•  
  •  
 

Bone Bulletin

Abstract

Introduction

Wearable sensor technology and smartphone based care platforms have been identified as promising technologies for postoperative care after joint arthroplasties.1,2 In fact, in lower extremity arthroplasty, these technologies have already demonstrated their effectiveness in a multitude of randomized trials.1,2 One recent multicenter RCT with 401 patients undergoing primary total knee arthroplasty (TKA) showed that self-directed rehabilitation using a smartwatch and mobile application (mymobility with Apple Watch, Zimmer Biomet) achieved comparable range of motion and patient-reported outcomes measures (PROMs) to standard post-operative rehab without increasing adverse events.1 Related evidence also indicates that “episode of care” smartphone applications can meaningfully reduce in person healthcare utilization without apparent short term adverse effects.2 In one RCT, patients who used a care application on their smartphone following a hip or knee arthroplasty had significantly fewer postoperative physical therapy sessions (60.6% vs 94.4%) and emergency department visits (2.5% vs 8.2%) than controls, while simultaneously attaining similar functional outcomes at 90 days.2 In aggregate, current data supports the notion that digital rehabilitation guidance and remote monitoring can, in carefully selected patients, replace some traditional clinical visits.1,2

While data supporting the use of these technologies in hip and knee arthroplasty continues to grow, the evidence for wearables in shoulder arthroplasty is smaller and less mature. A 2025 systematic scoping review that synthesized the data for the use of wearable sensors for functional outcome measurement after reverse shoulder arthroplasty (RSA) found only six studies that met the inclusion criteria.3 Among those included, the studies were evenly divided between those that measured shoulder motion and those that measured activity counts or intensity of upper limb activity.3 Across these studies, there was substantial variation in study design and outcome measures.3 This gap likely reflects a number of factors, including the variety of rehabilitation approaches and endpoints, and the lack of standardization of sensor location and duration of wear in the shoulder arthroplasty literature.3

Machine learning (ML) has shown potential in shoulder arthroplasty, though mainly for preoperative prediction rather than postoperative monitoring.4,12 A 2020 systematic review found that ML algorithms can predict PROMs, range of motion (ROM), and complications with reasonable accuracy, sometimes outperforming conventional risk scores.4 However, predicting expected outcomes based on baseline data is different from real time identification of recovery course deviations, which represents the main challenge in postoperative monitoring. Meeting this challenge depends on resolving two underlying problems: defining what constitutes a clinically meaningful deviation and determining what to do when one is detected.

Download

Included in

Orthopedics Commons

Share

COinS