Application of Machine Learning Techniques in Hernia Classification
Dan Kipkosgei Kogei *
Department of Mathematics, Physics and Computing, Moi University, Kenya.
Subby Mino Mackenzie
Department of Mathematics, Physics and Computing, Moi University, Kenya.
*Author to whom correspondence should be addressed.
Abstract
Machine learning is a branch of artificial intelligence (AI) concerned with designing systems that can learn from data and make predictions or decisions without explicit programming. The rapid progress in machine learning in recent years has been attributed to advances in computing capacity, the growth of large-scale datasets, and the development of novel algorithms. A hernia is a medical condition that occurs when an organ or tissue pushes through a weak spot in the surrounding muscle or connective tissue. Hernia develop due to a combination of muscle weakness and strain. Some common causes include congenital weakness, where some individuals are born with weak abdominal muscles, making them more prone to hernias. Accurate and early classification of hernia is crucial for effective treatment planning and management. Machine learning (ML) techniques have been increasingly applied in medical diagnosis due to their ability to identify complex patterns in clinical data. However, existing studies on hernia classification have been limited in their methodological approaches, that is, most studies have utilized the k-Nearest Neighbors (KNN) algorithm for hernia classification leaving other robust classification algorithms. The study used publicly available secondary data obtained from the University of California Irvine (UCI) machine learning data repository with the sample size of n=310 observations. The objective was achieved by applying K- nearest neighbor, the artificial neural networks, the support vector machines, random forest and the naïve bayes algorithms to the hernia data obtain from the UCI machine learning repository. For each method, data was splitted in the ratio 70: 30, that is, 70% of the data was used for training while 30% of the data was used for testing. KNN was optimized when k=6, while the random forest technique outperformed all the algorithms. This study recommends that the supervised random forest algorithm should be used while classifying hernia conditions because of its high accuracy, sensitivity, specificity and kappa values for proper medical diagnostics and treatment. Using the hybrid machine learning models such as RF-SVM, ANN-SVM, RF-NBC and deep learning methods could be possible future research areas.
Keywords: Hernia Classification, artificial intelligence, complex computing work, Machine learning classification, learning patterns