The main site for managing clients.
The graph for viewing pain level history.
An example of the conversation between bot and user.
Parents often under-medicate their children after surgery. This is due to a number of reasons such as losing track of time, not assessing pain properly, believing that the pain isn't as bad as the child says, or being concerned about over-medicating (http://pediatrics.aappublications.org/content/125/6/e1372). Pain is bad and can cause problems with the perioperative outcome. This app will help parents assess their child's pain level and determine if medication is required.
What it does
A doctor initially sets up the client, which triggers the bot process. The bot will initiate conversation with the parent and walk them through the pain assessment and medication process. All of the data is collected and stored. In the future, the data could be used to make better decisions about if the medication is not reducing the child's pain.
How I built it
I used the Microsoft Bot Framework and Azure Cloud to host the bot, trigger functions, data, and website. Twillio is used for sending SMS messages.
Challenges I ran into
The Microsoft Bot Framework is a new technology and doesn't work very well with proactive conversations (bot starts instead of reactive where the client starts the conversation). Testing the bot framework was difficult as well. It is also tricky to imagine how a conversation should progress and write it approriately.
Accomplishments that I'm proud of
I managed to get proactive conversations working. The trigger functions, bot, and website all communicate to each other.
What I learned
What's next for Emdee
I would like to implement Project LUIS for machine learning and have less constricted conversations.