Falls and fall-related injuries can have devastating health consequences and form a growing economic burden for the healthcare system. To identify individuals at risk for preventive measures and therapies, fall risk assessment scores have been developed. However, they are costly in terms of time and effort and rely on the subjective interpretation of a skilled professional making them less suitable for frequent assessment or in a screening situation.

Small wearable sensors as activity monitor can objectively provide movement information during daily-life tasks. It is the aim of this study is to evaluate whether the activity parameters from wearable monitors correlate with fall risk scores and may predict conventional assessment scores.

Physical activity data were collected from nineteen home-dwelling frail elderly (n=19, female=10; age=81±5.6 years, GFI=5.4±1.9, MMSE=27.4±1.5) during waking hours of 4 consecutive days, wearing a wearable 9-axis activity monitor (56×40×15mm, 25g) on the lateral side of the right thigh. The signal was analysed using self-developed, previously validated algorithms (Matlab) producing the following parameters: time spent walking, step count, sit-stand-transfer counts, mean cadence (steps/min), count of stair uses and intensity counts >1.5G.

Conventional fall risk assessment was performed using the Tinetti sore (range: 0–28=best), a widely used tool directly determining the likelihood of falls and the Short Physical Performance Battery (SPPB, range: 0–12=best) which measures lower extremity performance as a validated proxy of fall risk. The anxiety to fall during activities of daily living was assessed using the self-reported Short Falls Efficacy Scale-International (FES-I, range: 7–28=worst).

Correlations between activity parameters and conventional scores were tested using Pearson's r.

The activity parameters (daily means) for the 19 participants were 70.8min (SD=28.7; min-max= 22.8–126.6) of walking, 4427 steps (SD=2344; min-max= 1391–8269) with a cadence 79.3 steps per minute (SD=17.1; min-max=52.8–103.9) and 33.3 sit-stand transfers (SD=9.7; min-max=8.8–48.0).

The average Tinetti score was 21.2 (SD=5.1; min-max=10.0–27.0), with SPPB scoring 7.8 (SD=2.4; min-max=3.0–12.0), and FES-I 4.6 (SD=5.1; min-max=7.0–23.0).

Strong (r≥0.6) and significant correlations existed between the walking cadence and the Tinetti (r=.60, p=<.01) and SPPB (r=.71, p=<.01) scores. No other correlations were found between the activity parameters and the Tinetti, SPPB and none with the psychological FES-I questionnaire.

Conventional fall risk scores and activity data are comparable to literature values and thus representative of home-dwelling frail elderly including a wide range covered for both dimensions.

No quantitative activity measure had a predictive value for fall risk assessment. Strongly correlated with Tinetti and SPPB, objectively measured cadence as a qualitative parameter seems a useful parameter for remotely identifying fall risk in frail elderly. The perceived anxiety to falls was not correlated to quantitative and qualitative activity parameters suggesting that this psychological aspect hardly affects activity.

Wearable activity monitors seem a valid tool to assess fall risk remotely and thus allow low cost, frequent and large group screening of frail elderly towards a health economically viable tool for a growing societal need. The predictive quality of activity monitored data may be increased by deriving additional qualitative measures from the activity data.

Besides eliminating pain, restoring activity is a major goal in orthopaedic interventions including joint replacement or trauma surgery following falls in frail elderly, both treatments of highest socio-economic impact.

In joint replacement and even more so in frail elderly at risk of falling, turns are assessed in clinical tests such as the TUG (Timed Get-up-and-Go), Tinetti, or SPPB so that classifying turning movements in the free field with wearable activity monitors promises clinically valuable objective diagnostic or outcome parameters.

It is the aim of this study to validate a computationally simple turn detection algorithm for a leg-worn activity monitor comprising 3D gyroscopes.

A previously developed and validated activity classification algorithm for thigh-worn accelerometers was extended by adding a turn detection algorithm to its decision tree structure and using the 3D gyroscope of a new 9-axis IMU (56×40×15mm, 25g, f=50Hz,).

Based on published principles (El-Gohary et al. Sensors 2014), the turn detection algorithm filters the x-axis (thigh) for noise and walking (Butterworth low-pass, 2^ndorder with a cut-off at 4Hz and 4^thorder with a cut-off at 0.3Hz) before using a rotational speed threshold of 15deg/s to identify a turn and taking the bi-lateral zero-crossings as start and stop markers to integrate the turning angle.

For validation, a test subject wore an activity monitor on both thighs and performed a total of 57 turns of various types (walking, on-the-spot, fast/slow), ranges (45 to 360deg) and directions (left/right) in free order while being video-taped. An independent observer annotated the video so that the algorithmic counts could be compared to n=114 turns. Video-observation was compared to the algorithmic classification in a confusion matrix and the detection accuracy (true positives) was calculated.

In addition, 4-day continuous activity measures from 4 test subjects (2 healthy, 2 frail elderly) were compared.

Overall, only 5/114 turns were undetected producing a 96% detection accuracy. No false positives were classified. However, when detection accuracy was calculated for turning angle intervals (45°: 30–67.5°; 90°: 67.5–135°; 180°: 135–270°; 360°: 270–450°), accuracy for all interval classifications combined dropped to 83.3% with equal values for left and right turns. For the 180° and 360°, accuracy was 100% while for the shorter 45° and 90° turns accuracy was 75% and 71% only, mainly because subsequent turns were not separated.

Healthy subjects performed between 470 (office worker) and 823 (house wife) turns/day while frail elderly scored 128 (high fall risk) to 487 turns/day (low fall risk). Turns/day and steps/day were not correlated. In healthy subjects ca. 50% of turns were in the 45° category compared to only ca. 35% in frail elderly.

Turn detection for a thigh-worn IMU activity monitor using a computationally simple algorithm is feasible with high general detection accuracy. The classification and separation of subsequent short turns can be further improved.

In multi-day measurement, turns/day and the distribution of short and long turns seem to be a largely independent activity parameter compared to step counts and may improve objective assessment of fall risk or arthroplasty outcome.

Eliminating pain and restoring physical activity are the main goals of total hip arthroplasty (THA). Despite the high relevance of activity as a rehabilitation goal of and criterion for discharge, in-hospital activity between operation and discharge has hardly been investigated in orthopaedic patients.

Therefore, the aim of this study was to a) measure for reference the level of in-hospital physical activity in patient undergoing a current rapid discharge protocol, b) compare these values to a conventional discharge protocol and c) test correlations with pre-operative activities and self-reported outcomes for possible predictors for rapid recovery and discharge.

Patients (n=19, M:F: 5:14, age 65 ±5.7 years) with osteoarthritis treated with an elective primary THA underwent a rapid recovery protocol with discharge on day 3 after surgery (day 0). Physical activity was measured using a 3D accelerometer (64×25×13mm, 18g) worn on laterally on the unaffected upper leg. The signal was analysed using self-developed, validated algorithms (Matlab) calculating: Time on Feet (ToF), steps, sit-stand-transfers (SST), mean cadence (steps/min), walking bouts, longest walk (steps).

For the in-hospital period (am: ca. 8–13h; pm: ca. 13–20h) activity was calculated for day 1 (D1) and 2 (D2). Pre-operative activity at home was reported as the daily averages of a 4-day period. Patient self-report included the HOOS, SQUASH (activity) and Forgotten Joint Score (FJS) questionnaires.

In-hospital activity of this protocol was compared to previously collected data of an older (2011), standard conventional discharge protocol (day 4/5, n=40, age 71 ±7 years, M:F 16:24).

All activity parameters increased continuously between in-hospital days and subsequent am and pm periods. E.g. Time-on-feet increased most steeply and tripled from 21.6 ±14.4min at D1am to 62.6 ±33.4min at D2pm. Mean Steps increased almost as steep from 252 to 655 respectively. SST doubled from 4.9 to 10.5. All these values were sign. higher (+63 to 649%) than the conventional protocol data.

Cadence as a qualitative measure only increased slowly (+22%) (34.8 to 42.3steps/min) equalling conventional protocol values. The longest walking bout did not increase during the in-hospital period. Gender, age and BMI had no influence on in-hospital activity.

High pre-op activity (ToF, steps) was a predictor for high in-hospital activity for steps and SST's at D2pm (R=0.508 to R=0.723). Pre-op self-report was no predictor for any activity parameter.

In-hospital recovery of activity is steep following a cascade of easy (ToF) to demanding (SST) tasks to quality (cadence). High standard deviations show that recovering activity is highly individual possibly demanding personalised support or goals (feedback).

Quantitative parameters were all higher in the rapid versus the conventional discharge protocol indicating that fast activation is possible and safe. Equal cadence for both protocols shows that functional capacity cannot be easily accelerated.

Pre-op activity is only a weak predictor of in-hospital recovery, indicating that surgical trauma affects patients similarly, but subjects may be identified for personalized physiotherapy or faster discharge.

Reference values and correlations from this study can be used to optimize or shorten in-hospital rehabilitation via personalization, pre-hab, fast-track surgery or biofeedback.

Introduction: Physical activity is a major outcome in total hip arthroplasty (THA) and discharge criterion. Increasing immediate post-op activity may accelerate discharge, enable fast track surgery and improve general rehabilitation. Preliminary evidence (O'Halloran P.D. et al. 2015) shows that feedback via motivational interviewing can result in clinically meaningful improvements of physical activity. It was the aim of this study to use wearable sensor activity monitors to provide and study the effect of biofeedback on THA patients' activity levels. It was hypothesized that biofeedback would increase in-hospital and post-discharge activity versus controls.

Methods: In this pilot study, 18 patients with osteoarthritis receiving elective primary THA followed by a rapid recovery protocol with discharge on day 3 after surgery (day 0) were randomized to the feedback group (n=9, M/F: 4:5, age 63.3 ± 5.9 years, BMI 26.9 ± 5.1) or a non-feedback control group (n=9, M/F: 0:9, age 66.9 ± 5.1 years, BMI 27.1 ± 4.0). Physical activity was measured using a wearable sensor and parameters (Time-on-Feet (ToF), steps, sit-stand-transfers (SST), mean cadence (steps/min)) were calculated using a previously validated algorithms (Matlab). For the in-hospital period data was calculated twice daily (am, ca. 8–13:00h and pm, ca. 13–20:00h) of day 1 (D1) and 2 (D2). The feedback group had parameters reported back twice (morning, lunch) using bar charts comparing visually and numerically their values (without motivational instructions) to a previously measured reference group (n=40, age 71 ±7 years, M:F 16:24) of a conventional discharge protocol (day 4/5). Activity measures continued from discharge (D3) until day 5 (D5) at home.

Results: Randomization resulted in matched groups regarding age and BMI, but not gender. The first post-op activity assessment (D1am) was identical between groups. Also thereafter similar values with no significant differences in any parameter were seen, e.g. the time-on-feet at D2PM was 59.2 ±31.7min (feedback) versus 62.9 ±39.2min (controls). Also on the day of discharge and beyond, no effect from the in-hospital feedback was measured. For both groups the course of activity recovery showed a distinct drop on day 4 following a highly active day of discharge (D3). On day 5, activity levels only recovered partially. For both groups, all quantitative activity parameters were significantly higher than the reference values used for feedback. Only cadence as a qualitative measure was the same like reference values.

Discussion: Biofeedback using activity values from a body-worn monitor did not increase in-hospital or immediate post-op home activity levels compared to a control group when using the investigated feedback protocol. In general, while the day of discharge steeply boosts patient activity, the day after at home results in an activity drop to near in-patient levels before discharge. In a fast track surgery protocol, it may be of value to avoid this drop via patient education or home physiotherapy. Biofeedback using activity monitors to increase immediate post-op activity for fast track surgery or improved recovery may only be effective when feedback goals are set higher, are personalised or have additional motivational context.

Summary

Physical activity monitoring using a single accelerometer works reliably in clinical practice and is of added value as clinical outcome tool, as it provides objective and more precise information about a patient's activity compared to currently used questionnaires.

Summary

Upper extremity activity was similar in patients and healthy subjects, showing no significant asymmetry between arms within subjects. Further improvements (e.g. thresholds, filters, inclinometer function) are needed to show the clinical value of AM for patients suffering shoulder complaints.

Summary

Movement analysis (IMA) and activity monitoring (AM) using a body-fixed inertia-sensor can discriminate patients with ankle injuries from controls and between patients of different pathology or post-injury time. Weak correlations with PROMs show its added value in objectifying outcome assessment.

Summary

A single 3D accelerometer is accurate in measuring upper-extremity activity durations, rest periods and intensities, suggesting its feasibility for daily life measurements with patients. Further enhancements are feasible to reduce residual false classifications of intensity from certain activities.