As cost-effective measures become increasingly implemented in the US healthcare system, changes in patient-reported outcome measure (PROM) scores can be utilized to indicate patient satisfaction following procedures including total knee arthroplasty (TKA). The primary aim of this study was to develop and evaluate machine learning algorithms to predict achievement of the minimal clinically important difference (MCID) for the Knee Injury and Osteoarthritis Outcome Score-Physical Function Short Form (KOOS-PS) at 1-year following TKA. A retrospective review of primary TKA patients between 2016 and 2018 was performed. Variables considered for prediction included demographics and preoperative PROMs. The KOOS-PS MCID was calculated via a distribution-based method. Five machine learning algorithms were developed and tested by discrimination, calibration, Brier score, and decision curve analysis. Among the 744 patients who met the inclusion criteria, 385 (72.8%) patients achieved the MCID. The elastic-net penalized logistic regression model was selected as the best performing model (c-statistic 0.77, calibration intercept −0.02, calibration slope 1.15, and Brier score 0.14). The most important variables for MCID achievement were preoperative KOOS-PS score, preoperative VAS Pain, preoperative opioid use, preoperative PROMIS global mental health score, age, and sex. Algorithms were incorporated into an open-access digital application available at https://sorg-apps.shinyapps.io/tka_koos_mcid/. This study is the first to predict the probability of achieving the KOOS-PS MCID following TKA using a machine learning-based approach. The results were used to develop a clinical decision aid based on commonly collected predictive variables to preoperatively predict an individual patient's likelihood of attaining an acceptable outcome following TKA.