on demonstrated value: time saved, more reliable documentation, better-directed review, and useful support for care decisions and professional development. If those benefits can be established, DentAI could become both an operating capability within BOCA and a source of products and services for the wider dental profession.
06 / WHAT NEEDS TO BE DEMONSTRATED
Useful intelligence has to earn confidence in practice.
DentAI’s proposed capabilities need to be evaluated for both accuracy and usefulness. Can it interpret dental information reliably, identify meaningful discrepancies, and help people act on its findings without creating unnecessary alerts or additional work?
Model development would require appropriately authorized training data, reliable clinical labeling, and sufficient representation of the patients, procedures, and imaging conditions DentAI is intended to support. Evaluation would need to use cases kept separate from training to test whether the system performs reliably on unfamiliar information. Further training and updates would require continued validation, with research participation and permission to use records for model training clearly explained.
Testing would need to examine performance across different patients, procedures, practices, and imaging conditions. Coding suggestions must accurately reflect documented care. Clinical assistance and outcome comparisons need sufficient evidence and context to support their intended use. Practitioners and reviewers must be able to understand the basis for a finding, identify mistakes, and correct them.
Privacy protections and research permissions must work as intended. Equally important is testing with dental teams to establish whether the tools fit everyday workflows, reduce administrative effort, and improve turnaround time.
Each capability should advance on the strength of that evidence. The goal is dental intelligence that protects shared resources, supports professional judgment, and returns useful information to the people providing and receiving care.