A Quiet Project About Loud Feelings

🧠 Mental Health Prediction Model:

Some projects start with a spark of genius.
This one started with a slow, tired conversation and a random comment that hit deeper than expected.

So I was in a pre-admission applications meeting with a university in Europe, discussing the language requirements and all that, when one of the admission panel members mentioned how a lot of international students quietly fall apart. They’re juggling language classes, coursework, jobs, time zones, expectations… and homesickness that hits differently when your family is thousands of miles away.

That moment just stuck with me, not in a dramatic way but more like a soft echo.
I kept thinking about it.

Because honestly I’ve had loads of moments like that too. You wake up heavy. You’re trying your best, but you’re barely holding it together. Nothing looks wrong, but you feel off. And you keep going, because you think everyone else has it figured out.

And that’s how this project came to be, and luckily for me someone (BekBrace) on YouTube had already made a video on how to go about creating it.

It’s called the Mental Health Prediction Model, and it’s exactly what it sounds like: it’s an app that uses your input to predict potential mental health risks using machine learning. It doesn’t perform diagnoses or offer magic solutions. It’s just a tool. A quiet nudge. Like your mind saying, “Hey, maybe check in with yourself.”

💻 The Build (Tech Side)

I started with the basics:

• A virtual environment, requirements.txt file with dependencies:
  • FastAPI, Streamlit, scikit-learn, pandas, numpy, SQLAlchemy
• Built the backend with FastAPI
• Frontend runs on Streamlit

To run the app:

python main.py

streamlit run src/frontend.py

It’s lightweight and local and i also have it online using free hosting resources, which can take a while to boot up but the foundation is solid  and scalable.

 

🎯 Why I Built It

This isn’t about how well I can code or replicate existing projects. It’s about building and collaborating to build something that makes life a little easier for someone, somewhere.”

A tool for those moments when you’re not “sick” but you’re not okay either. When you’re just… slowly unraveling under expectations that no one sees. Especially for students and immigrants navigating new systems, new cultures, and new versions of themselves.

The idea is simple:
If we can flag mental strain early, not dramatically, just gently, people can get the right kind of help sooner. It reduces the weight on counselors, shortens long waitlists, and makes space for people to feel seen even before they ask.

“If you are distressed by anything external, the pain is not due to the thing itself, but to your estimate of it.” – Marcus Aurelius

We don’t always need saving. Sometimes, we just need a signal. This model is one small way to send that signal earlier. this is a link to the gitbub repo  Newbie4newthings/Mental_health_prediction: This is a ML Mental health prediction basic model

 

🧭 What’s Next?

This project was built on dummy/test data. The next step is finding real-world datasets to work with. I’d like to:

• Analyze actual patterns from student or workplace well-being reports
• Train the model on broader datasets
• Collaborate with institutions or orgs looking for lightweight mental health support tools

This could be used in schools, remote work platforms, community centers, or anywhere people are quietly burning out.

If you’re reading this and you know someone working in education, HR, or mental health who might be interested, feel free to connect with me.

 

🌱 Final Thought

I’m not a therapist. I’m not even sure I’m always okay.
But I know how easy it is to pretend to be okay. And I know how powerful it is when something or someone reminds you to slow down, breathe, and check in with yourself.

That’s all this app is. A soft reminder.
A mental health mirror. A digital pause button.

No capes. No noise. Just code with care in it.

“He who has a why to live can bear almost any how.” – Nietzsche