We were inspired to address the problem of mental health on a large scale, which has been getting increasing recognition over the past few years but still has few affordable, readily available treatment options.
What it does
Therapize connects people seeking mental health with volunteer professionals, aiming to increase access to mental health services, which the University of Cypress found in a 2020 study is positively correlated with improvements in quality of life, morbidity, mortality, and life expectancy. Patients can choose therapists based on their rating (generated by asking for reviews after sessions and performing a sentiment analysis of those reviews with a Hugging Face BERT API), personality type, and time available to chat. The chat interface, built with daily.co, also helps therapists perform their jobs by performing a mood analysis using the DeepFace library on patients' faces and updating the therapists on the predicted feelings of patients on a minutely basis. We use Firebase to store data about professionals who are available at given times.
How we built it
We built a React web app and outfitted it with a number of features, including mechanisms for therapists to understand their patients emotion and to leave reviews for therapists which implement API's for convolutional neural networks and bi-directional transformer based neural networks. Moreover, when a therapist logs onto our page, their information, including personality type, rating, and time available, gets added to a firebase database and displayed so that patients can select a therapist with whom they best match.
Challenges we ran into
Integrating all of our functionality into a single UI was the most difficult part of the project. Setting up video conference rooms on a website was also not easy and we went through multiple phases and attempted to use multiple libraries and interfaces before ultimately succeeding.
Accomplishments that we're proud of
We are proud of overcoming the challenge discussed above and more generally having a number of independent features calling advanced and interesting API's to improve user experience.
What we learned
We learned a lot about building a multi-layer website with React and attaching Python functionality; we also developed experience with numerous libraries and firebase and figured out how to facilitate video calls on a website with daily.co.
What's next for Therapize
We hope to add a voice sentiment analyzer to further alert therapists to the mood of patients. We also would like to put up Therapize on a regular domain instead of running it on local host. Lastly, we want to create some kind of incentive for patients and therapists alike to use our service. Potentially, we could mint custom NFT's and distribute them to users' crypto wallets. We would ensure that the NFT's are on a proof of stake blockchain so that we had no negative environmental impacts.