Article
Renato Banzai · Aug 27, 2020 2m read

Using FHIR to Interact with Natural Language

Whats NLP Stands For?

NLP stands for Natural Language Processing which is a field of Artificial Intelligence with a lot of complexity and
techniques to in short words "understand what are you talking about".

And FHIR is...???

FHIR stands for Fast Healthcare Interoperability Resources and is a standard to data structures for healthcare. There are
some good articles here explainig better how FHIR interact with Intersystems IRIS.

My Solution

picture

Chatbot to query FHIR

The most common approach in chatbots is use machine learning to train the model based in old conversations. But when the
chatbot need to use "live" data the challenge is improved. If you already know how a machine learning model works, you know
that is most of cases the trained model is a kind of "hadcorde" application with all decision embedded inside. If you need
more data without retraining the model you will need to create by yourself the integrations and put your mind to work part
in machine learning and part in software engineering.

First Approach

To show more about FHIR I have decided to first work on questions that a healthcare professional could do to a chatbot.
And to do this I keep the software engineering simple, using just javascript and regular expressions to look for intentions
and entities on the chat.

Regular Expressions

picture

This is my favorite subject in processing texts. Regex are a powerful way to work text and match patterns. With regular
expressions I can search for what patient the person chatting wants to know about and what question or chart he wants to see.

If this article help you or you like the content vote:

This application is at the current contest on open exchange, you can vote in my application fhir-chatbot here

30
2 0 2 209

Replies

Thanks, Renato!

You introduced the graph! Nice work!