Ambulatory measurement of knee motion and physical activity: preliminary evaluation of a smart activity monitor

The complexity of human physical activity has made it challenging to produce a validated, accurate, and cost-effective technique to quantify activities of daily living [1‐3]. The value of a sophisticated gait lab is well-established, but gait labs are expensive, require trained personnel, and may not simulate normal environments. Of the portable devices, those with accelerometers are effective in monitoring human activity when that activity is known [4‐14]. These devices are attractive because they are small, non-invasive, and inexpensive. Unfortunately, many are not "smart" enough to determine the type of physical activity (e.g. stair ascent vs. walking on level ground) that the subject is performing. Pedometers are generally not sensitive to differences in stride length and are less accurate when worn by obese patients [15]. Actometers and wrist/ankle devices can provide qualitative data via "on" and "off" switches, but they are limited in their ability to record quantitative data [14, 16, 17]. Foot-contact monitors and electronic load transducers are problematic in their technical and practical limitations, and no reports exist in the literature regarding their accuracy in measuring human physical activity[18].

The Intelligent Device for Energy Expenditure and Activity system (IDEEA™, MiniSun, LLC), a microcomputer-based portable physical activity measurement device, allows detection of multiple gaits, limb movements, and postures (walk, run, up stairs, down stairs, stand, sit, step, jump, lie, recline, transition, etc.). It also analyzes gait, speed, distance, power, work, and energy expenditure. The IDEEA system's accuracy has been evaluated in previous investigations [18, 19]. Their validation protocol required subjects to perform a series of choreographed activities. The timing of these various activities was then recorded and compared to the IDEEA. The IDEEA system was found to be accurate in measuring energy expenditure, postures and limb movements, and speed of walking and running. The original IDEEA, as described in the investigation above, was modified for our study by adding two electrogoniometers.

Various rehabilitation protocols and prosthetic knee designs assume that knee flexion as measured in laboratories reflects ambulatory knee range of motion, but this assumption, to our knowledge, has not been tested. In particular, implant manufacturers now produce "high flexion" total knee designs that may safely permit up to 150° knee flexion [20‐22], but whether even healthy subjects employ these ranges of motion has not been investigated outside gait laboratories. In the present study we investigated the validity of the modified IDEEA system's ability to accurately detect physical activities and knee flexion angles compared to the Massachusetts General Hospital Biomotion Laboratory (BML). The BML permits full body analyses of kinematics and kinetics using the Selspot/TRACK data acquisition system during standing and locomotion activities, with precision and accuracy of < 1 mm position and < 1° orientation [23]. We hypothesized that 1) knee flexion angles as reported by the IDEEA would correlate with knee flexion angles as recorded by the BML and 2) the IDEEA would accurately detect activities performed during short choreographed sessions outside the BML. Validation of the modified IDEEA in healthy subjects would corroborate previous investigations, and in addition provide the error boundaries for use in determining knee flexion angles and activities of daily living in the outpatient setting. The ability to evaluate these parameters at home has numerous potential applications for patients with disorders of the musculoskeletal and neurological systems.

The subjects used the IDEEA (including time during testing in the BML) for an average of 17.4 +/- 9.7 hours (range, 7.5–33 hours). Two subjects wore the device overnight while sleeping. Figure 5 summarizes the various activities performed by all subjects while awake. Subjects took an average of 8,441 +/- 4,785 steps (range, 4,369–14,715 steps) per session. Table 1 quantifies the various activity parameters recorded by the IDEEA system during each subject's entire data collection period.

The pooled correlations between the BML and the IDEEA system knee flexion angles were .98 +/- .01 for gait, .98 +/- .02 for stair descent, and .97 +/- .03 for step up/down (Figures 6, 7, 8). Four of 5 subjects flexed their knees >120° at any time during their data collection periods. Two subjects recorded knee flexion >160°, both during sitting with their foot underneath their contralateral buttock. Time spent at >120° of knee flexion averaged 58 +/- 39 seconds (range, 0–267 seconds). This time spent at knee flexion >120° represented, on average, 0.17 % of each subject's testing session. Figure 9 shows the number of occurrences of various knee flexion angles, for one subject, during the data collection period outside the BML.

In testing during the choreographed sessions outside the gait laboratory, the IDEEA accurately reported activity for all 5 subjects in all trials (Table 2). While testing in the BML protocol, the IDEEA system accurately identified the gait trial in all five subjects. It correctly identified stair descent in three of the five subjects. For the stair descent trials of the other two subjects, the IDEEA incorrectly reported their activity as "walking". The IDEEA identified stepping up/down correctly in only two of the five subjects. It incorrectly reported "walking" during the step up/down trials for the other three subjects (Table 3).

The frequency that healthy subjects use deep knee flexion outside the laboratory setting is currently unknown. We sought to validate the modified IDEEA system by using the Selspot/TRACK data acquisition system in a laboratory and in subjects' natural environments. In the laboratory, we limited our examination to knee motion and 3 activities in 5 healthy subjects, as it is these data that hold the most relevance to prosthetic device research. Our results indicate that the IDEEA system, when compared to the gold-standard gait laboratory, is able to accurately report knee flexion angles. During BML testing it was able to accurately report gait, but it was less accurate in recording stair descent and step up/down. However, during testing outside the BML, the IDEEA accurately identified walking, running, standing, sitting, stair ascent, stair descent, and lying in the five subjects.

The inability of the IDEEA system to accurately detect stair descent and step up/down during laboratory testing may be due to the limitations inherent in our protocol. The size of the data collection area restricts both the activity type and duration of activity that can be evaluated. This size constraint limits our stair model to 4 steps. MiniSun states that 4 steps are too few to allow the IDEEA to confirm this activity. We know that we diminished the estimate of accuracy by limiting the maximum time available for activity detection. However, this was done in an effort to perform a highly standardized data analysis, as has previously been the basis for prosthetic knee design. Moreover, during the stepping protocol, subjects stepped forwards and backwards, a task IDEEA is not currently designed to detect.

Our results of testing for activity identification outside the BML protocol corroborate previous investigations [18, 19]. In their validation of the IDEEA system, the authors used a timed protocol of specific activities to measure postures, limb movements, and jumping. They evaluated stair ascent and descent by timing subjects on the stairs at two different speeds. In combining the IDEEA with the flexible electrogoniometers, we have created a tool capable of determining, among others, the amount of knee flexion needed for activities that are commonly-performed outside the gait laboratory.

The pooled correlations between IDEEA and BML knee angles during step up/down, gait, and descending stairs ranged from .93 to .99. These data suggest that the IDEEA accurately measures knee motion during these three activities. The electrogoniometers proved to be durable, as none of the devices failed to collect data after the subjects exited the gait laboratory.

The data recorded by the modified IDEEA system confirm that some patients flex their knee >120° and that the system is capable of recording deep knee flexion angles up to 160°. Four of 5 subjects flexed their knee > 120° during routine activities. Two subjects flexed their knees >160° while sitting on a chair with their foot curled under the contralateral buttock. On average, these five subjects spent 0.17 % of their testing session with their knees in >120° of knee flexion. These data must be interpreted cautiously, as we only tested 5 subjects who performed jobs that do not regularly require deep knee flexion. The subjects may have used knee flexion >120° more often if they had been evaluated on a weekend or holiday.

We found that the modified IDEEA system, compared to the Selspot/TRACK data acquisition system in the MGH BML, accurately reported healthy subjects' knee range of motion. The IDEEA system was also able to accurately detect walking, running, standing, sitting, stair climbing, stair descent, and lying during choreographed activities outside the BML. The results of the present study, in conjunction with previous reports, support the use of the modified IDEEA system in the outpatient setting. In the future, we plan to use the IDEEA system to evaluate knee motion and frequency and duration of activities of daily living in patients who have had total knee arthroplasty. This approach may eventually allow for the assessment of surgical outcomes for different prosthetic designs.