I Wandered Lonely in the Cloud: A Review of Loneliness,
Social Isolation and Digital Footprint Data
Dominic Reedman-Flint
1
, John Harvey
1a
, James Goulding
1b
and Gary Priestnall
2
1
N/LAB, Nottingham University Business School, University of Nottingham, U.K.
2
School of Geography, University of Nottingham, U.K.
Keywords: Social Isolation, Loneliness, Human-Computer Interaction, Digital Trace Data.
Abstract: The harm that social isolation and loneliness can have on physical, mental, and emotional well-being is now
well evidenced. With social distancing and remote working now commonplace, the dangers of loneliness are
ever more acute. Consequently, information technologies have taken on renewed importance to support
healthy communication and reduce the negative impacts of social isolation. However, existing literature
remains highly conflicted as to the relationship between technology use and its impact on loneliness. This is
perhaps understandable: measures of loneliness have traditionally been examined within clinical settings, far
removed from the everyday realities of computational interactions. Yet data logged about such interactions
now offers potential to help identify isolation and loneliness and support those experiencing resulting health
issues. We present a scoping review of this domain, focusing on detection of loneliness and social isolation
through digital data. We interrogate a corpus of published articles from the HCI literature, identifying a series
of methodological, epistemological, and ethical tensions therein, as well as emerging opportunities for future
empirical study. We identify a need to examine such phenomena via actual behavioural data, rather than
reliance on historical proxies such as age and gender, to help modernize our understanding of this growing
social ill.
1 INTRODUCTION
"The world is the closed door. It is a barrier. And at
the same time it is the way through. Two prisoners
whose cells adjoin communicate with each other by
knocking on the wall. The wall is the thing which
separates them but it is also their means of
communication. … Every separation is a link."
The quote above by philosopher Simone Weil
(1997) draws on Plato’s concept of metaxu, meant
here as something that both separates and connects
simultaneously. It was originally intended to describe
the challenge of communion with God, but its
metaphor of a prison wall also serves as a useful
analogy for how contemporary researchers describe
technology and its effect on communication between
people more generally. Recent works summarising
loneliness and online social interaction have shown
how information technologies, and in particular
a
https://orcid.org/0000-0003-4188-1900
b
https://orcid.org/0000-0002-8892-6398
interconnected digital devices, appear in the literature
as both the cause of, and solution to, the growing
issue of loneliness (Dunbar, 2021; Hertz, 2021).
Much of the literature is therefore conflicted,
illustrating that the causal relationship between the
two categories is both complex and likely bi-
directional; such that the aetiology of loneliness is at
least partially context dependent. Technology
transforms social relations, and in doing so, creates
new opportunities for alienation and communion
alike.
Despite the complexity involved in potential
manifestations of loneliness, research conducted by
social neuroscientists (Cacioppo and Patrick, 2008)
provides scientific evidence that loneliness causes
physiological events that wreak havoc on our health.
Persistent loneliness leaves a mark via stress
hormones, immune function and cardiovascular
function (Knox and Uvnäs-Moberg, 1998) with a
cumulative effect that brings health outcomes similar
Reedman-Flint, D., Harvey, J., Goulding, J. and Pr iestnall, G.
I Wandered Lonely in the Cloud: A Review of Loneliness, Social Isolation and Digital Footprint Data.
DOI: 10.5220/0011578500003323
In Proceedings of the 6th International Conference on Computer-Human Interaction Research and Applications (CHIRA 2022), pages 225-235
ISBN: 978-989-758-609-5; ISSN: 2184-3244
Copyright
c
2022 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
225
to being a smoker (Holt-Lunstad et al., 2010).
Furthermore, loneliness exacerbates the risk of
experiencing additional subsequent problems and
altering behaviour insofar as it increases the
likelihood ‘of indulging in risky habits such as drug
taking and plays a role in mental disorders such as
anxiety and paranoia’ (Griffin, 2010). Despite this,
many people experiencing loneliness do not interface
with medicalized settings; and few receive clinical
diagnosis or support. Due to the hidden nature of the
problem, estimates of loneliness are often absent from
national statistics despite potential to inform social
policy. As a result, researchers have often turned to
demographic proxies for loneliness, with much work
throughout the 20th century seeking to theorise the
phenomena, and the nature of those people
predisposed to it. In particular, the elderly (over 65s)
and disabled, widely cited in academia as being most
affected, underpin most analyses of loneliness
prevalence. Yet the UK’s Office of National Statistics
(ONS, 2018) and BBC, via the world’s largest
empirical study of loneliness (BBC, 2019), recently
showed that loneliness is far from constrained to these
communities; and affects the population in far more
ways than previously recognised
Recently, the issue has been further exacerbated
by Covid-19, and our increasing exposure to
conditions conducive to loneliness (Bu et al., 2020).
Dramatic societal change, lockdowns, and enforced
social distancing has spawned a flurry of research to
help identify those most at risk of loneliness and its
myriad health consequences (Ernst et al., 2022). Yet,
the mass transformation of work and socialization
into increasingly online settings has also created
opportunity for researchers: digital identification of
those at risk. Whether in cases where digital
technology exacerbates loneliness or in situations it
diminishes it, human-computer interactions leave
behind a rich corpus of digital logs. Such digital
footprint data, if handled responsibly, can offer new
avenues to help identify, characterize, and intervene
in this “wicked problem”. The development of big
data and ubiquitous digital architectures to log
everything from library records, transport
movements, health records, social interactions and
even shopping habits illustrates the wide range of
siloed information available; data that might be
harnessed to better understand vulnerability to
loneliness and to direct social policy accordingly.
With such potential, of course, comes risk:
approaches that depend on indirect, closed, or
proprietary data sources introduce new
methodological, ethical, and privacy challenges,
otherwise absent in clinical settings.
The overall aim of this work is therefore to
establish the contexts in which digital data are being
used when identifying people experiencing social
isolation or loneliness, or where data are being
inadvertently created by the interactions of those
people experiencing loneliness (particularly those
who are doing so in populations not typically
recognised within clinical settings). Unlike review
papers that have attempted to map the efficacy of
digital technologies as loneliness interventions from
a public health perspective (Shah et al, 2019; Ibarra et
al, 2020), we provide a scoping review of how social
isolation and loneliness interventions are already
conceived, discussed, and enacted within the
computing literatures. By examining epistemological
and methodological bases of previous research we
aim to expose some of the underlying intellectual
commitments and assumptions currently being made
in the field and contextualize ongoing debate. We first
outline the methodology and scoping approach used
to achieve this, before discussing emerging genres in
the field and the key analytical differences that
separate researchers. Finally, we discuss emerging
issues and empirical gaps in the domain that, if
addressed, can serve as the basis for addressing this
phenomenon.
2 METHDOLOGY
While ‘loneliness’ and ‘social isolation’ are well
accepted terms in everyday usage, definitions in
academic contexts vary significantly by scientific
discipline and empirical focus. Significant existing
work has sought to characterise loneliness and social
isolation through study of their respective aetiologies
from a public health perspective (Holt-Lunstad, 2017;
Elovainio et al., 2017; Stepto et al, 2013; Lubben,
2017). Various constructs are described in the
literature suggesting interconnectedness between the
conditions; and in much work the terms ‘loneliness’
or ‘social isolation’ are used interchangeably -
particularly in research which does not aim to
conceptualise the respective concepts. Yet there is
limited consensus here, with other research viewing
the conditions as entirely distinct and to be considered
independently (Matthews et al, 2016); and in
psychological and clinical research it is far more
common to differentiate the constructs. Social
isolation is typically described as the circumstance of
being physically alone or otherwise detached from
contact with friends, family, or society; Loneliness, in
contrast, is described as a negative psychological
response to such situations, commonly portrayed as
CHIRA 2022 - 6th International Conference on Computer-Human Interaction Research and Applications
226
‘a subjective, unwelcome feeling of lack or loss of
companionship... [occurring] when we have a
mismatch between the quantity and quality of social
relationships that we have, and those that we want’
(Perlman and Peplau, 1981). Academic definitions
are contested for both terms, with some authors
speaking of multiple sub-types of loneliness (Weiss,
1975), yet a separation is commonly recognised
between them. As our intention is to synthesize
literature around these concepts, we do not challenge
such definitions - while the overlapping of language
creates challenges for comprehensive summarisation
of the field, it also highlights the need for consilience
and transdisciplinarity to support further advances.
Archetypal definitions of loneliness and social
isolation primarily came from diagnostic scales
developed in the latter part of the 20th century. Weiss
(1975) saw ‘social loneliness as a lack of or negative
change in social connections below a desired level,
whereas ‘emotional loneliness’ was a lack of deep,
meaningful (i.e. romantic or familial) connection.
This multidimensional approach to the study of
loneliness is not, however, reflected in the widely
used UCLA Loneliness Scale (Russell, 1996) which
treats loneliness as a unidimensional construct. While
this approach has been contested (Marangoni and
Ickes, 1989), Russell et al. (1984) have argued that,
despite different forms existing, that the UCLA scale
adequately summarises these different loneliness
states. UCLALS is the most widely used
measurement for loneliness, but several alternative
well-cited measures exist. Almost uniformly,
however, these predate widespread adoption of the
Internet and digital devices - whether de Jong
Gierveld’s Scale (De Jong-Gierveld, 1987), the
Social and Emotional Loneliness Scale for Adults
(DiTommaso and Spinner, 1993), Differential
Loneliness (Schmidt and Sermat, 1983) or the
Loneliness Rating Scale (Scalise et al, 1984).
One of the key, outstanding questions for digital
identification of loneliness, is therefore the extent to
which present-day research should revisit such
measures, given the rapid integration of technology
into our daily lives over the past two decades (e.g.,
mobile private messaging, social networking, and
livestreamed video). The impact the digital world has
had on our experience of loneliness and social
isolation remains unclear; do traditional metrics still
hold; and to what extent have traditional proxies
become anachronistic? Loneliness has predominantly
been identified in clinical settings through surveys
and self-diagnosis, or via direct medical assessment.
Do such environments overlook people in need of
support but who do not seek out practitioners, either
being unable or stigmatised from doing so? To
consider questions of this nature, Munn et al. (2018)
have advanced scoping reviews as useful tools to
examine emerging evidence; especially when it is still
unclear what other, more specific questions can be
valuably posed and addressed by future empirical
work. Such reviews aim to not only report on
evidence that informs practice in the field, but to
consider the way research has been conducted, and in
particular ‘in contrast to traditional literature reviews
scoping reviews are informed by an a priori protocol;
Are systematic; Aim to be transparent and
reproducible; and ensure data is extracted and
presented in a structured way’ (Munn et al., 2018).
This scoping review focuses on two databases
containing peer-reviewed papers in computing and its
associated sub-disciplines, the Association for
Computing Machinery (ACM) and the IEEE Xplore
(IEEE) digital libraries respectively. Both libraries
were searched, isolating abstracts containing the
terms ‘loneliness’ or ‘social isolation’. This produced
a total of 401 results, and the resulting corpus was
screened for non-English, inaccessible, or non-peer
reviewed articles or conference papers. Each article
was then reviewed to ensure that social isolation or
loneliness was not peripheral but a relevant empirical
or conceptual focus of the work. Papers either: (1)
identified social isolation or loneliness as part of their
sampling procedure; (2) used social isolation or
loneliness as a dependent or independent variable
within analysis; or (3) social isolation or loneliness
was specifically being explored or researched through
an inductive or conceptual approach. This process is
summarised via the PRISMA diagram in Figure 1.
The final output corpus yielded 52 articles, which
were taken forward to fine-grained analysis. Each
article was decomposed, and re-summarised into a
short commentary before then being analysed in
relation to a series of methodological questions.
Answers to these questions were tabulated, serving as
the basis thematic identification discussions between
the research team, who collaboratively identified
genres and epistemological commitments made by
the works.
The methodological research questions
considered are: (1) Is the work empirical (data
gathering) or conceptual? (2) Data provenance - if
data are gathered what is collected and how? (3)
Sampling - who is the focus of the research / who is
thought to be lonely? (4) Does the study aim to
identify people experiencing issues, intervene to help
people, or to conceptualize what loneliness is? (5) Is
loneliness or social isolation measured and if so, how;
is it a pre-existing measure or newly developed? (6)
I Wandered Lonely in the Cloud: A Review of Loneliness, Social Isolation and Digital Footprint Data
227
Is an expert identified within labelling (or not); who
gets to identify those being labelled lonely?; (7) What
methods were used to analyse loneliness data
generated? (8) what metadata is likely to exist as part
of the measures chosen e.g., social network data,
spatial data, or time series data; and (9) What
constitutes success and who are the beneficiaries?
The answers to these questions were then used to (a)
identify genres of work with clustered formulations
of loneliness problems/challenges addressed via
digital technology; (b) to examine contrasting
methodology and epistemology of studies in the field,
and (c) to indicate promising avenues for future
research.
3 ANALYSIS
52 papers were included in the final corpus. Figure 1
shows the earliest three papers from the corpus
(Zhang, 2009; Zhou et al., 2009; Waterworth and
Ballesteros, 2009) appeared in 2009, with a gradual
annual increase until 2020, which saw 12 relevant
papers published (search conducted 04/2021). 37
articles used established loneliness measures, of
which 16 used the UCLA loneliness scale. The vast
majority of papers that measured loneliness or social
isolation directly used survey questions using Likert
Scales or specific protocols (e.g., UCLA or De Jong
Gierveld scales). Those papers that did not use
established measures did so primarily due to an
inductive or exploratory qualitative focus, or because
of emphasis on intervention development (e.g. use of
robots) rather than intervention assessment.
A striking feature of the corpus is that two specific
demographics dominate the empirical work (the
elderly – 19 papers; and student populations – 9
papers). Such is their emphasis within the corpus that
in the following section we examine each of these
categories in turn as distinct genres. It is, however,
worth noting that this bimodal split is based upon a
priori demographic sampling decisions rather than
selections made due to indicators of loneliness or
social isolation. Most students graduate in their early
twenties, whereas retirees (particularly those in
nursing homes, a common setting of loneliness
research) normally exceed 65 years. This >40-year
gap reflects a large section of the populace absent
from research. Within the corpus 15 articles
explicitly sought to develop methods to identify
lonely or socially isolated people. 12 papers focused
explicitly on loneliness or social isolation
interventions. The remainder contained a mixture of
inductive, exploratory, and hypothetico-deductive
approaches in which loneliness or social isolation
featured.
In the following section, we describe genres
resulting from the analysis. Genres are structured
from clusters of research that frame problems of
loneliness, isolation, and digital technology in similar
ways. Given the recent emergence of much of the
work and the inevitable overlapping of some themes
within papers the genres should not be thought of as
internally consistent movements to which the authors
are aligned, rather the aim here is to illustrate
similarity of agendas and offer a lens through which
some of the key debates can be seen. There are five
identifiable genres within the corpus that describe all
but one paper (related to testing a loneliness scale in
Italy (Senese et al., 2020), albeit unrelated to digital
footprints); these are now discussed in turn.
Genre 1: The Elderly
18 papers contained an explicit focus on loneliness or
social isolation in the elderly, making it the largest
genre in the corpus (Broadbent et al. 2018; Eldib et
al.,2015; Yang and Bath, 2018; Pedell et al., 2010;
Baecker et al., 2014; Light et al., 2017; Zadeh et al.,
2020; Martinez et al., 2017; Austin et al., 2016; Ring
et al., 2013; Ha and Hoang, 2017; Bacciu et al., 2016;
Noguchi et al., 2018; Chang and Kalawsky, 2017;
Yoshida et al., 2018; Mulvenna et al., 2017; Petersen
et al., 2013; Waterworth et al., 2009). In these, older
people are identified as an at-risk group due to the
intersection of multiple life events - for example:
leaving the workforce through retirement; losing
family through bereavement; having children leave
home; or being forced into care homes for health
reasons. In each case, elderly people see
transformations of their social networks, losing
opportunities for meaningful social interaction. Such
features of ageing are well known; and because elderly
populations often tend to have limited geospatial
mobility and higher chances of interfacing with
medicalised settings, visibility of loneliness and
isolation is increased, allowing for more easily targeted
interventions. It is therefore understandable that this
genre also features the highest relative share of
interventions detailed within the corpus. In clinical
practice, algorithmic risk-based estimates of loneliness
in the elderly are likely to perform well, particularly
where personal data is available and where trust in the
robustness and explainability of the method can be
secured. Indeed, recent work showcases machine-
learning approaches already yielding reliable
performance (Yang and Bath, 2018). However, current
risk-based methods typically depend on presentation
or referral within a clinical setting.
CHIRA 2022 - 6th International Conference on Computer-Human Interaction Research and Applications
228
Figure 1: (Left) Review Screening Process; (Right) Yearly breakdown of publications within final corpus.
Sampling issues of this nature raise concerns about
potentially under-served sub-populations (e.g., those
living alone), not evident in clinical settings; and who
may be more effectively identified via non-traditional
means (e.g., via digital footprint data).
Genre 2: The Student Experience
Academics have historically been criticised for an
overreliance on sampling students as part of
psychological and clinical research, but in this
instance, the need for research on the experience of
loneliness within student populations is of clear
relevance. Students often move cities to attend
university, with a sizeable minority moving
internationally to an unfamiliar place to live and work
amongst unfamiliar people. A range of papers use
students as an explicit empirical focus or do so
implicitly by only sampling from student populations
(Fuentes et al., 2016; Joyner et al., 2020; Zhou et al.,
2020; Zhang, 2010; Kindness et al., 2013; Lu and
Yao, 2010; Zhou et al., 2009; Xu et al., 2015; Ferrer
et al., 2020). Migration is a key cause of loneliness
for this cohort. Students can be made lonely and
isolated via their own movement, which fragments
prior social networks (in hope of the formation of new
ones), while limiting familial support structures. The
methodological tendency within this genre tends
towards consideration of online data sources useful
for identification of mental health problems, whereas
for the elderly genre research methods tend to focus
on interventions, and data generated by physical
technology in situ. Most papers in the corpus leverage
established survey measures to identify loneliness;
but only three papers seek to identify loneliness as a
dependent variable with the remainder focusing more
on interpersonal relationships and wellbeing. One
article (Zhou et al., 2020) notes, for instance that
‘Previous studies on loneliness have mainly focused
on using questionnaire-based loneliness scales, e.g.,
UCLA scale, for the measurement of loneliness.
Nevertheless, the lonely may prevent reporting their
real conditions since they are afraid of information
disclosure, discrimination, unfair treatment, thus it
makes the information accumulated by these
questionnaires unreliable.’ The challenge of reliable
sampling frames again raises the possibility to
supplement traditional approaches using alternative
health surveillance methods (particularly those that
examine social network or communication data - e.g.,
classroom collaboration network data). Incorporating
behavioural observation data with surveys is likely to
help scale prevalence estimates of loneliness across
broader populations (behavioural vs demographic
proxies) although risk-based probabilistic approaches
are better suited to population prevalence estimation
than deterministic assessments used in clinical
practice. There is likely a useful bifurcation in data
and methods used for population and individual level
assessment worthy of further inquiry, particularly due
to the sensitive nature of social network data.
Genre 3: Online Services, the Web, and Apps as
Windows into the Experience of Loneliness
A growing number of papers (Brueckner, 2020; De
Choudhury et al., 2014; Joseph et al., 2014; Weinert
et al., 2014; Jeong et al., 2015; Taylor et al., 2017;
I Wandered Lonely in the Cloud: A Review of Loneliness, Social Isolation and Digital Footprint Data
229
Burke et al., 2010; Ananto and Young, 2021; Jeong et
al., 2016; Galunder et al., 2018; Pulekar and Agu,
2016; Rabani et al., 2020; Kaur et al., 2020; Wu et al.,
2016; Lu and Yao, 2010) draw on diverse data
sources being used from popular Web and Smart
Device applications such as (messaging, phone calls,
web browser logs, gaming, teleworking applications,
Facebook, Instagram, Twitter, Grindr, and Amazon
Kindle) to explore loneliness within the respective
user populations. 7 articles in the corpus used
established survey measures, of which 5 applied the
UCLA survey. This genre is characterised by big data
and behavioural analytics approaches, with
commercial Web services often being repurposed for
prosocial reasons. For example Gao et al. (2019)
correlate usage of a dating website with loneliness
measures, regressing features engineered from
profiles, postings, and check-ins with ‘surveys [of]
psychological states by four professional
questionnaires measuring different kinds of mental
disorders: depression, loneliness, anxiety, and stress’.
Mixed method approaches of this nature, that
combine structured data (e.g. clinical surveys) with
wide-ranging and often unstructured social network
data, hold promise for prevalence risk estimates
within their user population; though are likely to
introduce potentially hidden forms of survivorship
bias that require triangulation within a broader
population before they could be validated. The
linking of social network usage (both time spent,
words/sentiments used, and people contacted) is a
common area of study amongst this genre. An
illustrative example can be seen in a study (Kaur et
al., 2020) of over 140 million tweets to analyse
personality insight, emotion, and sentiment analysis
in relation to social isolation during the pandemic.
The premise is a straightforward one: how we
describe ourselves and who we interact with is liable
to change throughout our lives, with such transitions
being mirrored in our digital footprints online. If this
is true, then as the authors note, aggregated results
might be used to support ‘a public health indicator to
anticipate the possibility of social isolation and design
health policies accordingly.’ However, despite
showing initial promise, further work is needed to
evaluate the efficacy of such claims, particularly
where approaches can be replicated alongside robust
clinical measures.
Genre 4: Physical Technology, Robots, and
Anthropomorphic Interactions
A series of papers (Lazányi, 2016; Eyssel and Reich,
2013; Lou et al., 2019; Li et al., 2020a; Li et al.,
2020b) focus on the creation or delivery of physical
resources, often robotic interventions, to help
alleviate social isolation and/or loneliness. The focus
of these papers is primarily development of new
technology rather than identification of loneliness.
The genre emphasises intervention (or what has been
called ‘prosocial interaction design’ Harvey et al.,
2014), and specifically focuses on transforming the
physical environment around people likely to
experience loneliness. However, unlike interaction
design strategies that aim to pair people with other
people to foster communion and thus remove
loneliness, a far greater emphasis is placed on non-
human subjects and mastery of anthropomorphism.
The papers tend to focus on dyadic solutions to
loneliness i.e., creating a surrogate partner (either
high fidelity humanoids or non-humans such as pets
with human-like features Lou et al., 2019) with
whom a person can form meaningful attachment.
Though Eyssel and Reich (2013) found that people
experiencing loneliness may be more likely to
anthropomorphise robots, Li et al. (2020a; 2020b)
find that the relationship is not so clear and that a
more nuanced understanding of what loneliness is can
better serve to predict efficacy of interventions,
specifically through reference to ‘trait’ versus ‘state’
loneliness. Whilst this concept - effectively
representing chronic and temporary loneliness - is not
unique in clinical literature, it is within the papers
identified within this corpus and suggests distinct data
sources could be considered for more general
measures. This genre points to the idea that
multidimensional measures of loneliness in the digital
world may help to develop subsequent interventions,
it is therefore worthy of further research, particularly
where new measures account for online experience
and can be paired with longitudinal clinical outcomes.
Genre 5: Edge Communities
The smallest identifiable genre, covering 4 papers,
considers differing edge communities - people living
in more extreme physical or mental conditions,
including refugees (Almohamed and Vyas, 2016),
mental health patients (Bearse et al., 2020), cancer
patients (Jacobs et al., 2015), seafarers’, remote
communities in Greenland, and welfare claimants in
Northeast England (Jensen et al., 2020). Though these
groups represent relatively small sub-sets of the
broader population, they each nonetheless exhibit
distinct behavioural characteristics which might
prove useful in 1. generalized identification; and 2.
understanding variance across sub-populations. All
are characterised by transience i.e., people not
expected to remain a in lonely state in the long term.
They are, as Jensen et al. (2020) note, going through
CHIRA 2022 - 6th International Conference on Computer-Human Interaction Research and Applications
230
‘digital liminal’ states and thus the data generated by
these people is likely to experience sharp phase
transitions. Given these demographics remain more
likely to experience loneliness at some point than the
general population they serve as excellent case
studies for designing health surveillance to
understand broader prevalence statistics. Further
research is required to identify how digital footprints
align across liminal populations and whether
metadata can be ethically obtained, especially given
the precarious lives these groups experience.
3.1 Epistemological and
Methodological Tensions
Tension 1 - Sampling and Exclusion: As noted
earlier, research into loneliness and social isolation
has to date heavily emphasised student and elderly
populations. Some exceptions exist, but children and
those in the range 21-60 have seldom been considered
as research subjects for digital footprints and
loneliness a clear gap in the domain, made all-the-
more pressing due to changes in daily working
environments since COVID-19. Such an omission is
partly for good reason: students and the elderly are
stable populations, relatively accessible, and with
well reported challenges. Nonetheless, the corpus
excludes a huge portion of the average human life
course. Loneliness and social isolation do not, of
course, act across such neat demographics in practice.
A solution may be to develop encompassing
prevalence estimates derived from digital footprint
data.
Tension 2- Validity of Loneliness Measures: The
use of traditional scales, for example the UCLA or De
Jong Grieveld scales, may be anachronistic with the
forms of loneliness and social isolation occurring in
the modern world. Digital technology is enabling new
forms of rich and multi-faceted communication that
do not depend on being in the same place at the same
time. While the number ofcontacts we maintain has
increased due to technological innovations, has this
impacted on the shared social experience that
prevents loneliness? And is this divergence between
isolation and loneliness represented in the metrics
currently used? As most influential scales were
created prior to widespread Internet and smartphone
adoption they have little to say about the aspects of
social life now integral to work, play, and
socialisation. The notion of mixed modalities of
loneliness is something many authors note, but it is
rarely studied due to the absence of standardised
measures.
Tension 3 Clinical versus Non-clinical
Populations and Non-overlapping Data Sets: This
tension is best illustrated by the contrast between
genre 1 (the elderly) and genre 3 (online services and
their associated big data). In the former, researchers
have excellent access to a population known to be
more likely lonely and who also likely frequent
clinical settings. This demographic is therefore easier
to identify through formal methods, but they are
likely to have sparser digital footprints when
compared with the broader populace that could be
used for prevalence estimates. In contrast, online
services such as social media have rich behavioural
data illustrative of the interactions of lonely people
and often these data cut across multiple demographics
but pairing clinical measures is harder from a
methodological and ethical perspective. Here, again
care must still be taken to avoid exclusionary bias.
Machine learning models retain biases inherent to the
datasets they are trained with - and the availability of
digital footprint data (or lack thereof) in
representative communities such as the elderly must
be considered. As such hybrid approaches seem a
sensible solution – there is no one size fits all.
4 CONCLUSIONS
Via this scoping review, several tensions, research
gaps and opportunities have been evidenced.
Unexpectedly, it is gaps in the current literature, that
potentially offer the most insight, indicating the
impact digital footprints might have in improving
loneliness prevalence estimation and modernizing its
characterization. If leveraged responsibly,
behavioural datasets promise to advance
understanding for public health and policy and help
augment a domain that has historically been forced to
rely on coarse proxy data such as age and general or
limited clinical records. Social network, mobility, and
behavioral data all hold potential to support more-
encompassing prevalence statistics of loneliness. Yet
our scoping analysis highlights several challenges
and tensions therein, issues that require continued
scholarly attention. The following opportunities, if
invested in, may help attend to these challenges,
consolidate knowledge and advance maturity in the
field:
Opportunity 1 Sampling - The Need for
Longitudinal Study through the Life Course: To
make use of digital footprints researchers need to
ensure validity and efficacy, through ‘ground truths’
- labelling of loneliness in individuals, that can be
I Wandered Lonely in the Cloud: A Review of Loneliness, Social Isolation and Digital Footprint Data
231
paired to rich, observed behavioural data to determine
both indicators and antecedents. This is especially
relevant for middle-aged populations, who occur far
less in clinical contact records. Furthermore, inter-
and intra- person reliability measurements for
loneliness surveys are rarely conducted yet would be
highly valuable when pairing longitudinal digital
footprints for the purpose of identification and risk-
based prevalence estimates.
Opportunity 2 Validity - Studying Digital
Loneliness as its Own Experience: distinct from the
loneliness measured and identified via surveys such
as the UCLA, the field should encourage research of
extended multi-dimensional measures and
identification tools to recognise new experiences
across mixed modalities. Our contact lists are ever
extending however it is a lack of socially-shared
interactions which may be at the root of modern
experiences of loneliness.
Opportunity 3 – Make Data Useful, Open and
Transparent: As evidenced in Genre 3, much of the
data being studied by researchers is proprietary,
closed, and often commercial in nature.
Notwithstanding privacy concerns the need for open
data is a pre-requisite if the field is to develop. In
addition, only 3 papers include any spatial data, this
is surprisingly low because mobility/movement
tracking is often cited as a solution for loneliness
monitoring (particularly in the elderly). Broader
engagement of relevant communities, co-creation of
research studies, and focus on initiatives that engage
sufferers through open and transparent data sharing,
are required not only if we are to model loneliness
effectively but if we are to generate practical
interventions from model explanations, with real-
world impact. Kurt Vonnegut once reportedly said
‘What should young people do with their lives today?
Many things, obviously. But the most daring thing is
to create stable communities in which the terrible
disease of loneliness can be cured.’ To this we might
add, identifying and supporting those in need is a first
key step.
ACKNOWLEDGEMENTS
This article is dedicated to the memory of the primary
author who sadly died near completion of the
manuscript. Dr Reedman-Flint was supported by the
Horizon Centre for Doctoral Training at the
University of Nottingham (UKRI Grant No.
EP/L015463/1)
REFERENCES
Almohamed, A., & Vyas, D. (2016). Designing for the
Marginalized: A step towards understanding the lives
of refugees and asylum seekers. In Proceedings of the
2016 acm conference companion publication on
designing interactive systems (pp. 165-168).
Ananto, R. A., & Young, J. E. (2021). We can do better! an
initial survey highlighting an opportunity for more HRI
Work on loneliness. In Companion of the 2021
ACM/IEEE International Conference on Human-Robot
Interaction (pp. 457-462).
Austin, J., Dodge, H. H., Riley, T., Jacobs, P. G., Thielke,
S., & Kaye, J. (2016) A smart-home system to
unobtrusively and continuously assess loneliness in
older adults. IEEE journal of translational engineering
in health and medicine, 4, 1-11.
Baecker, R., Sellen, K., Crosskey, S., Boscart, V., &
Barbosa Neves, B. (2014). Technology to reduce social
isolation and loneliness. In Proceedings of the 16th
international ACM SIGACCESS conference on
Computers & accessibility (pp. 27-34).
Bacciu, D., Chessa, S., Ferro, E., Fortunati, L., Gallicchio,
C., La Rosa, D., ... & Vozzi, F. (2016). Detecting
socialization events in ageing people: The experience
of the doremi project. In 2016 12th International
Conference on Intelligent Environments (IE) (pp. 132-
135). IEEE
BBC (2019) BBC Radio 4 - The Anatomy Of Loneliness -
Who Feels Lonely? [Online] Available at:
https://bbc.in/3NOjIey [Accessed 15 May 2021].
Bearse, P., Manejwala, O., Mohammad, A. F., & Haque, I.
R. I. (2020). An Initial Feasibility Study to Identify
Loneliness Among Mental Health Patients from
Clinical Notes. In 2020 3rd International Conference on
Information and Computer Technologies (ICICT) (pp.
68-77). IEEE.
Broadbent, E., Ahn, H. S., Kerse, N., Peri, K., Sutherland,
C., Law, M., ... & Pandey, A. K. (2018). Can robots
improve the quality of life in people with dementia?. In
Proceedings of the Technology, Mind, and Society (pp.
1-3).
Brueckner, S. (2020). Captured by an Algorithm. In ACM
SIGGRAPH 2020 Art Gallery (pp. 458-459).
Bu, F., Steptoe, A., & Fancourt, D. (2020). Who is lonely
in lockdown? Cross-cohort analyses of predictors of
loneliness before and during the COVID-19 pandemic.
Public Health, 186, 31-34.
Burke, M., Marlow, C., & Lento, T. (2010). Social network
activity and social well-being. In Proceedings of the
SIGCHI conference on human factors in computing
systems (pp. 1909-1912).
Cacioppo, J. T., & Patrick, W. (2008). Loneliness: Human
nature and the need for social connection. WW Norton
& Company.
Chang, A. S., & Kalawsky, R. S. (2017). Future
configurable transport for the ageing population. In
2017 7th International Conference on Power
Electronics Systems and Applications-Smart Mobility,
Power Transfer & Security (PESA) (pp. 1-5). IEEE.
CHIRA 2022 - 6th International Conference on Computer-Human Interaction Research and Applications
232
De Choudhury, M., Counts, S., Horvitz, E. J., & Hoff, A.
(2014). Characterizing and predicting postpartum
depression from shared facebook data. In Proceedings
of the 17th ACM conference on Computer supported
cooperative work & social computing (pp. 626-638).
de Jong-Gierveld, J. (1987). Developing and testing a
model of loneliness. Journal of personality and social
psychology, 53(1), 119.
DiTommaso, E., & Spinner, B. (1993). The development
and initial validation of the Social and Emotional
Loneliness Scale for Adults (SELSA). Personality and
individual differences, 14(1), 127-134.
Dunbar, R., 2021. Friends: Understanding the power of our
most important relationships. Hachette UK.
Eldib, M., Deboeverie, F., Haerenborgh, D. V., Philips, W.,
& Aghajan, H. (2015). Detection of visitors in elderly
care using a low-resolution visual sensor network. In
Proceedings of the 9th International Conference on
Distributed Smart Cameras (pp. 56-61).
Elovainio, M., Hakulinen, C., Pulkki-Råback, L., Virtanen,
M., Josefsson, K., Jokela, M., ... & Kivimäki, M.
(2017). Contribution of risk factors to excess mortality
in isolated and lonely individuals: an analysis of data
from the UK Biobank cohort study. The Lancet Public
Health, 2(6), e260-e266.
Ernst, M., Niederer, D., Werner, A. M., Czaja, S. J., Mikton,
C., Ong, A. D., ... & Beutel, M. E. (2022). Loneliness
before and during the COVID-19 pandemic: A
systematic review with meta-analysis. American
Psychologist.
Eyssel, F., & Reich, N. (2013). Loneliness makes the heart
grow fonder (of robots)—On the effects of loneliness
on psychological anthropomorphism. In 2013 8th
acm/ieee international conference on human-robot
interaction (hri) (pp. 121-122). IEEE.
Ferrer, M., Mancha, V., Chumpitaz, M., Begazo, J., &
Chauca, M. (2020). Characterization of the impact of
SARS-CoV-2 pandemic social isolation on the
psychosocial well-being of public university students,
based on the GHQ-28 Scale. In Proceedings of the 4th
International Conference on Medical and Health
Informatics (pp. 295-301).
Fuentes, C., Rodríguez, I., & Herskovic, V. (2016). Making
Communication Frequency Tangible: How Green Is
My Tree?. In Proceedings of the TEI'16: Tenth
International Conference on Tangible, Embedded, and
Embodied Interaction (pp. 434-440).
Galunder, S. S., Gottlieb, J. F., Ladwig, J., Hamell, J.,
Keller, P. K., & Wu, P. (2018). A VR ecosystem for
telemedicine and non-intrusive cognitive and affective
assessment. In 2018 IEEE 6th International Conference
on Serious Games and Applications for Health
(SeGAH) (pp. 1-6). IEEE.
Gao, X., Zhang, C., Ma, L., Wang, Y., Wang, J., & Zhang,
D. (2019). Correlating msm's mental health with usage
behaviors on msm-specific social applications. In 2019
IEEE SmartWorld, Ubiquitous Intelligence &
Computing, Advanced & Trusted Computing, Scalable
Computing & Communications, Cloud & Big Data
Computing, Internet of People and Smart City
Innovation IEEE.
Griffin, J. (2010). The lonely society?. Mental Health
Foundation.
Ha, T. V., & Hoang, D. B. (2017). An assistive healthcare
platform for both social and service networking for
engaging elderly people. In 2017 23rd Asia-Pacific
Conference on Communications (APCC) (pp. 1-6).
IEEE.
Harvey, J., Golightly, D., & Smith, A. (2014). HCI as a
means to prosociality in the economy. In Proceedings
of the SIGCHI Conference on Human Factors in
Computing Systems (pp. 2955-2964).
Hertz, N. (2021). The lonely century: how to restore human
connection in a world that's pulling apart. Currency.
Holt-Lunstad, J. (2017). The potential public health
relevance of social isolation and loneliness: Prevalence,
epidemiology, and risk factors. Public Policy & Aging
Report, 27(4), 127-130.
Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010).
Social relationships and mortality risk: a meta-analytic
review. PLoS medicine, 7(7), e1000316.
Ibarra, F., Baez, M., Cernuzzi, L., & Casati, F. (2020). A
systematic review on technology-supported
interventions to improve old-age social wellbeing:
loneliness, social isolation, and connectedness. Journal
of healthcare engineering, 2020.
Jacobs, M. L., Clawson, J., & Mynatt, E. D. (2015).
Comparing health information sharing preferences of
cancer patients, doctors, and navigators. In Proceedings
of the 18th ACM Conference on Computer Supported
Cooperative Work & Social Computing (pp. 808-818).
Jensen, R. B., Coles-Kemp, L., Wendt, N., & Lewis, M.
(2020). Digital liminalities: Understanding isolated
communities on the edge. In Proceedings of the 2020
CHI Conference on Human Factors in Computing
Systems (pp. 1-14).
Jeong, E. J., Kim, D. J., & Lee, D. M. (2015). Game
addiction from psychosocial health perspective. In
Proceedings of the 17th International Conference on
Electronic Commerce 2015 (pp. 1-9).
Jeong, E. J., Kim, D. J., Lee, D. M., & Lee, H. R. (2016). A
study of digital game addiction from aggression,
loneliness and depression perspectives. In 2016 49Th
Hawaii International Conference On System Sciences
(HICSS) (pp. 3769-3780). IEEE.
Joseph, K., Carley, K. M., & Hong, J. I. (2014). Check-ins
in “Blau Space” Applying Blau’s Macrosociological
Theory to Foursquare Check-ins from New York City.
ACM Transactions on Intelligent Systems and
Technology (TIST), 5(3), 1-22.
Joyner, D. A., Wang, Q., Thakare, S., Jing, S., Goel, A., &
MacIntyre, B. (2020). The synchronicity paradox in
online education. In Proceedings of the Seventh ACM
Conference on Learning@ Scale (pp. 15-24).
Kaur, S., Kaul, P., & Zadeh, P. M. (2020). Study the impact
of covid-19 on twitter users with respect to social
isolation. In 2020 Seventh International Conference on
Social Networks Analysis, Management and Security
(SNAMS) (pp. 1-6). IEEE.
I Wandered Lonely in the Cloud: A Review of Loneliness, Social Isolation and Digital Footprint Data
233
Kindness, P., Mellish, C., & Masthoff, J. (2013). How
virtual teammate support types affect stress. In 2013
Humaine Association Conference on Affective
Computing and Intelligent Interaction (pp. 300-305).
IEEE.
Knox, S. S., & Uvnäs-Moberg, K. (1998). Social isolation
and cardiovascular disease: an atherosclerotic
pathway?. Psychoneuroendocrinology, 23(8), 877-890.
Lazányi, K. (2016). Investing in social support—Robots as
perfect partners?. In 2016 IEEE 14th International
Symposium on Intelligent Systems and Informatics
(SISY) (pp. 25-30). IEEE.
Li, S., Ni, S., & Peng, K. (2020a). The Role of Good Human
Uniqueness in Social Robot Anthropomorphism
Influenced by Chronic Loneliness. In 2020 IEEE
International Conference on Human-Machine Systems
(ICHMS) (pp. 1-6). IEEE.
Li, S., Xu, L., Yu, F., & Peng, K. (2020b). Does trait
loneliness predict rejection of social robots? In
Proceedings of the 2020 ACM/IEEE International
Conference on Human-Robot Interaction (pp. 271-
280).
Light, A., Howland, K., Hamilton, T., & Harley, D. A.
(2017). The meaing of place in supporting sociality. In
Proceedings of the 2017 Conference on Designing
Interactive Systems (pp. 1141-1152).
Lu, X., & Yao, J. (2010). The Influence of Internet
Interpersonal Communication to Relationship and
Loneliness of College Students. In 2010 International
Conference on Web Information Systems and Mining
(Vol. 2, pp. 386-390). IEEE.
Lou, C., Zhao, J., Li, X., Wei, H., Zhang, Y., & Zhao, H.
(2019). Pet Robot Emotional Interaction for Urban
Autism. In Proceedings of the 2019 2nd International
Conference on Intelligent Science and Technology (pp.
1-6).
Lubben, J. (2017). Addressing social isolation as a potent
killer!. Public Policy & Aging Report, 27(4), 136-138.
Marangoni, C., & Ickes, W. (1989). Loneliness: A
theoretical review with implications for measurement.
Journal of Social and Personal Relationships, 6(1), 93-
128.
Martinez, A., Ortiz, V., Estrada, H., & Gonzalez, M. (2017)
A predictive model for automatic detection of social
isolation in older adults. In 2017 International
Conference on Intelligent Environments (IE) (pp. 68-
75). IEEE.
Matthews, T., Danese, A., Wertz, J., Odgers, C. L., Ambler,
A., Moffitt, T. E., & Arseneault, L. (2016). Social
isolation, loneliness and depression in young
adulthood: a behavioural genetic analysis. Social
psychiatry and psychiatric epidemiology, 51(3), 339-
348.
Mulvenna, M., Zheng, H., Bond, R., McAllister, P., Wang,
H., & Riestra, R. (2017). Participatory design-based
requirements elicitation involving people living with
dementia towards a home-based platform to monitor
emotional wellbeing. In 2017 IEEE International
Conference on Bioinformatics and Biomedicine
(BIBM) (pp. 2026-2030). IEEE.
Munn, Z., Peters, M. D., Stern, C., Tufanaru, C., McArthur,
A., & Aromataris, E. (2018). Systematic review or
scoping review? Guidance for authors when choosing
between a systematic or scoping review approach.
BMC medical research methodology, 18(1), 1-7.
Noguchi, Y., Kamide, H., & Tanaka, F. (2018). Effects on
the self-disclosure of elderly people by using a robot
which intermediates remote communication. In 2018
27th IEEE International Symposium on Robot and
Human Interactive Communication (RO-MAN) (pp.
612-617). IEEE.
ONS (2018) Loneliness [Online] available at:
https://bit.ly/3zd10ZL [Accessed 15 May 2021]
Pedell, S., Vetere, F., Kulik, L., Ozanne, E., & Gruner, A.
(2010). Social isolation of older people: the role of
domestic technologies. In Proceedings of the 22nd
Conference of the Computer-Human Interaction
Special Interest Group of Australia on Computer-
Human Interaction (pp. 164-167).
Perlman, D., & Peplau, L. A. (1981). Toward a social
psychology of loneliness. Personal relationships, 3, 31-
56.
Petersen, J., Austin, D., Kaye, J. A., Pavel, M., & Hayes, T.
L. (2013). Unobtrusive in-home detection of time spent
out-of-home with applications to loneliness and
physical activity. IEEE journal of biomedical and
health informatics, 18(5), 1590-1596.
Pulekar, G., & Agu, E. (2016). Autonomously sensing
loneliness and its interactions with personality traits
using smartphones. In 2016 IEEE healthcare innovation
point-of-care technologies conference (HI-POCT) (pp.
134-137). IEEE.
Rabani, S. T., Khan, Q. R., & Khanday, A. M. U. D. (2020).
Multi-Class Suicide Risk Prediction on Twitter Using
Machine Learning Techniques. In 2020 2nd
International Conference on Advances in Computing,
Communication Control and Networking (ICACCCN)
(pp. 128-134). IEEE.
Ring, L., Barry, B., Totzke, K., & Bickmore, T. (2013)
Addressing loneliness and isolation in older adults:
Proactive affective agents provide better support. In
2013 Humaine Association conference on affective
computing and intelligent interaction (pp. 61-66).
IEEE.
Russell, D. W. (1996). UCLA Loneliness Scale (Version 3):
Reliability, validity, and factor structure. Journal of
personality assessment, 66(1), 20-40.
Russell, D., Cutrona, C. E., Rose, J., & Yurko, K. (1984).
Social and emotional loneliness: an examination of
Weiss's typology of loneliness. Journal of personality
and social psychology, 46(6), 1313.
Scalise, J. J., Ginter, E. J., & Gerstein, L. H. (1984).
Multidimensional loneliness measure: the loneliness
rating scale (LRS). Journal of Personality Assessment,
48(5), 525-530.
Schmidt, N., & Sermat, V. (1983). Measuring loneliness in
different relationships. Journal of personality and social
psychology, 44(5), 1038.
Senese, V. P., Nasti, C., Mottola, F., Sergi, I., & Gnisci, A.
(2020, September). Validation and measurement
CHIRA 2022 - 6th International Conference on Computer-Human Interaction Research and Applications
234
invariance across gender and age of the Italian
Interpersonal Acceptance-Rejection Loneliness Scale.
In 2020 11th IEEE International Conference on
Cognitive Infocommunications (CogInfoCom) (pp.
000289-000294). IEEE
Shah, S. G. S., Nogueras, D., Van Woerden, H., &
Kiparoglou, V. (2019). Effectiveness of digital
technology interventions to reduce loneliness in adults:
a protocol for a systematic review and meta-analysis.
BMJ open, 9(9), e032455.
Steptoe, A., Shankar, A., Demakakos, P., & Wardle, J.
(2013). Social isolation, loneliness, and all-cause
mortality in older men and women. Proceedings of the
National Academy of Sciences, 110(15), 5797-5801.
Taylor, S. H., Hutson, J. A., & Alicea, T. R. (2017). Social
consequences of Grindr use: Extending the internet-
enhanced self-disclosure hypothesis. In Proceedings of
the 2017 CHI Conference on Human Factors in
Computing Systems (pp. 6645-6657).
Waterworth, J. A., Ballesteros, S., & Peter, C. (2009). User-
sensitive home-based systems for successful ageing. In
2009 2nd Conference on Human System Interactions
(pp. 542-545). IEEE.
Weil, S. (1997). Gravity and grace. U of Nebraska Press
Weinert, C., Maier, C., Laumer, S., & Weitzel, T. (2014).
Does teleworking negatively influence IT
professionals? An empirical analysis of IT personnel's
telework-enabled stress. In Proceedings of the 52nd
ACM conference on Computers and people research
(pp. 139-147).
Weiss, R. (1975). Loneliness: The experience of emotional
and social isolation. MIT press.
Wu, Y., Yuan, J., You, Q., & Luo, J. (2016). The effect of
pets on happiness: A data-driven approach via large-
scale social media. In 2016 IEEE International
Conference on Big Data (Big Data) (pp. 1889-1894).
IEEE.
Xu, D., Qian, L., Wang, Y., Wang, M., Shen, C., Zhang, T.,
& Zhang, J. (2015). Understanding the dynamic
relationships among interpersonal personality
characteristics, loneliness, and smart-phone use:
evidence from experience sampling. In 2015
International Conference on Computer Science and
Mechanical Automation (CSMA) (pp. 19-24). IEEE.
Yang, H., & Bath, P. A. (2018). Prediction of Loneliness in
Older People. In Proceedings of the 2nd International
Conference on Medical and Health Informatics (pp.
165-172).
Yoshida, M., Kleisarchaki, S., Gtirgen, L., & Nishi, H.
(2018). Indoor occupancy estimation via location-
aware HMM: An IoT approach. In 2018 IEEE 19th
International Symposium on" A World of Wireless,
Mobile and Multimedia Networks"(WoWMoM) (pp.
14-19). IEEE.
Zadeh, P. M., Khani, S., Pfaff, K., & Samet, S. (2020). A
computational model and algorithm to identify social
isolation in elderly population. In 2020 IEEE
Symposium on Computers and Communications
(ISCC) (pp. 1-6). IEEE.
Zhang, M. (2010). Exploring adolescent peer relationships
online and offline: an empirical and social network
analysis. In 2009 WRI International Conference on
Communications and Mobile Computing (Vol. 3, pp.
268-272). IEEE.
Zhou, Z., Sun, X., Chen, X., Fan, C., & Pan, Q. (2009). The
Relation between Different Constructs in Middle
Childhood Peer Interaction and Loneliness: A
Mediational Model. In 2009 3rd International
Conference on Bioinformatics and Biomedical
Engineering (pp. 1-4). IEEE.
Zhou, Q., Li, J., Tang, Y., & Wang, H. (2020). Discovering
the Lonely Among the Students with Weighted Graph
Neural Networks. In 2020 IEEE 32nd International
Conference on Tools with Artificial Intelligence
(ICTAI) (pp. 474-481). IEEE.
I Wandered Lonely in the Cloud: A Review of Loneliness, Social Isolation and Digital Footprint Data
235