The application of chatbots in perinatal mental health interventions: a scoping review protocol

Abstract

Introduction

Perinatal mental health problems are prevalent during pregnancy and the postpartum period and can have substantial consequences for both mothers and infants. Chatbot-based interventions represent a promising and scalable approach to perinatal mental health support. However, the evidence base remains heterogeneous and has not been systematically mapped. This scoping review aims to synthesise existing research on chatbot-based interventions for perinatal mental health, focusing on study characteristics, intervention features, outcomes and implementation challenges.

Methods and analysis

This scoping review protocol follows the Preferred Reporting Items for Systematic Reviews and Meta-analyses Protocols guidelines. A systematic search will be conducted from database inception to May 2026 to identify quantitative, qualitative and mixed-methods studies focusing on chatbot-based interventions for perinatal mental health. Two reviewers will independently screen studies and extract data, with discrepancies resolved through discussion or consultation with a third reviewer. Data will be synthesised descriptively.

Ethics and dissemination

This study is based on publicly available literature and does not require ethical approval. The findings will be disseminated through peer-reviewed journal publications and presented at academic conferences related to perinatal and child health and digital health.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This scoping review will use a comprehensive and systematic search strategy to map evidence on chatbot-based interventions for perinatal mental health.

• Multiple study designs will be included to capture a broad range of intervention approaches and real-world applications.

• Data extraction and reporting will follow established scoping review frameworks and Preferred Reporting Items for Systematic Reviews and Meta-analyses Protocols guidelines.

• Ethical and implementation issues related to artificial intelligence-based interventions will be explored.

• The quality of included studies will not be assessed, as this is beyond the scope of a scoping review.

Introduction

Pregnancy and the postpartum period are critical phases marked by substantial physiological, psychological and social transitions, during which women may be particularly vulnerable to mental health problems.¹ Common perinatal mental health problems include depression, anxiety and related disorders, with a substantial global burden that appears to be higher in low- and middle-income countries.^{2 3} Women experiencing depression or anxiety during pregnancy are more likely to experience postpartum mental health disorders and are at increased risk of adverse obstetric and neonatal outcomes, including preterm birth, low birth weight and intrauterine growth restriction, as well as suicidal behaviours.^4–6 Perinatal mental health problems may also negatively affect offspring social-emotional, cognitive, language, motor and adaptive development.⁷ Recognising these widespread impacts, international guidelines have emphasised perinatal mental health as a global public health priority.^8–11 The WHO specifically recommends integrating perinatal mental health screening and interventions into routine perinatal and child health services to improve outcomes for both mothers and infants. These recommendations underscore a growing global consensus that perinatal mental health is an essential component of overall public health.

Although several approaches have been developed to manage perinatal mental health, existing interventions remain limited in accessibility and scalability. Cognitive–behavioural therapy (CBT) has been delivered in individual, group, telephone and online formats to help women identify and modify maladaptive thoughts and behaviours.⁵ Interpersonal psychotherapy (IPT) focuses on enhancing social support and resolving interpersonal conflicts.¹² Traditional interventions for perinatal mental health such as CBT, IPT, psychological counselling and pharmacological treatment have been shown effective in relevant studies.¹³ However, in real-world settings, the accessibility of these interventions remains limited, as perinatal women may avoid seeking psychological counselling due to stigma,¹⁴ privacy concerns or fear of exposing family conflicts. In addition, perinatal women may prioritise the needs of other family members over their own mental health.^{15 16} Beyond individual-level factors, access may also be constrained by health-system and structural barriers, such as shortages of adequately trained professionals, unclear referral pathways, treatment costs and broader social and cultural inequities.^{17 18} Therefore, there is an urgent need to develop innovative delivery models that can expand the reach of established, evidence-based interventions such as CBT and IPT. Within this context, chatbot-based approaches represent an emerging mode of intervention delivery, offering new opportunities to improve the accessibility, personalisation and continuity of perinatal mental healthcare.

A chatbot is a software programme that interacts with users through text or voice and generates responses using natural language processing (NLP).^19–21 Chatbots can provide all-day support for perinatal women, generate dynamic feedback based on users’ emotions and needs and may serve as a feasible and acceptable approach to expanding access to mental health support.^{22 23} In addition, chatbot-based interventions may help alleviate pressure on limited healthcare resources through scalable support, although evidence on their cost-effectiveness remains limited.^{24 25} In recent years, several studies have explored chatbot-based interventions for perinatal mental health and provided preliminary evidence regarding their implementation and potential benefits. For example, Suharwardy et al²⁶ conducted a randomised controlled trial (RCT) on a mental health chatbot designed for postpartum women, which demonstrated high acceptability among participants and preliminary efficacy in reducing depressive symptoms compared with usual care.²⁶ Similarly, Nguyen et al²⁷ evaluated Rosie, a health education chatbot specifically tailored for women of colour during pregnancy and the postpartum period, through a randomised pilot study. The study confirmed the feasibility of the chatbot and observed favourable trends in perinatal mental health outcomes and infant healthcare utilisation compared with the control group.²⁷ Additionally, Chua et al²⁸ tested Parentbot, a comprehensive digital healthcare assistant, in an RCT involving heterosexual couples in Singapore. The results showed that Parentbot could enhance parenting self-efficacy and reduce perinatal anxiety.²⁸ As chatbot capabilities continue to advance, they are expected to play an increasingly vital role in healthcare, optimising women’s pregnancy and childbirth experiences and helping both mothers and their infants achieve their full health and well-being potential.^29–32

To date, evidence on chatbot-based interventions for perinatal mental health remains fragmented, with considerable variability in study designs, geographical settings, participant characteristics, intervention components and outcome measures. In addition, findings regarding the feasibility, acceptability, effectiveness and implementation challenges of these interventions have not yet been systematically synthesised. To address these gaps, this protocol outlines a scoping review that will systematically map the existing literature on chatbot-based interventions for perinatal mental health. Specifically, the review is designed to describe the characteristics of current studies, including study designs, geographical settings and participant populations; to examine how chatbots have been applied to support perinatal mental health during the perinatal period in terms of functions, delivery formats and therapeutic components; to summarise reported outcomes related to depression, anxiety, user engagement and acceptability; and ultimately to identify reported advantages, limitations and implementation challenges. By adhering to this predefined plan, the findings will provide a comprehensive overview of the current evidence base. Ultimately, this work is intended to help inform future research as well as the development of digital mental health strategies for perinatal populations.

Methods

This scoping review protocol has been registered on the Open Science Framework (DOI: 10.17605/OSF.IO/C39P4).

This scoping review will be conducted in accordance with the methodology framework proposed by Arksey and O’Malley,³³ which involves five key stages: identifying the research question, identifying relevant studies, study selection, charting the data and collating, summarising and reporting the results. This scoping review protocol follows the Preferred Reporting Items for Systematic Reviews and Meta-analyses Protocols (PRISMA-P) guidelines (PRISMA-P checklist).³⁴

Stage 1: identifying the research question

The questions of this review will be as follows:

• What study designs, geographical settings and participant characteristics have been reported in existing research on chatbot-based interventions for perinatal mental health?

• How have chatbots been applied to support perinatal mental health, in terms of their functions, delivery formats and therapeutic components?

• What perinatal mental health outcomes have been assessed and what effects on depression, anxiety, user engagement and acceptability have been reported?

• What advantages, limitations and implementation challenges related to chatbot-based interventions have been identified across studies?

Together, these questions aim to systematically map the scope, characteristics and emerging evidence base of chatbot interventions in perinatal mental health. The systematic literature search will be conducted in May 2026. The databases will be searched from database inception to May 2026. Following completion of the literature search, study selection, data extraction and evidence synthesis will be undertaken.

Stage 2: identifying relevant studies

Following the systematic search strategy outlined in the Cochrane Handbook, a comprehensive search will be conducted across both Chinese and international databases using a combination of subject headings and free-text terms. The databases to be searched will include¹: Chinese databases: Chinese Biomedical Literature Database, Chinese National Knowledge Infrastructure, Wan Fang Database, Chinese Science and Technology Periodical Database²; English databases: PubMed, Medline, Embase, Cochrane Central Register of Controlled Trials, Cumulated Index to Nursing and Allied Health Literature, Psychological Information, Web of Science, Institute of Electrical and Electronics Engineers Xplore and Scopus.

In addition, a snowballing approach will be applied by manually screening the reference lists of the included studies to identify any relevant literature not captured in the database search.

A systematic search will be conducted from database inception to May 2026. The search terms will cover concepts related to conversational agents (eg, chatbots, virtual assistants, dialogue systems), maternal or perinatal populations (eg, pregnancy, postpartum, perinatal period) and mental health outcomes (eg, depression, anxiety, stress). The search strategy will be developed through an iterative process, including preliminary scoping of the literature, identification of relevant keywords and controlled vocabulary (eg, Medical Subject Headings (MeSH) terms) and refinement of search terms in consultation with the research team. The Chinese and English search strings will be further refined according to the characteristics of each database. Full search strategies for all databases will be provided in the supplementary file (online supplemental material).

All retrieved records will be imported into EndNote V.21 for deduplication, after which the cleaned dataset will be uploaded to Rayyan for title and abstract screening, followed by full-text review.

Stage 3: study selection

The inclusion criteria for this study were developed based on the population–concept–context framework¹: population: pregnant or postpartum women (within 12 months postpartum), aged 18 years or older, with no history of psychiatric disorders or severe physical comorbidities²; concept: psychological interventions delivered via chatbots equipped with NLP technology that enable bidirectional communication. Eligible studies will include quantitative, qualitative and mixed-methods designs. Outcomes of interest will include mental health-related measures (eg, depression, anxiety, stress), as well as secondary outcomes such as intervention adherence, engagement, user satisfaction or acceptability, feasibility metrics, maternal and infant clinical outcomes and any reported adverse events or safety concerns³; context: mental health interventions conducted in healthcare institutions, community settings, home environments or online platforms, without restrictions on country or cultural background. Studies published in English and Chinese from database inception to May 2026 will be included.

Exclusion criteria: (1) interventions lacking natural language interaction functions; (2) non-original research, including reviews, commentaries, poster presentations or theoretical papers. Conference papers with complete methods and data are eligible, whereas abstracts or poster-only reports will be excluded unless full texts can be obtained from authors; studies remaining without full texts will be recorded as ‘unretrieved’; (3) studies with unextractable data, duplicate publications or unavailable full texts.

After duplicate records are removed in EndNote V.21, the deduplicated dataset will be imported into Rayyan for screening. Study selection will be conducted in two stages: title and abstract screening, followed by full-text review. Two reviewers will independently complete both stages. Prior to formal screening, a pilot screening of a sample of studies will be conducted to ensure consistent application of the eligibility criteria. Any disagreements will be resolved through discussion or consultation with a third reviewer.

During the initial screening phase, studies that are clearly irrelevant will be excluded based on their titles and abstracts. If the information provided was insufficient to determine eligibility, the article was included in the full-text review. Any discrepancies between the two reviewers were resolved through discussion or consultation with a third researcher. The study selection process will be illustrated using a PRISMA flow diagram (figure 1).

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for the scoping review process (to be filled in final scoping review). *Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools.

Stage 4: charting the data

Data will be extracted from the studies included in the scoping review by two independent reviewers using a data extraction tool developed by the reviewers and based on the research questions. Extracted information will include author(s), year and type of publication, setting, aims, study design, population characteristics and key findings relevant to the review questions and objectives. A draft data extraction table is provided in the online supplemental file 1.

Prior to formal data extraction, the draft data extraction tool will be pilot tested on a small sample of included studies to ensure its clarity and consistency. Based on this pilot testing, the tool will be refined as necessary during the early stage of the extraction process. Any disagreements that arise between the reviewers will be resolved through discussion or with a third reviewer. Authors of papers will be contacted to request missing or additional data, where required. Modifications will be detailed in the scoping review.

Stage 5: collating, summarising and reporting the results

The extracted data will be analysed primarily using descriptive methods. Quantitative data will be summarised using frequencies and proportions to describe the distribution of studies according to study design, publication year, geographical setting, participant characteristics. In addition, characteristics of chatbot interventions, including their functions, delivery formats and therapeutic components, will be summarised. Reported outcomes related to perinatal mental health, such as depression, anxiety, user engagement and acceptability, will also be synthesised. Where appropriate, findings will be stratified by study design or intervention type to facilitate comparisons across studies.

Qualitative data will be analysed using Braun and Clarke’s six phases reflexive thematic analysis to identify the main themes,³⁵ influencing factors and barriers related to chatbot-based interventions for perinatal mental health. This process includes: (1) familiarising the data; (2) generating initial codes; (3) searching for themes; (4) reviewing themes; (5) defining and naming themes; (6) producing the final narrative synthesis. Two reviewers will independently code the data and compare their coding frameworks. Any disagreements will be resolved through discussion, and if necessary, adjudication by a third reviewer. NVivo will be used to support data coding and theme organisation throughout the analysis.

The results will be presented in tables, figures and narrative summaries to map the distribution and key characteristics of the existing evidence. The analytical methods and presentation formats may be adjusted appropriately according to the characteristics of the results.

Discussion

This scoping review will systematically synthesise global evidence on chatbot-based interventions for perinatal mental health. It aims to map existing research types, geographical distribution and participant characteristics; describe intervention forms, theoretical foundations and contexts; synthesise findings on effectiveness and user experience; and summarise clinical and technological strengths and challenges. The findings of this review may help identify key research gaps and contribute to the scientific, standardised and sustainable development of digital mental health interventions for perinatal women.

Ethics and dissemination

As this study is a literature review and does not involve individual privacy or primary data collection, ethical approval is not required. The study findings will be disseminated through conference presentations and publications in peer-reviewed journals and shared with professionals in the fields of perinatal and child health and digital health. These efforts aim to provide evidence to inform the future application of chatbots in perinatal mental health interventions.