Agent · 100% on-premise · our own reasoner

Patient Segmentation

12 value cohorts for targeted campaigns and clinical reactivation.

Request demo View the full agent network

What this agent does

Patient Segmentation is the agent that classifies the active patient base into 12 value cohorts (Recency, Frequency, Monetary). Each patient receives a score and a cohort label (champions, loyal, at risk, hibernating, etc.) that the manager uses to design targeted campaigns. All computation is local.

Why it matters

A mass campaign offering a dental cleaning to the whole base is noise for active patients and budget waste for inactive ones. Patient Segmentation lets you speak to at_risk with a different tone and offer than champion. Campaign ROI goes up and the base does not burn out.

How it works

The agent computes three normalized scores per patient: Recency (days since last visit), Frequency (annual visits) and Monetary (cumulative billing). It combines the three into a 12-cohort classification established by the product owner. It recomputes on each new appointment or invoice — real-time scoring with no batch.

Integration with the clinical workflow

Patient Segmentation passes the per-cohort patient list to Reception to generate campaigns with appropriate templates. To Patient Flow to feed the decision of which cohort to prioritize when there are gaps in the schedule. The manager receives a monthly dashboard with cohort evolution.

Autonomous decisions it makes

Recompute a patient's value on each new appointment or invoice
Assign cohort label according to the 12 defined value cohorts
Flag significant cohort changes (e.g. loyal → at_risk)
Expose filterable patient list per cohort to the manager
Always work on pseudonymized data for aggregate analysis

Inputs and outputs

Receives

· Patient appointment history (dates, attendance)
· Patient billing history
· Procedure type (for weighted Monetary)
· Clinic's value threshold configuration

Produces

· value score per patient (3 normalized values)
· Cohort label (1 of 12)
· Filterable list per cohort for Reception/manager
· Alert on significant cohort change

Production metrics

12

Cohorts

Real-time

Scoring

Pseudonymized

Analysis privacy

Collaborates with

Financial Forecast

Patient Flow Forecast

No-Show Prediction

Reception

Tech stack

Model

Normalized value scoring (no LLM)

Execution

Local CPU (lightweight computation)

Latency

Real-time on events, immediate query

Privacy

Aggregate analysis on pseudonymized data, GDPR-compliant

Frequently asked questions

Are the 12 cohorts fixed or configurable?+

They are configurable. The base catalog follows the standard value model (champions, loyal, potential, at_risk, hibernating, etc.) but thresholds and names can be adjusted per clinic. By default they come pre-calibrated based on real Microdental data.

Does segmentation let me send email campaigns automatically?+

The agent generates the lists. Campaign sending is coordinated with Reception (templates + email channel). The marketing lead approves before mass sending.

Can patients know which cohort they are in?+

No. The cohort is internal information for the manager and only used to personalize the offer. The patient never sees the label; they only receive the communication with appropriate tone and offer.

Is the patient's marketing consent respected?+

Yes. List generation automatically excludes patients who have revoked marketing consent. Clinical data remains available for clinical use, but is excluded from any campaign.