Mapping Knee Osteoarthritis Biomechanical Severity Prior to Total Knee Arthroplasty Using an Unsupervised Learning Framework

Description

In this presentation, Kathryn Young introduces a study titled "Mapping Knee OA Biochemical Severity Prior to TKA Using an Unsupervised Learning Framework." The study explores the effectiveness of total knee arthroplasty (TKA), noting that despite its success, patient satisfaction rates are disappointingly low at around 80%. Young discusses the variability in patient outcomes post-TKA, linking it to differences in demographics, health status, and severity of osteoarthritis (OA). The research utilizes a sophisticated data mining framework to analyze a data set of 936 observations from 497 knees, spanning the OA continuum from asymptomatic to post-TKA situations.



Four patient groups are classified: asymptomatic adults, moderate OA individuals, severe OA candidates one week pre-TKA, and TKA recipients one to two years post-surgery. The study captures kinematic and kinetic patterns in gait using force platforms synchronized with motion capture systems. Through Principal Component Analysis (PCA), the researchers identify key modes of variability and apply a self-organizing map (SOM) to visualize clustering within the data.



Results reveal three distinct clusters representing asymptomatic, moderate OA, and severe OA patients. The SOM provides insights into how OA severity correlates with various patient demographics and gait patterns. Additionally, the framework allows for individual tracking over time, enabling clinicians to monitor patient progression and identify sudden declines, which could inform future interventions. The study emphasizes the potential of this approach in enhancing understanding of OA severity and improving surgical candidacy predictions.

DOI: 10.1302/3114-220769

Specialties

Conferences