Cardiovascular Disease Prediction using Various Machine Learning Algorithms

Abstract: Almost one-third of all deaths caused around the world were
caused due to cardiovascular diseases. Even if death was not the result, much
cost is incurred during the treatment of such diseases. But much of these
deaths and treatments could have been prevented with prior action. Advance
knowledge of the symptoms and consequently proper care can lead us to
avoid such diseases. Thus, current research proposes a highly effective model
to predict the presence of heart diseases. Bad eating habits, smoking, stress,
and genetics are some of the factors that influence our body mechanisms,
which actually cause various irregularities in our hearts and thus adversely
affect our bodies. The body mechanisms influenced by external factors have
been included to prepare an efficient model to predict the probability of
cardiovascular diseases. UCI repository dataset has been utilized for the
training and testing purpose in our model. Then accordingly, five different
algorithms namely Logistic Regression, Support Vector Machine, MultiLayer Perceptron (MLP) Classifier with Principal Component Analysis
(PCA), Deep Neural Network, Bootstrap Aggregation using Random Forests
are executed on our filtered dataset to find which one is the optimum out of
all of them. Pre-processing techniques have been extensively used to filter
out the dataset. The data processing along with the different models
employed make this a sound paper, which could be utilized for real-world
cases without any prior modification. Different places around the world
would take different factors into account, hence our model can be used as it
takes all critical factors from several datasets.