Project Title

A brief description of what this project does and who it's for

Conversational Agent for Asthmatic Patients (Master Research Project)

This project is demonstration of a general purpose Conversational Agent, currently domain constrained on Asthma. However, much of the methodologies followed in this project are extendable to different chronic diseases.

The project makes use of Sentence Transformers and GPT3 to answer user queries related to Asthma. Because of dependencies on different libraries like PyTorch that makes the environment of the project larger, a smaller requirements.txt file is attached.

The project can be run via docker compose up without having to build the environment locally, as the environment will then be setup inside the engine container.

Typical questions that the Conversational Agent can answer are as follows:

- I did not sleep well yesterday, and my Asthma has worsened. What should I do?
- Is not a good time to go out for a walk?
- Can I go swimming if I have Asthma?
- What happens to my data that is collected?

The questions can be passed as raw json body either via postman or any other alternatives. Below is the format of the json body that has to be passed:

API endpoint: 127.0.0.1:5000/api/v1/answer_davinci (Note: The endpoint can be changed to use the curie model by modifying it to answer_curie)

NOTE: It is advisable to keep the default port 5000 open for the application to run, as this port is set in a custom docker image that is created. If the port is occupied, you may open it with the following command:

sudo kill -9 `sudo lsof -t -i:5000`

JSON body:

{
    "question": "Can I go outside to play if I don't have asthma?"
}

Since there is no UI integrated at the moment as this is just the backend, unfortunately sending POST requests is the only way to get responses from the Conversational Agent.

Installation

Clone the project

  git clone https://github.com/tanmaychimurkar/ca-asthma

Go to the project directory

  cd ca-asthma

Note: Before running the script, we need to make sure to create a .env file which contains the following API keys as variables:

weather_api_key: (key from here: https://docs.ambeedata.com/)

openai_api_key: (key from here: https://beta.openai.com/signup)

The .env file has to be created in the project root where the docker-compose.yaml file exists

1. Run via entrypoint.py

  cd src
  pip install -r requirements.txt
  python entrypoint.py

Note: If running via entrypoint gives the error ModuleError: No module named src, then the following commands have to be run from the project root (from inside the activated virtual environment):

  python -m pip install --upgrade pip
  python -m pip install build
  python -m build --sdist
  python -m build --wheel
  pip install -e .
  python -m nltk.downloader all

After these steps, re-run the entrypoint and the code should then be working.

2. Run via docker

  docker compose up --build

Note: For running via docker, the default Flask port 5000 has to be open to accept connections, since the flask application is being run from the custom docker image that we have uploaded to docker hub.

Responses and the API execution time can be seen on mongo-express UI when run via docker at http://localhost:8081/