Welcome to the BMI Check Tutorial!
Here you will learn how to use more advanced features of the platform. Together we'll develop a more complex system, working with multiple types agents. You will also see how you can train new machine learning systems from scratch when you don’t have any initial data.
No matter how complex our systems get, we always want our agents to display these three core characteristics in communication with end users:
- I know you (and that's why I can take care of your specific needs)
- I have the expertise (and that's why my advice is relevant)
- I care about you (and that's why I'm proactively reaching out to you).
This combination brings human-to-machine communication to another level, and makes digital agents emotionally intelligent.
In this demo, we’re implementing a health assistant for identifying potential obesity issues in children. This is a real-life problem that is fairly well-defined, but not as straight-forward in its calculation as it is for adults. That's why at first it requires expert knowledge from a doctor who will evaluate the BMI queries themselves.
As patients flood the doctor with requests, it will be our goal to relieve the doctor’s workload. In the background, we'll fit a machine learning model based on the responses we obtain from the human expert in this communication process, and once it can replicate the doctor’s decisions with sufficient confidence, it will take over larger and larger parts of the doctor's workload.
This is a slightly more complex use case compared to the first tutorial, but we’ll get through it together step by step, taking care to implement all three core agent characteristics. Let's have a look!