I am new to Rasa and was hoping to get some guidelines on how to recreate a simplified version of Symptom Checker, Check Your Symptoms in Real Time | Buoy just for learning purposes.
What I’d like to achieve is a dialogue flow in which the user initially specifies one symptom (or is instructed to provide a symptom), the assistant retrieves a few associated symptoms from some external service (associated symptoms model) and asks the user about those all symptoms as follow up questions. Once the followup questions are all answered, then the assistant passes all the symptoms present (and not present) to some disease prediction model based on the symptoms to output some potential diagnoses.
For now I would just like to keep things simple with either yes the symptom is present, or no the symptom is not present (as I’m not sure how to handle if a user were unsure whether the symptom is present).
Example conversation:
input> I think I have a cough
#the entity is recognized as cough either from nlu.md or a symptoms.txt lookup table
output> let’s determine some potential diagnoses
output> do you have a fever?
input> yes/no (or if this is possible via buttons then great)
output> do you feel dizzy?
…
output>possible diagnoses: bronchitis, benign cough, allergies, etc