Preliminary Introduction and Implementation of Novel Machine Learning Algorithm Utilising Pareto Principle: Classification of Small Biomedical Health-Related Datasets

Machine learning algorithms are usually developed to reveal useful insights from data resources, such as selecting appropriate treatments in healthcare. However, focusing on the “more is less principle” is essential as the algorithms are complex to implement and the data-driven perspective is challenging and overwhelming for users. The objective of this study is to introduce a comprehensive and accurate novel machine learning algorithm, the Pareto Principle, using Multi-objective Optimisation (MO) and Always Better Control (ABC) analysis. Many health-related datasets, such as immunotherapy and diabetes, are used as case studies, considering the broader classification application of this algorithm in several health domains. The achieved classification results are compared with related research publications. Utilising Pareto’s 80/20 Principle, area and time are found to be the most relevant attributes in the case of the immunotherapy dataset. Pareto Principle is recommended as it is effective, requiring minimum resources and memory in the initial stages of classification before algorithm deployment.