Reflect

• 

  dashboard - access all functionalities from here!

• 

  image -> text - digitized journal log

• 

  seeking help - locate mental health professionals in your area

Inspiration

The single ask of this hackathon was to build something that benefits people in the real world using real data. As students, we've seen how hard it can be to ask for help, and barriers to mental health access can delay and prevent individuals from seeking help. We think this is especially true for new students who are just starting their journey in college and still figuring out what resources are even available to them. We built Reflect to address that: a toolkit for tracking your mental health, processing your thoughts, and finding professional support near you when you're ready for it.

What it does

Reflect is a mental health web app designed to feel personal and intuitive. You can scan your handwritten diary entries and have them transcribed to text instantly using AI, so your thoughts can be digitized and saved in the app. From there, you can track your mood over time, write and save journal entries, and set personal milestones to celebrate your progress.

When you're ready to seek professional help, Reflect allows you to search for mental health facilities filtered by driving distance from your location, so whether you're new to the area or just don't know where to start, finding care and the mental health specialties offered near you is just one search away.

How we built it

The frontend is React + Vite with a tab-based single-page architecture. The backend is a Node.js Express API.

The "Talk to someone" feature pulls from the SAMHSA National Directory of Mental Health Facilities, a 9,000+ row federal dataset. We geocode user location and facility addresses with the Google Maps Geocoding API, then calculate both straight-line distance and driving distance via the Google Maps Distance Matrix API, so users can filter by how far they're actually willing to travel.

The “Scan a Diary Photo” feature lets you upload a photo of your handwritten journal. Our AI reads your handwriting and converts it to text. The transcription is saved in Supabase, so you can keep all your journal entries organized over time. You can also type entries directly. Both typed and scanned entries are stored in the same place for easy access.

AI assistance was used to help implement a few parts of the code and to solve some technical issues.

Challenges we ran into

Integrating three external APIs, especially Google Maps, and SAMHSA for finding therapists nearby was difficult. For the data analysis portion of our project, we also had trouble decoding column names in the NSDUH dataset and selecting the best model (random forest model) for predicting major depressive disorder with tight turnaround time.

Accomplishments that we're proud of

Building a tool which pulls from real federal mental health clinic data, incorporates AI for image to text processing, and allows for making the first step towards seeking mental health resources felt meaningful.

What we learned

• Lock scope early. The anonymous session system we ended up building was a great call because it saved us from a full auth rabbit hole.
• Things always take longer than expected, so it's important to leave enough time for presentation building and the dev post at the end!
• Divide and conquer -- decide who is going to do what, and make use of the whiteboards at Morgridge!

What's next for Reflect

We'd love to expand the SAMHSA dataset filtering by insurance type, specialty, or availability. We would also be interested to learn more about data privacy for health apps. We would also like to further build out the mood tracking data to help users analyze their health trends over time.

Built With

• api
• express.js
• google-maps
• node.js
• react
• supabase
• vite