Validity of pervasive computing based continuous physical activity assessment in community-dwelling old and oldest-old

Narayan Schütz, Hugo Saner, Beatrice Rudin, Angela Botros, Bruno Pais, Valérie Santschi, Philipp Buluschek, Daniel Gatica-Perez, Prabitha Urwyler, Laura Marchal-Crespo, René M. Müri & Tobias Nef

Abstract

In older adults, physical activity is crucial for healthy aging and associated with numerous health indicators and outcomes. Regular assessments of physical activity can help detect early health-related changes and manage physical activity targeted interventions. The quantification of physical activity, however, is difficult as commonly used self-reported measures are biased and rather unprecise point in time measurements. Modern alternatives are commonly based on wearable technologies which are accurate but suffer from usability and compliance issues. In this study, we assessed the potential of an unobtrusive ambient-sensor based system for continuous, long-term physical activity quantification. Towards this goal, we analysed one year of longitudinal sensor- and medical-records stemming from thirteen community-dwelling old and oldest old subjects. Based on the sensor data the daily number of room-transitions as well as the raw sensor activity were calculated. We did find the number of room-transitions, and to some degree also the raw sensor activity, to capture numerous known associations of physical activity with cognitive, well-being and motor health indicators and outcomes. The results of this study indicate that such low-cost unobtrusive ambient-sensor systems can provide an adequate approximation of older adults’ overall physical activity, sufficient to capture relevant associations with health indicators and outcomes.

Introduction

It is commonly known and widely accepted that physical activity positively influences health. There is strong scientific evidence that physical activity reduces the risk for a variety of health outcomes like high blood pressure, type 2 diabetes, cancer, weight gain, falls, depression, loss of cognitive function or functional ability in seniors^1,2. While these findings are of high relevance for all age groups, they are of special importance for the growing number of old and even more so for the oldest-old adults – especially since physical activity is a modifiable risk factor^3,4. In addition, seniors are more likely to suffer from chronic diseases, experience falls or face significant cognitive decline. They are also more prone to a sedentary lifestyle⁵ and results of cardiorespiratory fitness measures even suggest an age-related acceleration in decline⁶, which might also be detectable by physical activity.

While it is evident that moderate-to-vigorous-intensity physical activity is usually better, research suggests that light- and moderate-intensity physical activity is still better than no physical activity in terms of health benefits². This is important for seniors as they may often find it difficult to engage in high-intensity physical activities such as running or aerobic exercise. Light- and moderate-intensity physical activities like cooking, vacuuming or other everyday activities, constitute an important and often integral part in older adult’s total physical activity. Measuring this type of physical activity is rather difficult but may be very important for the early detection of preventable physical activity decline or to monitor the course of interventions. Today, physical activity assessments are often based on self-reporting which is not only prone to response bias but also suffers from recall bias – especially with declining memory^4,7,8,9. Frequently used alternatives are accelerometer or pedometer based^7,10. While these provide objective physical activity measures in free-living conditions, they must be worn, which becomes cumbersome in long-term assessments of several months or even years and is thus often accompanied by wear-time dependent non-compliance issues¹⁰.

Advances in technology made pervasive computing feasible for technology assisted healthy aging by embedding smart microprocessor-driven computing devices in everyday objects (as for instance seen in appliances of smart homes)¹¹. A growing body of groundbreaking research shows that such systems are not only feasible and well accepted by seniors but are also useful for the detection of emergency situations or early changes in health status^9,12,13. A frequently used and increasingly commercialized technology is passive infrared (PIR) motion sensing, which is both inexpensive and unobtrusive, to an extent that people tend to forget about it^14,15. In this context, PIR motion sensors work by detecting the presence of a person’s motion in an equipped room¹⁶. Besides safety applications^17,18,19,20, most work in this direction primarily targeted cognitive outcomes. Galambos et al. for instance showed that changes in PIR-sensor derived motion density maps correspond to exacerbations of depression and dementia²¹. In a similar manner Hayes et al. demonstrated that variability in PIR-sensor derived activity and gait-speed data differed between cognitively normal subjects and those with mild cognitive impairment (MCI)²². Similarly, Urwyler et al. highlighted the difference between sensor derived activities of daily living patterns in healthy and MCI subjects²³.

In this work, we assess the potential of PIR-sensors in the light of physical activity. In particular, we explore the validity and potential of unobtrusive, continuous PIR-sensor readings for physical activity quantification, targeting in-home light- and moderate-intensity physical activity. Towards this goal, we analyzed the behavior of PIR-sensor based (physical) activity metrics and compared them with a multitude of cognitive, well-being and motor-function related assessments to see whether this approximation to physical activity sufficiently captures known effects of physical activity on commonly used health indicators and outcomes. The data for the analysis stems from a naturalistic sample of thirteen community dwelling old and oldest-old Swiss subjects (age = 90.9 ± 4.3 years, female = 69.23%) from the StrongAge cohort in Olten (Switzerland). All analyzed subjects shared the same apartment layout. The subjects were monitored for the duration of one year. Simultaneously, a battery of standardized clinical tests and assessments were performed repeatedly. The resulting data was aggregated and analyzed in terms of baseline differences. In addition, physical activity data from a subject with rapid health decline was evaluated and visualized in a case study format.

Results

Over roughly one year, more than 89’389 person-hours were recorded from the homes of thirteen old and oldest-old participants (age = 90.9 ± 4.3 years) (Table 1), all sharing the same apartment layout and sensor placement. During the same period, classic assessments of multiple health outcomes have been assessed. Two normalized PIR-sensor derived measures of physical activity were calculated. First, the daily sensor activity – measuring the time the sensors were detecting activity (Equation (1)). Second, the normalized daily number of room-transitions (measuring the hourly number of transitions between different rooms) (Equation (2)). Here, we present the resulting associations and observations between these sensor-based physical activity metrics and the classic clinical assessments (Fig. 1).

Visual Correlation Matrix of the four sensor-derived physical activity metrics and the clinical assessments). Shown is a visual representation^h of the respective correlations as measured by the Spearman’s rank correlation coefficients (ρ) based on an α = 0.05ⁱ. The sensor-derived physical activity metrics (rows) represent the mean and the coefficient of variation (CV) of the daily measurements over the whole monitoring duration. The size as well as colour-intensity signal the correlation strength, where red means a strong positive and blue a strong negative correlation. ^aTimed Up & Go (TUG)²⁷ (Counting = while additionally counting backwards from 100; Cup = while holding a full cup of water). ^bGeriatric Depression Scale (GDS)²⁵. ^cTinetti Performance-oriented mobility assessment (POMA)²⁸. ^dMontreal cognitive assessment (MoCA)²⁴. ^eKnee extensor strength (Knee). ^fHip flexor strength (Hip). ^gVisual analogue scale: measuring perceived health based on the EQ-5D-3L system (EQ-VAS)²⁶. ^hcreated using the R package “corrplot”^{34 i}*<0.05; **<0.01.

‍