Inspiration
During the pandemic, all of us underwent serious lifestyle changes. We realized that a major problem in our lives has been the growth of things to work on with limited pathways to vent or stress. Anxiet-ease was made to help all those students, employees, parents and any other group who have had their work increased over the pandemic leading to further stress on their lives.
What it does
Our application provides easy access to resources that can help ease your anxiety. Our resources include anxiety easing activities like yoga and meditation, various exercises, different cooking recipes as well as calming music. We have also provided a steady flow of animal pictures using the unsplash API, so that the user never runs out of animal pictures. The app also takes into account the user's preferences and suggests the most effective calming techniques based on previous anxiety attacks.
How we built it
We have a website, hosted on Heroku and designed on Figma, which includes a profile section, calendar (with monthly and weekly views) and a page with all the activities. We take user-based data and link it to a Colab notebook acting as a backend which uses time of attack and the delta change of how the user feels before and after the attack to predict the next possible attack. This data is then displayed on the calendar. We have an IOS application which includes all these features along with google sign in to link your google calendar so as to provide further data for the model to work with.
Challenges we ran into
Hosting a website after coding it completely in HTML took a fair amount of time. Linking the Unsplash API to our app to provide users with new pictures continuously and instantaneously was a challenge.
Accomplishments that we're proud of
Having a fully functional website and application, a working prediction based Machine-Learning Algorithm and the curated relaxation techniques.
What we learned
Converting Figma designs to working HTML code, connecting APIs to SwiftUI for the application, Single shot prediction of sequences.
What's next for Anxiet-ease
The next step for the app would be to fully automate the entirety of the process by connecting all the separate entities to establish a unidimensional workflow all the way from the website inputs to the calendar. Another major improvement which we regard as highly important is developing an ML network which takes in the other data from the user such as smart watch position, heart-rate etc. to provide more accurate estimates of when to expect an attack. This would be in the form of a classifier with a softmax output to give weights to each option. We would also consider how much better the user feels and link that with the activity used to calm. This would greatly improve the efficacy of the system as a whole and have a considerable impact on the final application. However, given that the crisis exists in the now, there is a great need for urgent deployment of this application even in its crudest form with the basic learning algorithms.
Log in or sign up for Devpost to join the conversation.