Using unsupervised machine learning to predict quality of life after total knee arthroplasty.

BACKGROUND: Patient reported outcome measures (PROMs) are an important metric to assess total knee arthroplasty (TKA) patients' global health. The purpose of this study was to use a machine learning algorithm to identify patient features that impact PROMs after TKA.

METHODS: There were 636 TKA patients included in the study. The patient data was retrospectively reviewed from patients enrolled in the University of Ottawa, Division of Orthopaedic Surgery patient database between 2018 to 2022. Their mean age was 68 years (range, 39 to 92), 56.7% women, who had a mean BMI of 31.17 (range, 16 to 58). Patient demographics and the Functional Comorbidity Index were collected alongside Patient Reported Outcome Measures Information System Global Health v1.2 (PROMIS GH-P) physical component scores measured pre-operatively, and at 3-months and 1-year after TKA. An unsupervised machine learning algorithm (spectral clustering) was used to identify patient features impacting PROMIS GH-P scores at the various time points.

RESULTS: The algorithm identified five unique patient clusters. The groupings varied by demographics, comorbidities, and pain scores. The patient clusters were associated with predictable trends in PROMIS GH-P scores at the 3- month and 1-year post-operative time points. Notably, patients who had the worst pre-operative PROMIS GH-P scores (cluster 5) had the most improvement after TKA, whereas patients who had higher global health rating pre-operatively had more modest improvement (clusters 1, 2, and 3). There were two out of five patient clusters (cluster 4 and 5) that showed improvement in PROMIS GH-P scores that met a minimally clinically important difference at 1-year post operative.

CONCLUSION: The unsupervised machine learning algorithm identified patient clusters that had predictable changes in PROMs after TKA. This algorithm can be applied to other data sets or used as a baseline from which to measure the change of an intervention on this single center's TKA population. It is a positive step towards providing precision medical care for each of our arthroplasty patients.