This study looks the different physical activity patterns in patients after knee or hip replacement surgery. The researchers utilize wrist-worn accelerometers from the UK Biobank to track the daily step count in 237 patients for a period of 4-12 months after the surgery. They used machine learning techniques called k-means and PAM clustering to pick out different groups of patients. They were classified into a high performing group that had an average of over 10,000 steps per day and a low-performing group that had an average of less than 6,000 steps per day. These groups had big differences in multiple ways. Patients that were in the high-performing group were younger, had fewer comorbities, and a lower BMI and were more likely to have a total hip replacement  and not a total knee replacement. Additionally, patients in the high-performing group had higher educational attainment and lower deprivation index scores, showing that socioeconomic factors also played a role in recovery. The two clustering methods (k-means and PAM) showed weak but positive agreement with each other, which means they generally identified similar patterns. The discussion emphasized that recovery after joint replacement surgery is influenced by many different factors, not just physical characteristics like age and BMI, but also by social and behavioral factors like education level and alcohol consumption, which highlights the importance of taking a personalized approach to post-operative care.