JMIR Publications

ABSTRACT

Background:

smartphones and wearable devices such as fitness trackers or smart watches as well as ambient sensors have opened up new possibilities for researchers and clinicians to gather users’ data remotely. This technology can acquire data about individuals, their daily routines, and behaviors in real time, which can be used to identify various outcomes, including loneliness and social isolation.

Objective:

This scoping review aims to identify and synthesize recent scientific studies that used passive sensing techniques, such as in-home ambient sensors, smartphones, and wearable device sensors, to collect data about users' daily routines and behaviors to detect loneliness or social isolation. Additionally, it examines various aspects of these studies, especially population, privacy, and validation issues.

Methods:

A scoping review has been undertaken following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension for scoping reviews (PRISMA-ScR). Studies on the topic under investigation have been identified through 7 databases (IEEE Xplore, Scopus, ACM, Cochrane Library, PubMed, Web of Science, EMBASE). The identified studies have been screened for the following information: the type of passive sensing detection methods for loneliness and social isolation, the targeted population, the reliability of the detection systems, and the challenges and limitations of these detection systems.

Results:

A total of 33,254 papers were identified after conducting an initial search. After screening for the inclusion and exclusion criteria, 28 studies have been included in this scoping review. The majority of studies used smartphone and wearable technology to detect loneliness, 21 out of 28 included studies have used a validated reference standard to assess the accuracy of passively collected data for detecting loneliness or social isolation.

Conclusions:

Despite growing usage of passive sensing technologies for detecting loneliness and social isolation, some substantial gaps remain in this domain. There is a lack of robust studies that have assessed loneliness detection against validated reference standards. A population heterogeneity issue exists among several studies, which indicates that different user behaviors can affect loneliness differently in different populations based on their demographics, making loneliness detection a very complex challenge. Numerous studies have mixed loneliness/isolation with other related concepts such as depression. Additionally, relatively few studies have addressed privacy and ethical concerns, with most studies having overlooked these concerns.