Classification of ECG beats using optimized decision tree and adaptive boosted optimized decision tree

Rapid advancements in our innovation have encouraged early analysis of diseases in the clinical area. The primary diagnostic tool for detecting cardiovascular diseases is electrocardiogram. We have proposed two classifiers for classifying six types of heartbeats. Among these six types, one is normal, and the other five depict cardiac arrhythmia beats. For tackling complex order issues, decision trees are a notable methodology in all artificial intelligence procedures. The standard decision tree algorithms are unable to process imprecise, uncertain and incomplete data. Optimizing hyperparameters helps to overcome these flaws. Hence, an optimized decision tree and adaptive boosted optimized decision tree are proposed to deal with the data with uncertain class labels and attribute values. Moreover, we consider the instance where the uncertain data are characterized and accomplished over the evidence theory. The proposed method is validated and tested on MIT-BIH arrhythmia database. Our proposed model is effective and promising and outperformed all state-of-the-art techniques with an accuracy of 98.77%.

Automated summary

This study explores early analysis of cardiovascular diseases in the clinical area and proposes a decision tree method that can effectively classify heartbeat types. The research uses optimized hyperparameters to overcome the challenges posed by uncertain data and ultimately validates and achieves an accuracy of 98.77% on the MIT-BIH arrhythmia database.