How we built Carebot
Carebot is a Facebook messenger bot that helps college students find therapists and mental health specialists under their insurance plan, built by my buddy Matthew Zhang and me. Using a conversational and friendly tone, Carebot helps narrow down your search by insurance plan, distance, gender and language preferences. We built it over a weekend at a hackathon in April hosted by ProPublica, Better Doctor and Yelp. The challenge was: how can we create and identify prototype tools to help patients access health care from high-quality, affordable providers? And in your solution, how can you use the Yelp, Better Doctor, and ProPublica APIs?
Matt and I are both juniors at Northwestern, where mental health is a big issue on our campus and many others. In the 2016–2017 school year alone, two students committed suicide. For a while, Northwestern’s counseling and psychological services (CAPS) only gave students 12 free sessions of counseling. Thankfully, there will no longer be a limit in the 2017–2018 school year. But now CAPS is the only resource students have on campus. There used to be free counseling available at the Women’s Center but the university eliminated the service last year.
How in the world, then, are young students struggling with mental health for the the first time supposed to find professional help? Especially when financial resources are limited? Well, that’s where we hope Carebot comes in. Matt and I first spent a couple of hours at the hackathon calling our friends and conducting user interviews. Here are some main findings we found:
College students aren’t really looking to find a primary care doctor or a physician on campus. If they look for a doctor, it’s usually for mental health reasons
Financial costs are a huge barrier to entry. Students want to find a mental health specialist within their insurance plan or their parents’ insurance plan
Gender, location, and reviews are really important in making your decision
It’s scary and overwhelming to filter what’s best for you on your own
There’s no easy way to determine the “quality” of the doctor. Yelp reviews are helpful
Students don’t really know which questions to ask when they call for a consultation.
These user research interviews were formative in helping us revise our “how might we” statement and the problem we were trying to solve. This part included lots of post-it design jams. We finally narrowed our question from this:
How might we help patients access health care from high-quality, affordable providers?
How might we use familiar technology to help college students find the right mental health specialist near them?
We decided to make a Facebook Messenger bot because a recent Pew Research study showed that 88 percent of 18–29 year olds are active on Facebook. We also dived into Facebook’s Developer docs and thought that the bot offered great ways to narrow down a user’s search with features like buttons, location finder, and the ability to directly call an office once you’ve found a good match. It was a great balance between familiar technology meets easy call for action.
Afterwards, we wrote a sample conversation and learned how to make a bot in Node. We used the Better Doctor API to find psychologists and psychiatrists by insurance plan and distance, and then the Yelp API to get any reviews. Here’s a little snippet of what the final version looked like:
One feature that I was really excited about was the last message Carebot gives you, which is a button that makes it easy for the user to make that leap and call the office, and three sample questions that are good to ask before you schedule an appointment. These questions were absolutely inspired by the conversations we had with our friends about their experiences and what they wish they knew. By making the barrier to entry as low as possible, we hope that users feel confident and empowered enough to take action by the end of the conversation.
Our prototype ended up winning the $1,000 cash prize from Yelp for best use of their API and the award for best use of the Better Doctor API. But the best part was the feeling that we designed a prototype bot that understood the needs of a young college student. I kept thinking like “We didn’t win today. User empathy won today.” And it’s so true. The time we spent on the phone with our friends and listening to their experiences to revise our “how might we” question absolutely paid off. It was also super validating to see that Matt and I could build something from ideation to prototype, even when we’re the youngest ones at a hackathon with the least amount of professional experience (we are also ~pleasantly~ awkward).
Thanks for reading!