IEEE Technology and Society Magazine - March 2023 - 28

Feature
mental health applications-those that are not used
in conjunction with other mental health services-
have even more mixed outcomes. In a systematic
review and meta-analysis of standalone applications
for depression, anxiety, substance use, self-injury,
and sleep problems conducted in 2019 [28], the only
significant positive effects noted were for applications
addressing depression, with some indications
for efficacy in applications for smoking cessation
and sleep problems. These effects were relatively
small, and the authors warn against drawing generalizable
conclusions. Furthermore, as they note, there
is a need to discuss the " potential harm of currently
available apps, which might keep users away from
evidence-based interventions while bearing the substantial
risk of being ineffective " [28, p. 118].
Previous literature in clinical medical research is
sober in its claims about the prospect of pointing to " an
app for that " [29] as a silver bullet solution for addressing
mental health concerns. We are at the developmental
stages of an experimental intervention that has the
propensity to markedly shift how people access care,
from who to get help from to what to get help from [69].
Evidence around the safety and efficacy of chatbots or
AI conversational assistants for mental health is even
more limited [30]. A 2020 systematic review of 12 studies
found conflicting results [31]. While there was some
evidence supporting the effectiveness of chatbots for
improving symptoms of some conditions, including
depression and stress, evidence was either not statistically
significant or conflicting regarding psychological
well-being and anxiety. Furthermore, statistical significance,
the reviews note, does not equate to clinical significance.
In other words, measures of efficacy in these
studies may not translate into meaningful differences
in practice. Making unfounded claims now about app
capabilities, especially more advanced AI chatbots,
would be irresponsible.
Some researchers may be optimistic about the
use of digital mental health apps, with commercial
entities, in particular, touting a range of benefits for
self-help services [32], such as the convenience of
on-demand access to resources [33], the ability to
address a range of mental health concerns holistically
using a single application, and the potential to
experience improved sleep and a reduction in stress
and anxiety. However, it is important to state that
these claims are often made prior to clinical validation.
In other words, it is highly premature to declare
the apps a " win " for those living with mental health
28
conditions. There may be some potential for apps, to
address the mental health crisis, but a sober research
perspective is necessary as in [34]. Our assessments
of the practical applications of these digital mental
health apps among diverse populations must be evidence
based.
The stakes related to mental health mean that
it is risking lives to start declaring AI-based mental
health apps successful without a proven clinical
basis-even when everything else in the wider society
seems to be moving toward a digital and mobile
interface. Likewise, digital transformation does not
equate to commensurate social transformation and
certainly not health-related transformation even if
it might well revolutionize how we live and receive
personalized health and well-being solutions. Fundamentally,
as Abd-Alrazaq et al. [31] warn, chatbots
should augment, not replace, traditional healthcare
provisions. However, while mental health applications
may be intended to be used as supplements
rather than replacements, the shortage of clinicians
noted above may lead to this scenario regardless of
designer intention. But we also need clinicians who
are well-trained in the strengths and weaknesses of
these kinds of applications.
Furthermore, we must ask some fundamental
questions about where this technology is potentially
leading us, and whether that direction is where we
want to go. We must ask more pertinent questions
related to the root causes of this mental health crisis.
Certainly, it is advisable, before we commit to technological
solutions, that we examine the possible
fallout of such technologies. It is not enough to build
frameworks for evaluation (though we obviously
need those [35]), conduct systematic reviews and
meta-analyses, or to engage users in clinical surveys,
and commensurate focus group panels.
We must dig deeper into the fundamental philosophical
question of what it means to be human and
our collective responsibility to respond to people in
need. Might misguided hype or excitement about
these possible mental health " tech fixes " conveniently
serve to relieve pressure on governments for expending
resources on social services such as preventative
mental health efforts, or on evidence-based intervention
strategies and support? Will the misguided lure of
supposedly easier, faster-but unproven-tech-based
services cause attention to shift from more robust
efforts to solve these problems, or from finding ways
to attract more talent into the fields of psychology,
IEEE Technology and Society Magazine
IEEE Technology and Society Magazine - March 2023

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - March 2023
