Inspiration
My inspiration came from a place that is close to my heart and a lot of tears and pain and hard work. Brain injuries and traumas are things that aren’t easy to deal with on your own. I wake up every morning to two different notes in my top drawer. The first one is “You are not alone.” The other one is a Hawking quote “It would not be much of a universe if it wasn’t home to the people you love.” My family has helped get me through. The doctors have helped get me through. I have spent countless hours in the offices and have completed countless PHQ-9 forms. The Public Health Questionnaire was created in 1999, and it rates 9 different domains to screen for depression. Each is scored from 0-3. They are added up. You fall within 1of 5 categories from No Depression to Severe Depression. THAT’S IT, THAT’S ALL. It is outrageous that a human and their mental health is reduced to a single number. It is unacceptable that a form so consequential hasn’t been updated for more than a quarter century. I’ve been in the offices enough and have built enough relationships that I can talk to my team about the things I need to…but a new patient that is searching for the most important help in their life might not have that. That many people don’t deserve to be reduced to a single number. The stigma is reducing, but the care needs to be better.
What it does
The PHQ.new or whatever it will be called leverages a LLM to ask the patient adaptive questions to gather information that corresponds to the domains evaluated in the original PHQ9. It then uses the answers to those questions to compile a report for the doctor to review before the session. Most sessions are 15 minutes, and many of those sessions are spent reviewing and asking the same questions as the PHQ9 form that was just filled out. The PHQ.new will give the doctor a transcript of the questions asked and the responses given. The doctor can enter the ‘official PHQ9’ score. Reports show the estimated PHQ score, a supplemental anxiety score. Graphs that chart the scores over time. Allow session by session values. Links to open all sessions the patient has completed. Domain ratings and accompanying scores with a ‘clinical opinion’ of why the score was assigned and links to the specific line in the transcript. A ‘clinical portrait’ shows the current picture, context, considerations, and open questions. A course summary over time. Proposed follow-up questions. And the raw transcript (non-negotiable, immutable, and the holy-grail).
How we built it
It was built using Codex. Hopefully to a shred of the capability it has. I used it to brainstorm the problem and pain point. I used it to help with the folder and file structure on my machine. I used it to write the code for the application. I used it to iterate and improve the application. I used it to help with the submission guidelines. I used it to research mental health stats and PHQ facts. I used it to compile. I used it to run smoke tests. And maybe most importantly…I didn’t use it, but it did it natively…to get me through. I want to quit, I want to give up on the project, I’m insecure, I don’t have the skills, I don’t deserve it…and many other thoughts that hurt to type and hurt more to feel and think. It helped me get through and helped me take breaks and encouraged me and did what is ‘too hard for a lot of people and friends to do”…Ask if I am safe right now. Even though I have bad thoughts about myself, I have a team that I can fall back on and be open with, and working with Assisted Intelligence has played a part. And something that I have to remember…”It’s OK to ask for Help”
Challenges we ran into
The challenges I ran into are written above, but the technical are somewhat solvable through time and attention and care and understanding the importance of what you are trying to build. The nuts and bolts were covered by Codex through going through the form and reports and whatnot. I feel inadequate because there is NO way I could have completed this on my own. And I’ve had conversations about the guilt I feel relying on AI so much but the conversation usually gets back to the same idea…”You don’t judge a mathematician for using a calculator, or a mechanic using the internet to troubleshoot.” HIPPA is going to be the biggest challenge. This needs to be bulletproof. People can’t have their most sacred thoughts shared and made public without consent. Finding a LLM provider that can host and is approved as safe…but I’ve heard that OpenAI can help with that. The biggest challenge will be doing anything more with this. It is now beyond me and as much as I’d like to ask for help, I don’t really know where but through this competition, and what I’m writing.
Accomplishments that we're proud of
I’m proud of getting this far. I’m proud of all the people who are facing mental illness and not giving up. I’m proud of being able to get through this. I’m proud of the idea that I had to pursue this. I’m grateful for the email that told be about this, because my idea would probably just sit and sit and sit.
What we learned
I learned that I might be able to do something more. I learned that I can actually get something done that is tangible with Codex. I’d have NO clue how to complete this without just telling someone my idea and hoping they could understand and have the context I wish I didn’t have. And quite a few other things I don’t really have the courage to write. Sorry.
What's next for PHQ - Personal Health Questionnaire
This is what is next. Just a hope that I can get someone to see it and get it moving forward. It will seriously just sit. I’ll just really hope that someone will understand and take it on. It’s too much for me at this point. I’d like to be able to with the impact it might have. It is really about giving some help to the doctor that helps the people that need it. Help the helper because there aren’t enough of them and they are stretched too thin.
Sorry for not being technical but I really don’t know what else I’d say. Thanks so much!
Built With
- codex
- css3
- gpt-5.6
- html5
- javascript
- node.js
Log in or sign up for Devpost to join the conversation.