Cardiovascular disease prediction

• 

  cardiovascular disease

Inspiration

I have been passionate about Machine Learning and I found this hackathon as an opportunity to try out my skills. Cardiovascular diseases (CVDs) are the leading cause of death globally, taking an estimated 17.9 million lives each year , according to sources. Thus it is very important to treat them at the earliest stage and cure them with the help of medicine and technology that is available.

What it does

It is a simple Machine Learning model which predicts if a person can have cardiovascular diseases based on a number of attributes. Giving these values into the predictor, it predicts if a person is at the risk of this disease.

How we built it

As a Machine Learning model, it tries out Decision tree first. But Random forest model gives better accuracy of 74%. Exploratory data analysis gives further light into the factors determining it, such as diastolic and systolic blood pressure.

Challenges we ran into

Determining which attributes are more important for the model was difficult. Exploratory Data Analysis helped to shed light into it and helped to find the trends.

Accomplishments that we're proud of

This system currently provides an accuracy of 74%. During initial analysis, a lower accuracy was expected. But using Random forest model increased the accuracy. This system helps us to predict the disease at an early stage. As a beginner in ML, I feel proud of the model created and how it can help the world in curing.

What we learned

We learnt about various classifying models in Machine Learning. The best model was used for creating this system. While determining the model to use, it was observed that the decision tree model gave a lower accuracy than the random forest model.

What's next for Cardiovascular disease prediction

We have to increase the accuracy. The False Negatives are a risk to our system. We intend to decrease them to the lowest levels. Further trials and learning might also predict a better model for our problem.

Built With

• eda
• machine-learning
• pandas
• python
• seaborn