Smoke Detector

Command line prompt of output
Email to the user
Email to the sponsor

Inspiration

Smoking is a serious addiction that impacts a large portion of the population in the US. Quitting can be difficult and seem impossible for addicts, especially when many of them face difficulties through relapses. This is why we created an at-home assistant to track people's mental state and frequency of thoughts about smoking in order to help people fight the addiction and provide them with the support they need.

What makes the Smoke Detector special?

Early stages of relapse are usually characterized by emotional distress which makes the person at risk of falling deeper into relapse and eventually succumbing to physical relapse. Our device is meant to be able to identify when people are at these early stages and stop the relapse process before it progresses too far.

What it does

The Smoke Detector will assess if the user is at risk of falling into a relapse based on what they talk about in their day-to-day lives. If the Smoke Detector believes the user is at mild to moderate-risk of a relapse, it will notify the user and give positive reminders to encourage them to continue sobriety. If the user is at high risk of relapse, the user's selected friend/family member will be notified and encouraged to reach out to and support the user.

How we built it

We used speech recognition to convert speech to text. Once this is done, the text is both analyzed for frequency of keywords (such as "smoke" or "cigarette") and fed into the Google Cloud Natural Language API. The API will then return a sentiment analysis to the program and the program will assess if the user is at risk of relapse based on the frequency of keywords and the sentiment analysis.

Challenges we ran into

Some challenges we faced were setting up our development environment and working with new technologies and language for the first time.

Accomplishments that we're proud of

For most of our team, this is our first Hackathon so we are proud of the completion of the project and gaining experience in applying our knowledge.

What we learned

During this project, we learned Python, speech-to-text recognition, sentiment analysis, command line, how to work with APIs, how to send emails in Python, and time-management skills.

What's next for Smoke Detector

-Calibration: have the system calibrate to the person’s overall sentiment and respond to changes

-Use sentiment trend analysis to detect changes and spikes in sentiment

-Hardware: make it compatible with a smart assistant

-Make it compatible with more addictions

Built With

google-cloud
google.cloud.language
natural-language-processing
python
raspberry-pi
speech-to-text-recognition

Submitted to

Knight Hacks

Created by

I set up our development environment and was in charge of recording data with the Google cloud api

Jonathan Huang
Researched background for project and adjusted thresholds

Kristen Ling
I helped with much of the the development of the back-end of our program. Specifically by integrating an automated email functionality.

AJ Martin
I worked with Google Cloud's natural language and speech recognition APIs along with the logic of the system. It was my first time using a third-party library and it was also a first time for me coding in Python. It was a pretty hard project, but I ended up learning so much and building so much interest in speech recognition and Python language.

Uygar Baran
First time joining a Hackathon!