A friend expressed how straining COVID has been on his mental health. He is reluctant to visit healthcare facilities to avoid contracting COVID. Because he lives with 86 year old grandmother, he is afraid of passing the virus to her since she is susceptible to adverse effects of the virus. As such, he does not have access to the help that he needs.
What it does
MindfuL is a virtual chatbot web application that provides immediate psychiatric help for those in need. Services include advice, uplifting messages, references to informative material, contact information for support groups and specialists, and emergency Services. Dialogue was consulted by practicing psychologists and linguists to obtain optimal results.
How we built it
MindfuL is powered by Google Cloud Platform, .NET Winforms, and Twilio. The front-end is a Winforms app built with C# and the backend server is a Flask app written in Python on a Google Cloud Compute Engine instance. The app uses the Twilio SMS API and the following Google Cloud services:
Dialogflow Firebase AutoML Sentiment Analysis Custom Search Translate Maps Compute Engine
Challenges we ran into
Setting up the individual components posed some challenges, especially bypassing some firewalls for the Compute Engine instance. Then capturing the nuances of psycho-analysis to obtain the best results posed some issues as well.
Accomplishments that we're proud of
The accuracy of the sentiment analysis model provided some promising results. We were able to effectively diagnose many instances of depressing and suicidal verbiage. This is promising for the future development of this product. We see the potential with this project to really help a lot of people.
What we learned
What we learned is the complexity of psycho-analytics and the abstract nature of the human mind. Despite the difficulties of this, we learned about the commonalities of human behavior to build a system that can effectively pick-up on behavioral patterns. With the proper intents, we found that we can generate positive conversations.
What's next for MindfuL
To optimize the performance of MindfuL, the following will be done:
- Obtain more professionally-reviewed & contributed data
- Improve sentiment analysis to better grade rhetoric
- Improve Dialogflow for optimal user experience
- Login feature to maintain past conversations
- More accurate predictions of user intents
- More applicable material for assistance
- Add computer vision capabilities to analyze common physical cues of depression
- Add social network to create a supportive community
- Engaging activities to inspire confidence and raise morale
- YouTube videos for How-Tos and inspirational videos