Subsampling Supported by Artificial Immune Systems to Improve Data Balancing in Health Settings

Data imbalance presents a significant challenge in analyzing and classifying datasets, especially when it comes to subsampling techniques. One-Sided Selection (OSS) is a common approach, however its effectiveness may be limited due to the inefficient removal of redundant examples from the majority class. This paper proposes an alternative solution, called Subsampling Supported by Artificial Immune Systems (SSAIS), which incorporates principles of detecting patterns from artificial immune networks for a more informed selection in the OSS. Dealing with unbalanced data remains an ongoing challenge in the healthcare field, for this reason, this article uses a dataset related to diabetes management as an example to illustrate the effectiveness of the proposed methodology. The results corroborated the superior efficiency of the proposed method compared to the traditional OSS in eliminating redundancies