Comparison of classification machine learning algorithms for damage detection in simulated total knee replacements

Total knee arthroplasty (TKA) is currently one of the most common orthopedic surgeries in the United States. While the surgery is generally highly successful, revision due to pain and failure is costly, and can have adverse impacts on patient outcomes. The possibility exists to reduce rates of catastrophic failure by early detection of damage in total knee replacements (TKR). Previous work has been done to establish the ability of a structural health monitoring (SHM) technique known as the electromechanical impedance (EMI) method to detect certain types of damage prevalent in TKRs. In the previous work, 19 simulated TKRs were constructed and artificially damaged, impedance spectrum measurements were taken, and healthy and damaged data was compared to determine if significant differences between these impedance responses exist. The current study expands upon the previous work by exploring classification machine learning (ML) techniques to translate the differences in impedance responses into discrete damage classes. The goal of this work is to determine ideal classification technique(s) for identifying and classifying damage within the aforementioned TKR systems. To this end, several algorithms are trained on the aforementioned impedance data, and the results of a leave-one-out crossvalidation scheme are compared for accuracy, among other common ML performance metrics.