-
Poster
Inspiration
More people are suffering from cardiovascular disease, so I want to make a model that can help people predict their risk of having cardiovascular disease and get examination if they have a high risk.
What it does
It studies 60000 cases of people having and not having cardiovascular disease and generates a model that can predict whether or not a person has that disease.
How I built it
I compared different models and choose the one that has the highest accuracy.
Challenges I ran into
The ANN model I built at first suffered overfitting, the difference in the accuracy of training and testing is always between 2%, not as good as the example I saw on slack. Then, I preprocessed the data, standardized and normalized them, and made some changes to my model to address the overfitting problem. I changed the epoch number and add more patience, then the model is much better.
Accomplishments that I'm proud of
Increase 1.5% of the accuracy of the ANN model (spent a lot of time on it ). Employed many models to compare their fitness to this problem.
What I learned
How to deal with the data, like deleting the outliers and duplicated data; how to show the model visually
What's next for Cardiovascular Disease Prediction
The ANN model can be improved to reach 72 percent accuracy.
Log in or sign up for Devpost to join the conversation.