WellBud

• 

  This is the message received from twlio

• 

  This is twlio phone call number

Inspiration

This is a Mental Health Track. We were inspired to work on this project as we felt that issues relating mental health is very alarming and thus we wanted to apply our knowledge of data science and machine learning to solve this issues. Covid times has made us experience similar situations at some times or the other and therefore our goal was to work on this project.

What it does

This project uses the English twitter messages data fetched from an open-source (Kaggle) and runs a sentiment analysis using Natural Language Processing on these messages. Moreover, a Deep Learning model is trained and tested to perform the sentiment analysis to filter out if the emotion is anger, fear, joy, love, sadness, or surprise. Further, if the emotion is found to be of sadness or fear for the user, a chatbot is triggered which helps to lightens the mood of the user. The chatbot works to give the user an option to either have received a motivational message on their phone or listen to a soothing song through a call through twilio.

How we built it

We worked on the English messages from twitter as the training and testing data for developing our Deep Learning Model to implement sentiment analysis to predict the emotion to be anger, fear, joy, love, sadness, or surprise. Moreover, we have implemented all the preprocessing tasks of Natural Language Processing of tokenizing, vectorization, etc. to format the data to be utilized for model development. Further, if the sentiment is fear or sadness, we called a chatbot to lighten the mood of the user by giving them option to either receive a text message with a motivational quote or listen to a calming music over a call through twilio. Overall program is written in python programming language.

Challenges we ran into

We ran into some challenges such as the model's accuracy was coming to be low and thus it was giving wrong predictions. But then we fixed it by hyperparameter tuning. Also, we wanted to incorporate a feature wherein the user can get an option to talk to a doctor nearby. For this we thought to use a database of doctors nearby but due to lack of time we were not able to implement it.

Accomplishments that we're proud of

We are proud that we were able to complete our project and implement 90% of what we have thought to implement on this project. This was our first hackathon experience and we are proud that we worked on deep learning and chatbot technologies.

What we learned

We have learned a lot about Natural Language Processing, Deep Learning model development and chatbot applications. Moreover, we honed our understanding of hyper-parameter tuning and accuracy of the model.

What's next for MentalHealthChatbot

As a next step we have the idea of linking a database that has a list of therapist/ psychiatrists clinic's address and phone number nearby the user's location. This will reflect as the 3rd option in the chatbot for the user to talk to someone to share their feelings with.

Built With

• keras
• machine-learning
• natural-language-processing
• numpy
• pandas
• python
• scikit-learn
• twilio