APPLYING MACHINE LEARNING (THE K-MEANS ALGORITHM) TO CLUSTERING AND ANALYZING SYNOVIAL FLUID CONTENTS AMONG DIFFERENT AGES AND GENDERS IN HEALTHY AND OSTEOARTHRITIS PATIENTS

Abstract

Machine learning, a subset of AI, has made a significant impact on the medical field by improving the speed and accuracy of test results. Among the many discrete ML tools, k-means is a type of data clustering that uses unsupervised ML to divide unclassified data into different groups with similar variances. This dissertation applied the k-means clustering algorithm to analyze synovial fluid compositions of healthy people and osteoarthritis (OA) patients, focusing on four components: hyaluronic acid (HA), chondroitin sulfate (C6S, C4S), and the C6S ratio. The main objective was to identify distinct patterns and clusters within these datasets based on age and gender. Data was extracted from two previously published research studies. The first dataset comprised 187 healthy participants, with ages ranging from 10 to 90 years. The second dataset consisted of 133 OA participants with ages ranging from 55 to 90 years. Applying ML algorithms, specifically k-means clustering, the MATLAB program was used for data analysis. The findings showed the k-means clustering successfully highlighted age- and gender-related synovial fluid concentration patterns. In addition, for both healthy and OA groups, younger people had higher levels of synovial fluid components, which decreased with age. In healthy people, HA levels were high among younger people but decreased with age. In the OA group, HA levels increased in older patients. These findings confirmed the potential of synovial fluid concentration in diagnosing joint health. These findings also asserted the utility of ML techniques, such as k-means clustering, in medical data analysis.