Data Pre-Processing Technique for Enhancing Healthcare Data Quality Using Artificial Intelligence

Abstract

Healthcare datasets frequently contain large dimensional, distorted, uneven, missing, and imbalanced data. These difficulties may lower the effectiveness of machine learning algorithms. Before using machine learning algorithms for healthcare datasets, pre-processing is necessary to ensure the data is adequate for learning. The data pre-processing is essential to improve the performance of classification or prediction. This paper proposes a data pre-processing technique for enhancing healthcare data quality using artificial intelligence. The pre-processing includes handling missing values, outlier detection and handling imbalanced data. The missing values are imputed using the KNN-based approach, the outliers are detected using a cluster-based algorithm, and SMOTE and the Random resampling approach can rebalance the imbalanced data. Different machine learning classification algorithms are used to analyze the data quality. The real-time healthcare dataset is used to evaluate the performance of the proposed approach using accuracy, sensitivity, specificity, precision and f-measure. This research shows that the pre-processing techniques chosen have a considerable positive impact on the model's performance when comparing the model's efficiency with and without pre-processed data.