-
-
Function that takes in training data to recognize keywords like the cause behind their stress via co:here
-
Sentiment analysis that helps us determine the user issues, causation and story vagueness
-
Once a user has talked about what's on their mind, they're given a story that relates to their situation based on event and issue
-
Training for NPL by feeding in sentences and the categorization we want i.e isolating the cause of the users stress
-
Training co:here to summarize large amounts of text into a summary we can post
-
Page that the user starts on
-
Sharing a negative experience
-
Sharing a positive experience
-
NPL autofills based off of user inputs as it can recognize the issue they're going though and the causation behind it
Inspiration
Mental health issues are something that affect every demographic, ethnic group, and gender. The problem is that lack of quality attention paid to those in need. Providing facilities such as counselling and mental therapy is phenomenal but a key challenge is getting individuals to avail such resources.
Persons with mental health issues are sometime reluctant to speak with another human because they've felt ostracized from the "normal" community all their lives. This does not mean they don't deserve other options to express their emotions.
Our team personally knows young students we consider friends and family who will not go and speak with councillors about their situation. This alone inspired us to want to help not only the people in our lives but others as well.
What it does
L . A . D has two types of anonymous users:
1 ) Users sharing personal success stories to help others 2 ) Users sharing their current situation for help.
success stories are anonymously saved and shared with users who are seeking help regarding their current situation by having the power to share.
A user seeking help does not speak to a human but rather our AI L . A . D, a master of NLP, that extracts key features concerning their situation from chat and provides success stories that truly relate to the users current situation. This is because others have submitted genuine, real, and raw stories aimed at helping others who may be experiencing something similar. Many individuals who overcome tough circumstances are passionate about helping others in the same situation and users in a negative state may relate more to a person nearby their age, gender, ethnic
How we built it
The AI was trained using a supervised learning model from data found in online databases and forms related to mental health. The data related to individuals experiencing loneliness, depression, stress, anxiety, and suicidal thoughts was stored in csv files to be fed into our model which utilized the co:here API for NLP. With the NLP features L . A . D is able to categorize an individuals mental state with some quantitative confidence along with cause.
The front end was built in flask for easy rendering and bootstrap for styling.
Challenges we ran into
-Relevant data collection for model training -Data cleaning -AI integration with front end
Accomplishments that we're proud of
-Scratching the surface of machine learning and AI -Creating a effective classification model in under 30 hours -Having a finished product -Learning new technologies such as flask and co:here -Creating a product to help those in need, including our friends
What we learned
two of our members never used flask before. This project not only got them introduced to the tech stack, but also the problem resolving part of it, as we encountered code breaks and bugs often. This project got all the members motivated to learn and perform better in making projects.
What's next for HTV PRODUCTIONS
At HTV PRODUCTIONS, we believe in quality and best way to help the needy. We plan on finishing the project and having a proper system that helps the people with mental issues, and provide a community they never thought could exist.
Log in or sign up for Devpost to join the conversation.