Rapport: patient behaviour replicas
WYnova developed patient behaviour replicas for Rosemark Rapport that predict the response before a message is sent, with +63% projected engagement by segment.
- Client
- Rapport
- Sector
- Health
- Service
- AI Solution Builder
- Time to MVP
- 8 weeks
projected engagement by segment
up to the MVP
The challenge
The same message rarely has the same effect on every patient. To find out what worked in each segment, the team had to send, wait and measure.
Key obstacles
-
General messages
The same text for very different patient profiles.
-
Late feedback
The effect of each message was only known after it was sent.
-
Heterogeneous segments
Very different behaviour patterns across segments.
The solution
WYnova custom-built patient behaviour replicas: models that reproduce each segment's response, so that messages can be personalised and engagement predicted before sending. The MVP was completed 8 weeks after kick-off.
-
01
Data
Profiles and interaction history for each segment.
-
02
RepliesAI
Models that replicate the behaviour of each patient group.
-
03
MessagesAI
Message variations tailored to each segment.
-
04
ForecastAI
Estimated engagement for each variation, prior to submission.
-
05
Human validation
The client's team selects and approves the message to be sent.
With each segment's response predicted before sending, personalisation no longer depends on trial and error.
Results
+63% projected engagement by segment, with messages personalised for each patient profile. The MVP was completed in 8 weeks.
What we learned
In healthcare communication, simulating each segment's response makes it possible to choose the message based on a prediction, rather than waiting for the results of each send.
Related cases and services
More cases and solutions in AI for healthcare and pharmaceuticals.
