Today, over 40% of Americans suffer from obesity. Obesity and excessive weight have been associated with severe risk of illness (CDC). We find this problem especially relevant given the current COVID-19 pandemic, as many people have found themselves confined indoors with little to no exercise. By providing a predictive model to gauge the impact a person's lifestyle has on their health, we hope to aid in preventing health risks that arise from obesity and excessive weight.
What it does
O-Test allows users to interact with a trained predictive model to determine their obesity risk factor.
How we built it
This project was created with a wide range of technologies. The frontend was developed using React, Firebase, Axios, and Material-UI. The backend consists of a REST service created with Flask and hosted on Google Cloud Platform. Our ML predictive model was crafted using NumPy, Pandas, and Matplotlib with a dataset acquired from Kaggle.
Challenges we ran into
Training our model proved difficult due to the limited amount of data we had access to as well as the necessity of considering many factors that may be associated with obesity. We also decided to challenge ourselves by using technologies that we were largely unfamiliar with. A lot of our time was spent experimenting with Pandas to try to build as accurate of a model as we could. User authentication was also difficult to implement, as was setting up the hosting services for our backend.
Accomplishments that we're proud of
Our team had a diverse background that allowed each of us to specialize in different parts of our project. This allowed us to cover each other's weaknesses and maximize our productivity and knowledge gained. We were especially proud of our ability to work together and bring together each of the different parts of the project that made up the cohesive whole. In particular, we are proud of our mostly polished frontend design as well as our model's surprising accuracy.
What we learned
Besides the new frameworks and tools that we picked up for this project, we learned to work together as a team and to incorporate our wide range of skills to provide a solid product.
What's next for O-Test
We hope to develop a more accurate model with cross-validation in the future. We would also like to smooth out the frontend even more to give users the best possible experience.