JMIR Publications

ABSTRACT

Background:

The world population is aging, challenging governments and healthcare systems to handle these increasing needs. Tracking performance of Activities of Daily Living (ADLs) using ADL recognition has potential to facilitate aging in place strategies, allowing older adults to live in their homes longer and providing their families/caregivers the ability to monitor changes in health status. However, the ADL recognition literature historically has evaluated systems in controlled settings with data from younger populations, creating the question of whether these systems will work in real-world conditions for older populations.

Objective:

This scoping review seeks to establish the state-of-the-art for recognizing basic ADLs using wearable sensors and identify how many of these works include older adults.

Methods:

We conducted a scoping review using the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines to identify studies focused on ADL recognition using wearable sensors within the PubMed, Association of Computing Machinery Digital Library, and Google Scholar databases. This review focuses on studies that were published in the last 5 years to identify current trends and include at least one of the basic ADLs defined by Katz et al.: bathing, dressing, toileting, transferring, continence, and feeding. Studies that include older adults are highlighted.

Results:

56 studies satisfied the inclusion criteria; only 8 studies included older adults despite most studies identifying this population as a focus for their research. Most works focused on eating (n=25), hygiene (n=22), drinking (n=18), or transitions (n=13). Few works included toileting (n=3), dressing (n=1), or bathing (n=1) activities. Of the 8 studies that included older adults, 5 focused on recognition performance while 3 focused on user experience and system acceptability.

Conclusions:

Basic ADLs are unevenly covered in the ADL recognition literature. Few studies focus on bathing, dressing and toileting most likely due to privacy concerns and attributes making them difficult to detect. Most systems do not evaluate their performance on the intended audience of older adults. We summarize the state-of-the-art for recognizing each category of basic ADLs and identify barriers to widespread adoption such as the high cost of annotating data and the lack of publicly available datasets. Future research directions to address these gaps include activity recognition systems that provide an established benchmark for the quality of the activity performed, developing standalone devices that automatically log ADL performance overcoming barriers of data annotation, and including older adults in activity recognition studies.