Gait Lab

Inspiration

While a large chunk of the project comes from my personal experiences as an Electronics student, I had previously worked on something similar, a standard hardcoded MATLAB solution. That was the most direct form of inspiration that I had - I wanted to build something for user health. That, and a close friend's parents having a sciatic bending problem, prompted me to build something like this.

What it does

I built a system that can monitor your gait movement and classify between natural and unnatural. That, along with a posture analyser for the detection of sciatic bending.

How I built it

I built the initial DSP codes in MATLAB, converted them to Python to process, used scikit-learn, and upon acquiring real-time datasets of both unnatural and natural walking, trained my Random Forest classifier model upon it. I used API bridges for LLM integration (Google Generative AI, 3.7-Flash).

Challenges I ran into

As interesting as it may sound, I ran into a load of problems! First, I faced problems with a broken frontend component, and also a problem with wrong classification! I debugged them live, and built around it!

Accomplishments that I'm proud of

Beyond building a hackathon project that "works", this project serves actual audiences. I didn't build it just for the pitch - I can improve and give it to my friend's parents to use! I wanted to build something for the user. The commons.

What I learned

I learnt so much in this creative journey. I learnt how solutions like these can meaningfully transform lives. On a personal level, this project was completely done solo, so I effectively learnt how working alone can actually push me harder to work on this solution. I learnt to do design.

What's next for Gait Lab

I'm actively working on the code! After the review, Gait Lab will see 2 major transformations:

1. A new Flutter app for mobile. I believe that really is essential for me to move forward.

2. Integrating the Phyphox part into the app. While it's a 2-click process to record from Phyphox, I am planning to bring the recording of X and Z accelerometry data to the app. While it makes measuring of gait easier, it is essential for the continuous monitoring of posture and giving relevant alerts!!

Built With

• butterworth
• dsp
• filters
• javascript
• python
• randomforest
• react
• uvicorn