Chair time is the whole economics
A dental practice's capacity is surgeries multiplied by hours. An unfilled slot cannot be recovered, and the fixed costs continue regardless.
Almost every worthwhile operational application therefore comes back to keeping chairs productively occupied: reducing no-shows, filling gaps quickly, scheduling appointment lengths accurately, and keeping patients in recall.
No-shows and short-notice cancellations
Patient history is the dominant signal, as in most appointment-based businesses. Beyond that, appointment type, how far ahead it was booked, time of day and the gap since the last visit all contribute.
- Target reminder effort at high-risk appointments rather than reminding everyone identically
- Keep a short-notice waiting list of patients wanting earlier appointments, prioritised by predicted availability
- Consider scheduling higher-risk appointments where a gap is least damaging
- Track whether reminders actually change behaviour - many practices have never measured this
That last point is worth acting on before any model. Reminder effectiveness is measurable with an experiment costing nothing, and the answer sometimes shows a channel is doing very little.
Recall compliance
Patients drifting out of recall is the quiet leak in most practices. They do not formally leave; they simply do not rebook, and by the time anyone notices it has been two years.
Predicting who is at risk of lapsing - from visit intervals, recall response history, treatment history and engagement - allows earlier, more personal contact. A call at three months overdue is far more effective than a fourth letter at eighteen months.
This tends to be the highest-return application in a dental practice, because a retained patient is worth considerably more than the cost of noticing they were slipping away.
Appointment length prediction
Booking every appointment of a type for the same duration guarantees the day runs late or leaves gaps. Actual time varies by patient, clinician and complexity.
| Input | Effect on duration |
|---|---|
| Clinician | Consistent individual differences |
| Patient history | Anxious or complex patients take longer |
| Treatment type and tooth | Substantial variation within one code |
| Time of day | Later sessions often run longer |
Better duration estimates improve the day for everyone - fewer overruns, less waiting, less stress at reception - and require no clinical judgement from the model.
Treatment plan acceptance, used properly
Predicting which plans patients accept is straightforward from history, and the appropriate use is improving how plans are explained and financed, not deciding who to offer treatment to.
Used to identify where patients commonly decline on cost, it supports better payment options and clearer explanation. Used to decide who hears about which treatment, it becomes clinically and ethically indefensible.
Radiographic diagnosis is a separate, regulated field with a much higher evidence bar and should not be bundled into a practice operations project.
The patient who quietly stopped coming is worth more than the one who cancelled - at least you knew about the cancellation.