From Motion to Metrics: How wearable technology is shaping the future of post-operative mobility assessment in Orthopaedics
Written By Shreya Kantamneni
Post-operative mobility assessment is at the cusp of transformation in Orthopaedics. Traditionally, this clinical landscape has been heavily sculpted by periodic clinical consultations, clinician-observed gait assessments and traditional patient-reported outcome measures (PROMs). Whilst it may be argued by some that these approaches remain clinically rational or rather robust, their inherent limitations raises questions about their viability against the backdrop of our dynamic, digital era. Ultimately, can we continue to justify such tools being routinely used in clinical practice, given their intrinsic risks of subjectivity and temporal sparsity?
Currently, the evolving sphere of wearable sensor technologies within Orthopaedics presents an unprecedented opportunity to shift from intermittent patient assessments to more continuous and ecologically valid monitoring. The two main domains which epitomise this movement are post-fracture rehabilitation and post-arthroplasty recovery. Collectively, these domains illustrate how potentially such wearable devices could provide tangible metrics to help inform insights from traditional PROMs and routine clinical examinations.
The Limitations of Traditional Assessment
The deep-rooted, conventional approach to monitoring post-operative mobility and rehabilitation is underpinned by episodic clinical assessments, involving visual inspection and techniques such as the “Timed Up and Go test”and goniometry (where the range of motion is manually measured at each joint). Additionally, these assessments are often further supplemented by patient-reported outcome measures (PROMs) at fixed intervals. Whilst this approach has served as the cornerstone of orthopaedic post-operative management for several decades, their inherent limitations constrain our understanding of patient recovery and restrict our ability to detect subtle deviations in the recovery trajectory of patients. Ultimately, the periodic nature of these clinical assessments renders them as momentary snapshots of functional outcomes. They do not necessarily capture the dynamic, non-linear nature of functional improvement or decline in the patient’s mobility.
On the other hand, PROMs may arguably fall short when the need for granular, real-time monitoring arises to fully understand mechanical loading, mobility patterns and joint function in patients. Garabedian et al. argue that perhaps PROMs are influenced by subjective interpretation by clinicians along with poor sensitivity; clinical scores, which may appear satisfactory on the surface, might not necessarily correlate with subtle functional deficits within patients. This gap in subjective reporting and objective functional assessment creates invisible blind spots that hinder the overall efficacy and precision of clinical decision-making.
Wearable Technology: Tiny Sensors, Big Data
The advent of wearable sensor technology, ranging from accelerometers and gyroscopes to smart insoles and IMU-based systems, provides the emerging possibility of overcoming the inherent limitations of this conventional approach. They can quantify variables such as step count, walking cadence, joint range of motion, weight-bearing force and postural stability. For example, Lebleu et al. highlight in their comprehensive analysis of 1,144 patients that wearable sensors can capture “the intricate dynamics of real-life scenarios” which “encompass physical and temporal variations throughout the day, week and so forth, particularly in the individual-level factors such as pain, stress, emotion and motivation”. This standpoint signals how wearable technology in orthopaedics marks a paradigm shift in how we fundamentally conceptualise recovery in patients; we can use the metrics derived from such wearable technologies to deliver personalised, data-driven rehabilitation plans for patients, following common procedures such as total hip arthroplasty and total knee arthroplasty.
The underlying mechanisms of wearable sensors
Currently, modern orthopaedic rehabilitation devices rely on inertial measurement units, which are sophisticated electronic sensors that combine gyroscopes, accelerometers and magnetometers to allow precise, three-dimensional kinematic testing. In particular, accelerometers detect changes in velocity whilst gyroscopes measure orientation and angular velocity. Additionally, magnetometers determine position relative to the Earth’s magnetic field.
Altogether, this motion data is collected continuously over the course of the patient’s daily activities, where these devices would record the parameters at granular temporal resolution, often capturing step-per-minute data instead of focusing solely on daily totals. Qin and Wen et al. underscore that collecting such high-resolution temporal data is essential with regards to the “long term and real-time monitoring of weight bearing movements in daily living activities such as standing and walking” because “it provides real-time feedback notifying patients and healthcare providers if weight-bearing above prescribed limits occurs”.
Emerging metrics for clinical assessment
Whilst the notion of the daily step count has historically served as the primary metric extracted from activity trackers, recent research however takes an alternative stance that parameters derived from step-per-minute data (known as cadence) offer superior clinical utility. For instance, the data of Lebleu et al. from post-arthroplasty patients illustrates that cadence-based measurements provide earlier and more reliable differentiation between recovery trajectories than total step counts.
Two examples of metrics which have been considered clinically meaningful are the ‘Peak-1 minute cadence (P1M)’ and ‘Peak 6-minute consecutive cadence (P6MC)’. In particular, P1M represents the highest step count achieved within any single minute throughout the day, serving as a crucial indicator of maximal exertion or burst activity capacity. This metric is representative of what researchers describe as one’s “best natural effort" (i.e. the free-living walking cadence of which an individual is capable). Altogether, the research findings signpost that P1M is highly dependent on age, physical function, physical activity level and body mass index (BMI), making it a sensitive marker of overall functional capacity.
On the other hand, P6MC, which is calculated using a sliding 6-minute window to identify the highest continuous activity period, provides an unsupervised equivalent to the clinical 6-minute walk test. The automated detection of P6MC from continuous activity data captures the patients’ optimal sustained walking capacity (during their normal daily routines), rather than requiring the patient to participate in supervised testing during scheduled clinic visits. In fact, studies have suggested strong correlations between P6MC measurements and standardised 6-minute walk test results, with one study by Saporito et al. reporting correlation coefficients of 0.70 between remote and laboratory-based assessments.
On the whole, the clinical relevance and significance of these cadence-based metrics becomes crystallised when their stability and sensitivity is thoroughly examined. For example, Lebleu et al.’s analysis indicated that intraweek variability for cadence measurements was approximately 40%, which is substantially lower than that of the 80% variability that was observed for total daily step counts. This enhanced stability may suggest that clinicians could view cadence metrics as reliable indicators of functional capacity, which are less likely to be influenced by the day-to-day fluctuations that complicate the way in which step count data is interpreted.
Clinical Insights into recovery trajectories
The integration of wearable technology within post-operative orthopaedic patient care has uncovered striking insights in patient recovery patterns which would have otherwise remained unnoticed using traditional clinical assessment methods. For example, Lebleu et al.’s findings demonstrate how cadence-based metrics (used within wearable sensor technology) can detect the differences in recovery earlier and more sensitively than step counts alone, across the total hip arthroplasty (THA) patient population. Lebleu et al. discovered that THA patients achieving a minimal clinically important difference returned to baseline step counts only marginally earlier than slower recoveries (33 vs 40 days). However, cadence metrics revealed the divergence sooner, with P6MC differentiating the groups by day 26 versus 32 days and P1M at 35 days, while slower recoverees failed to reach the baseline within the 60 day follow-up period. The higher sensitivity of cadence metrics can further be inferred through how the authors note that they “observed that early differences are detected using the P6MC and P1M in comparison with the total number of steps”.
The potential for remote monitoring
In addition to enhancing the quality of post-operative mobility assessment in orthopaedics, wearable technology enables clinicians to monitor patients remotely and proactively intervene when concerns arise regarding patient engagement during rehabilitation. Youssef et al. speculate in their review on digitalisation of orthopaedics that the COVID-19 pandemic has been the driving force in terms of “the accelerated use and implementation of digital tools and applications for direct patient care in the form of telemedicine.”
One of the key limitations of the conventional approach is the logistical challenge of requiring patients to attend frequent clinical consultations, which may especially result in substantial travel for those from rural areas to specialised orthopaedic care centres. Not only does wearable technology help address this, but it also enables continuous monitoring and detection of patterns that could emerge in the space between scheduled appointments, thereby allowing for more holistic assessment.
Moreover, another crucial element underpinning the impact of such wearable technology is the level of patient engagement it facilitates. For instance, Van der Walt et al. carried out a randomised controlled trial in 163 Total Joint Replacement patients, where comparison was made between the intervention group who received feedback from their Garmin Vivofit 2 activity tracker and the control group who did not receive feedback. Overall, it was found that the intervention group showed much higher mean daily step counts throughout the recovery period. This may signify how providing patients with real-time feedback into their own activity levels and allowing them the scope for visualising their progress (towards their personal recovery goals) may sustain their overall motivation and engagement with their rehabilitation.
On the contrary, not all studies seem to find consistent advantages from device feedback. For example, Kuiken et al. observed a counterintuitive pattern where TKA patients, who wore a knee-mounted goniometer (with feedback capability) demonstrated slightly higher activity rates on days without feedback (22.5 activity counts per hour) as opposed to days with feedback (15.1 counts per hour). However, one thing to note here is that this difference was not deemed statistically significant. It could therefore perhaps be argued here in this case that a heterogeneous relationship exists between feedback and the consequential behaviour as a result of a multitude of factors such as individual patient characteristics and the format in which the feedback is delivered.
The intersection between AI and clinical decision making
One of the questions that remains, in light of the benefits of wearable technology in post-operative mobility assessment, is how we can seamlessly integrate this into clinical workflows. This avenue has been expanded on by Tueni et al., who envision an “orthopaedic platform” that coherently connects virtual visits, pre-planning simulations, surgical execution and post-operative monitoring. Within this model, wearable sensor technology data is thought to flow automatically into electronic health records, where artificial intelligence algorithms would then systematically analyse patterns and flag any deviations from the expected recovery trajectories of patients.
Challenges, issues and limitations
Despite the wide array of advantages and benefits that wearable technology can offer within this field, there are several challenges that must be addressed before their widespread adoption into clinical practice.
One crucial factor to consider would be the accuracy and validity of using activity trackers. This has been highlighted by Battenberg et al., who tested multiple widely used devices in healthy participants, where they found that the waist-worn devices achieved greater than 90% accuracy for step counting across different activities, whilst wrist-worn devices fell below 90% accuracy for most activities. On top of that, the ankle-worn StepWatch Activity Monitor exceeded 95% accuracy for lower-cadence activities, but undercounted running by 25%. This altogether sheds light on how the measurement characteristics and limitations unique to each device must be thoroughly considered when interpreting the clinical data that they generate.
Within the cohort of post-arthroplasty patients, walking patterns straight after surgery can differ substantially from normal gait, with regard to altered cadence, step length and weight distribution, potentially impacting the degree of accuracy within the devices. Also, mobility aids such as crutches and Zimmer frames can consequently complicate the measurements derived from these devices. For instance, Qin and Wen et al. mention how “current limitations of many smart wearable devices include comfort issues during prolonged use, particularly for patients who have recently undergone lower limb fracture surgery”. They further add how “the weight, materials and fit of these devices may affect patient compliance and rehabilitation outcomes”. Qin and Wen et al. have also suggested that the “massive amount of data generated by smart wearable devices is difficult to integrate with existing hospital information systems (HIS) and electronic medical records (EMR)”, impeding the flow of information and “clinical decision support” due to persistent issues such as “non-uniform data formats, lack of interface standards, and poor cross-platform compatibility”(8).
Conclusion
Overall, integrating wearable technology into post-operative mobility assessment in orthopaedics reflects a fundamental reconceptualisation of how we understand, monitor and assess the trajectory of patient recovery. Through allowing continuous, objective and real-time assessment of functional capacity, these wearable sensors may potentially bridge the gap between episodic clinical evaluations and the patients’ actual lived experiences during rehabilitation.
Sophisticated metrics derived from continuous activity monitoring, particularly cadence-based metrics such as P6MC and P1M, offer a much higher degree of sensitivity in comparison with traditional measures (which tend to focus on overall step counts). This therefore would allow deviation from patients’ recovery trajectories to be identified much earlier on and the ability to distinguish between successful and problematic outcomes.
Moving forward, we must remain grounded in the fundamental purpose of these technologies in post-operative orthopaedic assessment, which is to improve overall outcomes for all patients in the most safe, standardised and precise manner possible.
Read the rest of the edition here
