Cited by 164Digital Mental Health InterventionsAnxiety, Depression, Psychometrics, Treatment, Cognitive ProcessesSocial Robot Interaction and HRI

Chatbot-Delivered Psychotherapy for Adults With Depressive and Anxiety Symptoms: A Systematic Review and Meta-Regression

Lim, Shi Min, Shiau, Chyi Wey Claudine, Cheng, Ling Jie, Lau, Ying·2022·Behavior Therapy

Abstract

Although psychotherapy is a well-established treatment for depression and anxiety, chatbot-delivered psychotherapy is an emerging field that has yet to be explored in depth. This review aims to (a) examine the effectiveness of chatbot-delivered psychotherapy in improving depressive symptoms among adults with depression or anxiety, and (b) evaluate the preferred features for the design of chatbot-delivered psychotherapy. Eight electronic databases were searched for relevant randomized controlled trials. Meta-analysis and random effects meta-regression was conducted using Comprehensive Meta-Analysis 3.0 software. Overall effect was measured using Hedges's g and determined using z statistics at significance level p < .05. Assessment of heterogeneity was done using χ2 and I2 tests. A meta-analysis of 11 trials revealed that chatbot-delivered psychotherapy significantly improved depressive symptoms (g = 0.54, 95% confidence interval [-0.66, -0.42], p < .001). Although no significant subgroup differences were detected, results revealed larger effect sizes for samples of clinically diagnosed anxiety or depression, chatbots with an embodiment, a combination of types of input and output formats, less than 10 sessions, problem-solving therapy, off-line platforms, and in different regions of the United States than their counterparts. Meta-regression did not identify significant covariates that had an impact on depressive symptoms. Chatbot-delivered psychotherapy can be adopted in health care institutions as an alternative treatment for depression and anxiety. More high-quality trials are warranted to confirm the effectiveness of chatbot-delivered psychotherapy on depressive symptoms. PROSPERO registration number: CRD42020153332.

Integrity signals

Registry

Crossref · 10.1016/j.beth.2021.09.007

Cites

84 of 84 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Chatbot eHealth meta-analyses to date have focused on specific health conditions, resulting in few studies focusing on comparing primary health effects.

    “Chatbot eHealth meta-analyses to-date have focused on specific health conditions resulting in few studies focusing on comparing primary health effects [37, 38].”
  • Previous reviews limited their focus to a subset of conversational agents based on the embodiment level.

    “Third, the previous reviews limited their focus to a subset of CAs based on the embodiment level, such as disembodied CAs13,20, CAs with virtual representation15, or with a physical representation23–25.”
  • Most previous reviews did not distinguish fully automated conversational agents from non- or semi-autonomous conversational agents.

    “Second, most of the previous reviews did not distinguish fully automated CAs from non- or semi-autonomous CAs13,20.”
  • A previous review showed medium to large effects of automated CA interventions when conducted on adults with clinical levels of anxiety or depression.

    “Indeed, when conducted on adults with clinical levels of anxiety or depression, a previous review showed medium to large effects of automated CAs interventions13.”
  • Rule-based AI chatbots show promise to address some of these limitations, by simulating human conversation using predefined scripts and algorithms such as decision trees, to deliver the benefits of DMHIs in a more dynamic and interactive way.

    “Rule-based AI chatbots show promise to address some of these limitations, by simulating human conversation using predefined scripts and algorithms such as decision trees, to deliver the benefits of DMHIs in a more dynamic and interactive way13,14.”

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Lim, S. M., Shiau, C. W. C., Cheng, L. J., & Lau, Y.
Affiliations
  • Tan Tock Seng Hospital (SG)
  • National University of Singapore (SG)
Corresponding author
Lau, Ying
Venue
Behavior Therapy
Publisher
Elsevier BV
Published
2022-03
Volume
53
Issue
2
Pages
334-347
ISSN
0005-7894 (ISSN-L)
OpenAlex
W4229082340
Cited by
163
References
84
Citations by year
2026: 35 · 2025: 47 · 2024: 49 · 2023: 26 · 2022: 6
FWCI
3.15
Type
journal article
Language
en
Open access
closed
Locations
  • Behavior Therapy (publishedVersion)
  • PubMed (publishedVersion)
Topics
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Anxiety, Depression, Psychometrics, Treatment, Cognitive Processes
  • Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
Keywords
depression, anxiety, depressive symptoms, problem-solving therapy, digital mental health interventions, treatment effectiveness
MeSH terms
Adult, Anxiety, Anxiety Disorders, Depression, Humans, Psychotherapy, Software

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review