Predictive Analysis of Imbalanced Cardiovascular Disease Using SMOTE

Abstract

Throughout the human life, health care is an inherent activity. The spectrum of cardiovascular disease (CVD) affecting the heart and nerves is a common description. Early cardiovascular infection identification techniques led to the resolution of decisions on progressions in high-risk patients which reduced their dangers. The health care industry contains bunches of clinical information, subsequently machine learning evaluations are required to settle on choices viably in the forecast of heart illnesses. Of late research has dove into joining these methods to give cross breed machine learning calculations. The data collected was a skewed distribution of data i.e majority samples (positive) are dominating minority samples (Negative). The dataset consists of 164 samples with positive class and 139 samples with negative class. The major challenging task is developing a model with skewed dataset and producing accurate results. So, in this research Synthetic minority oversampling technique (SMOTE) is used by the researchers to balance the positive and negative samples of the dataset. After applying SMOTE on the given dataset train set and test set divisions are performed for various classification techniques. In this research, researchers are also adding new test records to the training set, with this each new entry the model becomes more intelligent and ensures accurate results. In addition to this our undertaking proposes an expectation model to foresee whether the individuals have a CVD or not and furthermore, it Figureures the hazard level of the individual to get influenced via cardiovascular ailments and play it safe to dispose of it.