Prepared by the scanO editorial team. Editorial draft prepared 2 November 2026; independent specialist review pending.
Quick answer: AI dental screening for schools should be evaluated as a clinician-led hybrid workflow. Compare the whole encounter, record quality and parent handover against a well-run manual process. Do not compare an AI analysis time with the time for a complete dental examination or claim that software replaces the dentist.
Schools need a predictable event, parents need understandable information and dentists need records that support appropriate care. A device can help one part of that service while adding work elsewhere. The right comparison measures the whole pathway rather than counting photographs captured.
For the manual arm, specify registration, guardian permission, clinical assessment, findings record and report delivery. For the technology-supported arm, include those same tasks plus capture, retries, software review and any data handling. Keep clinical scope and referral criteria comparable.
The scanO air product description presents a screening and reporting workflow. Treat the demonstration as a starting point, then verify age suitability, access and the proposed school configuration. Brand claims about speed do not establish completed examinations per hour in your event.
Record arrival-to-completion time and staff time separately. A child may wait while the device processes another record, or a staff member may spend extra time matching reports later. A quick capture can coexist with a slow overall service.
Measure consent checks, registration, examination, capture, retries, counselling and report handover. Report the median and range where the sample supports it, together with incomplete encounters. Do not exclude every difficult case from the technology arm and then call the result representative.
Check whether the record identifies the correct child, distinguishes observations from diagnosis and records examination limitations. A legible paper form may outperform a digital record attached to the wrong profile. Equally, a structured digital report may reduce transcription where the workflow is implemented well.
Ask a parent-facing reviewer to read a synthetic report without explanation. Can they describe the next step? Do the words match the clinician's plan? A report is not effective merely because it has colour, a tooth diagram or an automated score.
An encounter may be incomplete because of distress, access, lighting, capture quality or missing permission. Define what the team does before the event. Do not force a child to complete a scan so the device's utilisation figure looks better.
The manual fallback needs an actual form, trained staff and a report route. It should not depend on staff improvising a notebook when the internet fails. Verify which features work offline and what later synchronisation requires for the exact product version.
The school can coordinate logistics without receiving each child's clinical detail. Confirm the guardian destination privately through the agreed process. Do not share named reports in a class group or use child images as promotional evidence of event success.
If a clinician identifies a time-sensitive concern, follow the clinical escalation process rather than waiting for a batch of automated reports. Record the contact, action and owner. A successful software delivery status is not proof that the parent understood or acted.
Agree the evaluation criteria before the supplier demonstration. Use synthetic records to test the system first, then run any real-world evaluation under appropriate consent and clinical supervision. Record the site, staffing, scope and reasons for exclusions so the results can be interpreted honestly.
Do not publish invented time savings, referral conversion or accuracy figures. The comparison table below deliberately leaves those as measurements to collect. A useful local result may be that record handover improved while total encounter time remained similar.
Success includes a workable service, understandable reports and appropriate follow-through. Higher treatment sales are not a suitable primary objective for a children's screening programme. Use the clinician's findings and parents' choices to guide subsequent care.
Ask the team whether technology removed a real bottleneck or moved it to another stage. If report matching remains manual, solve that before scaling to more classes. Review the school programme guide and use the parent report templates to keep the comparison grounded in the actual service.
If one group receives more staff, shorter forms or a different clinical scope, the time comparison is difficult to interpret. Record those differences and do not attribute them automatically to the technology. A fair operational test does not require pretending the workflows are identical; it requires making the differences visible.
Choose the unit of measurement before the event. Per attempted capture, per completed report and per completed clinical encounter are different denominators. If a child needs three attempts, all three belong in the capture workload even if only one final report is retained. Missing permission and distress should remain visible as separate reasons for incompletion.
Keep the same follow-up definition across workflows. Report delivery is one milestone, parent understanding another and attendance for further assessment a third. A system that sends reports automatically may improve the first without proving an improvement in the other two.
At the end of the pilot, list the conditions required for another event. These may include better record matching, a more private layout, an additional staff member or a clearer parent explanation. Assign owners and verify the changes rather than expanding while the same issues remain unresolved.
Review the fallback with staff who did not operate the demonstration. Can they complete the agreed paper record and handover if the device is unavailable? Can they explain why an incomplete scan is not a normal result? The fallback should be a rehearsed service, not an emergency improvisation.
Give the school an honest operational report. State the scope, dates, participation and known limitations without claiming diagnostic superiority from a local service test. If the pilot did not measure accuracy against a suitable reference standard, it should not publish an accuracy claim.
Finally, ask whether the observed benefit matters to the programme's objective. Improved documentation, fewer transcription errors or clearer reports can be worthwhile even without a dramatic speed increase. The purchase decision should reflect a demonstrated operational benefit and acceptable clinical boundaries, not the need to declare one method universally better.
MeasureManual workflowAI-supported hybrid workflowClinical scopeState examination and limitationsState the same scope; identify the software's additional roleTotal encounter timeObserve locallyObserve locally, including retriesStaff timeInclude transcription and reportsInclude setup, matching and exception handlingIncomplete encountersCount and explainCount capture failures and other incompletesParent report clarityTest with a synthetic reportTest with the same understanding questionsClinical reviewNamed examinerNamed examiner; software output reviewedReferral completionTrack under agreed permissionTrack using the same definition and windowPrivacy and accessSecure forms and handoverVerify access, destination and export controls
Pilot decision note: retain, revise or stop; observed benefit; unresolved risk; extra staffing; missing evidence; next review owner. Do not publish a superiority claim from a small uncontrolled pilot.
Operational check: Keep the original observation, the responsible decision-maker and the next action together. Review unresolved items explicitly; do not treat a missing response as evidence that the task was completed.
For further practical context, see how AI dental screening works. These resources support planning; adapt decisions to your clinic and review clinical requirements with qualified professionals.
