Inspiration
Our project was inspired by a shared family history of heart disease and heart health challenges. Despite the numerous modern advancements in cardiovascular science, we realized that prevention remains the best cure for heart-related issues. This idea, combined with our personal experiences, fueled our motivation to create a project that could assist in the early detection and prevention of heart disease. We're hoping to make a meaningful contribution in this area through ArteryIQ.
What it does
This application gathers input from the user regarding their basic health, health history, and dietary habits. The project then uses a Machine Learning Algorithm built using NumPy, Pandas, MatPlotLib, and Scikit-Learn, written in Python, to predict with an accuracy rating of 90% if the user is at risk for Heart Disease. The Machine Learning Algorithm was built on a dataset with over 250,000 entries, retrived from Kaggle.com and known as the Behavioral Risk Factor Surveillance System (BRFSS), a health-related telephone survey that is collected annually by the CDC. The Algorithm is trained on data from 2015.
How we built it
We began by developing the core of our project in Jupyter using Python. Our tech stack included Matplotlib, Pandas, Scikit-learn, and NumPy to implement and train the machine learning algorithm. After finalizing the model, we converted the Jupyter Notebook into a Python executable. Next, we debugged the backend, integrated the script with a data submission interface, and connected everything with the frontend. The end result is a fully functional application capable of processing data and offering insights into heart disease risk based on our model.
Challenges we ran into
The biggest hurdle we faced was the time constraint. With less than 48 hours to understand, synthesize, and apply an huge amount of information, we were racing against the clock. As relatively new developers, we were learning a lot of things for the first time, which added to the pressure to produce. Selecting the appropriate machine learning algorithm and optimizing it for our dataset was a particularly tough challenge, requiring us to experiment with various approaches.
Another significant challenge was piecing together the different components of our project—getting the backend, frontend, and the machine learning model to communicate effectively was complex and took us a while to connect everything together. However, overcoming these obstacles made the final success taste even sweeter.
Accomplishments that we're proud of
We're super proud to have completed our project and to achieve a 90% accuracy rating, especially with our limited data and time.
What we learned
Throughout the process, we gained so much knowledge about full-stack development. Both of us were relatively new to intensive hackathon environments, and this was our first experience building a full application in less than two days. We learned about machine learning algorithms, learned how to clean datasets, and integrated Python scripts to create executable programs. We also learned about frontend development, backend processes, and connecting everything together. This hackathon was a steep learning curve, but it gave us exposure to a wide range of technologies and techniques.
What's next for ArteryIQ
Hopefully we can take this application further and turn it into a proper piece of software, with higher accuracy rating and more data to train our algorithm on!
Log in or sign up for Devpost to join the conversation.