The original inspiration for Haven was our team’s recognition of the widespread issues of mental health, especially depression, in the world today. Our discussion of the way mental health issues are dealt with in this country led us to believe there are key gaps in the chain of care which need to be filled, and that’s where Haven comes in. Haven’s purpose is to fill the gap between primary providers of mental healthcare such as psychiatrists and psychologists and their patients. Generally, a patient will see their provider once a week at the most. Our thinking was that it seemed insufficient to only have one day per week for a mentally ill person to process their emotions, as they need it more than anyone else.

What it does

Welcome to Haven! Haven is a mental health app which monitors the mental and emotional status of patients while compiling this data for easy access for medical professionals. Haven also utilizes cloud-based lexical analysis and keyword-spotting based off of Microsoft’s research on diagnosis of depression with Artificial Intelligence to pre-diagnose potential mental illnesses.

Haven is the patient's best friend, it helps patients reflect on their feelings and emotions while allowing them to convey them to therapists or a loved one when they're not around. It also empowers the therapist to make decisions on real data vs the patients' reflections and emotions. The journal/status functionality of Haven allows a patient to reflect upon their own emotions, and attempt to help themselves throughout the week. In addition, the therapists will be able to view this data through a webpage to monitor the status of the patient throughout the week. However, it also serves the second purpose of being another avenue via which the patient’s provider can gain a greater understanding of their mental and emotional well being through the lexical cloud-based sentiment analysis on the patient’s statuses. Lastly, we also perform keyword-spotting on the patient’s statuses to attempt to pre-diagnose mental illnesses and potential emotional problems that the patient may be facing.

How we built it

The client-facing application is built on the Android platform with Android Studio while the health professional facing app is built for the web with Google Graphs API and Bootstrap and is hosted on Github Pages. We chose Firebase as the database system most suitable for this application because as this project scales, our queries will need to obtain a large amount of information from individuals on our database; non-relational databases are advantageous in this regard because they are more efficient in querying data when they’re all clustered under the same child nodes. The sentiment analysis algorithm was built using Google Cloud Platform’s sentiment analysis tools. We also utilized NodeJS to write a script that analyzes the status entries made by users while scanning for important keywords for mental illness pre-diagnosis purposes. Lastly, we used python as the “glue” script that runs all the scripts and manages the interactions between the different applications, interfaces, database, and scripts.

Challenges we ran into

The biggest challenge that we ran into was determining how we were going to handle all the user data that we collected. After some debating, decisions, and testing, we decided on Google Cloud's sentiment analysis based off of user's text.

Accomplishments that we're proud of

We're really proud of our integration of Google Cloud's sentiment analysis tools as well as our own development of a word-spotting algorithm based off of Microsoft's research in Artificial Intelligence and how AI can be used to pre-diagnose mental illnesses. We're also really proud of our nice user interface for both the professional-facing web application as well as the client-facing Android application.

What we learned

We learned a lot about cloud-computing analysis tools as well as about analyzing raw data from users. We definitely think the biggest educational takeaway for Haven was how to come up with methods and ways to analyze data and make sense of all the raw data with both tools and algorithms we develop ourselves.

What's next for Haven

Our mission at Team Haven is to not just improve diagnosis of mental health and increasing efficiency of communication between patients and medical professionals, but also raise awareness of mental health problems while welcoming everyone to contribute to our project. The Firebase database key is open read/write and can be found in the HavenScript files.

We hope people can find more ways to analyze and make sense of user data to help them pre-diagnose mental illnesses while helping medical professionals stay aware of all the data provided to them.

Share this project: