Anxiety, stress, and depression are on a disturbing upward trend in modern society. Often, people experience symptoms of these mental health issues at varying levels of severity. People might not be sure whether or not they actually need treatment. Because of barriers such as cost, lack of access, or mere intimidation, some individuals might hesitate to seek treatment when they actually need it. They thus miss out on the benefits of having a trained professional listen with empathy to their
What it does
Empact provides a variety of mental health services. The main feature is an intelligent, and most importantly, empathetic virtual assistant. While not a substitute for a licensed therapist, Empact is trained to recognize the symptoms of depression. After identifying the symptoms in the user, it suggests a potential diagnosis and recommends actions to take. If the user ever mentions they are in danger of self-harm, Empact immediately asks for a phone number and calls the user. In addition, Empact locates nearby options for mental health treatment, as well as promoting self-care by suggesting healthy practices.
How we built it
First, we researched the symptoms of depression and trained the IBM Watson Assistant instance to request and recognize them from the user. Then, we added a specific emphasis on potential self-harm or suicidal thoughts, in which case we use IBM Cloud Functions with Twilio to reach out to the user's phone directly. Finally, we embedded the chatbot on the Wix site, including a Google maps search for nearby mental health resources and the recommended best practices.
Challenges we ran into
We had difficulty training Watson to recommend such a large variety of symptoms and building the logic to continue to collect them. We also had difficulty integrating the Twilio request into the chat flow. We also struggled in integrating the custom use of the Google Maps API into our Wix site.
Accomplishments that we're proud of
We arrived as three complete strangers, but we quickly learned how to work effectively as a team. We divided up tasks and leveraged each other's strengths to accomplish a variety of tasks that none of us have never done before. We sought help when we needed it and learned a ton along the way. We will leave as not only teammates, but as friends.
What we learned
We learned how to use serverless architecture to process requests and responses from different API calls. We learned about training a chatbot to recognize certain key aspects in natural text.
What's next for Empact
We would like to add the ability to recognize more mental health conditions in addition to depression, such as anxiety. We would also like to integrate online resources such as remote therapy and suicide hotline services in addition to local mental health treatment options, which would be especially beneficial to users located far from affordable local options. We also hope to refine our project and submit it for IBM's 2019 Call for Code.