Feasibility of human-in-the-loop multimodal generative artificial intelligence chatbot for school-based adolescent mental health support

doi:10.5498/wjp.123904

Advanced Search

BPG is committed to discovery and dissemination of knowledge

Home / Archive / Volume 16, Issue 10

This Article

(1)

(0)

(18)

Peer-Review Report of This Article

CrossCheck and Google Search of This Article

Academic Rules and Norms of This Article

Supplementary Materials of This Article

Citation of this article

Corresponding Author of This Article

Research Domain of This Article

Article-Type of This Article

Open-Access Policy of This Article

Times Cited Counts in Google of This Article

Journal Information of This Article

Publication Name

World Journal of Psychiatry

ISSN

2220-3206

Publisher of This Article

Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA

Prospective Study

Copyright: ©Author(s) 2026. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.

World J Psychiatry. Oct 19, 2026; 16(10): 123904
Published online Oct 19, 2026. doi: 10.5498/wjp.123904

Feasibility of human-in-the-loop multimodal generative artificial intelligence chatbot for school-based adolescent mental health support

Xi-Wang Fan, Ting-Yu Lin, Ming-Hao Wang, Li-Cheng Wu, Xin-Xi Chen, Jun-Yao Chen, Yi-Zhen Wu, Yu Zheng, Tang Li, Hao Du, Yao-Xuan Wang, Yu-Jie Wang, Jing-Wen Zhang, Li-Wei Huang, Tian-Ze Fan, Bing Liu, Jie Liu, Pei Sun, Hui Zhao, Guo-Lin Ma, Qiang Cheng, Shi-Jun Li

Xi-Wang Fan, Ting-Yu Lin, Li-Cheng Wu, Xin-Xi Chen, Jun-Yao Chen, Yi-Zhen Wu, Yu Zheng, Hui Zhao, Clinical Research Center for Mental Disorders, Shanghai Pudong New Area Mental Health Center, School of Medicine, Tongji University, Shanghai 200124, China

Ming-Hao Wang, Tang Li, Department of SituTech AI Lab, Beijing Situ Wellbeing Technology Co., LTD, Beijing 100080, China

Hao Du, Yao-Xuan Wang, Yu-Jie Wang, Jing-Wen Zhang, Department of General Manager, Beijing Ruihong Embodied Intelligence Robot Technology Co., Ltd., Beijing 100176, China

Yao-Xuan Wang, Yu-Jie Wang, Jing-Wen Zhang, Department of Artificial Intelligence, Beijing Ruihong Embodied Intelligence Robot Technology Co., Ltd., Beijing 100176, China

Li-Wei Huang, Department of Management, Beijing Gaidi Intelligent Technology Co., Ltd., Beijing 100070, China

Tian-Ze Fan, Department of Radiology, The 72^nd Group Army Hospital of the Chinese People's Liberation Army, Zhejiang 313100, China

Bing Liu, Department of Radiology, Chinese PLA Medical School, Beijing 100853, China

Bing Liu, Shi-Jun Li, Department of Radiology, The First Medical Center of Chinese PLA General Hospital, Beijing 100853, China

Jie Liu, Beijing Key Laboratory of Design and Intelligent Machining Technology for High Precision Machine Tools, Beijing University of Technology, Beijing 100124, China

Pei Sun, Faculty of Health and Wellness, City University of Macau, Macao 999078, China

Pei Sun, Tsinghua Laboratory of Brain and Intelligence and Department of Psychological and Cognitive Science, Tsinghua University, Beijing 100084, China

Guo-Lin Ma, Department of Radiology, China-Japan Friend Hospital, Beijing 100029, China

Qiang Cheng, College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing 100124, China

Co-first authors: Xi-Wang Fan and Ting-Yu Lin.

Author contributions: Fan XW and Lin TY contribute equally to this study as co-first authors; Fan XW, Lin TY, Wang MH, Zheng Y, Chen JY, and Li SJ conceived and designed the study; Fan XW, Lin TY, Wang MH, Wu LC, Li T, Du H, Wang YX, Wang YJ, Zhang JW, Huang LW, Fan TZ, Liu B, Zhao H, and Li SJ contributed to study implementation, including participant recruitment, on-site coordination, technical support, and data acquisition; Chen XX and Lin TY conducted the statistical analyses; Lin TY, Chen XX, and Chen JY verified the analytic outputs; Fan XW, Lin TY, Wang MH, Chen XX, Wu YZ, Wu LC, Chen JY, Zheng Y, and Li SJ drafted the manuscript; all authors reviewed and revised the manuscript critically for important intellectual content and approved the final version for publication; authors affiliated with technology companies contributed to platform development, technical operations, technical troubleshooting, and/or implementation support according to their roles and they did not have sole authority over outcome definition, inferential statistical analysis, interpretation of clinical outcome findings, or manuscript conclusions; Fan XW and Li SJ accept responsibility for the work as a whole, including the integrity of the data, the accuracy of the analyses, and the transparency of role separation.

AI contribution statement: During revision, the authors used ChatGPT (OpenAI, GPT-5.6 SOL; web application; accessed in July 2026) solely for English-language editing of text drafted by the authors in the manuscript and the point-by-point response. The tool was used to improve grammar, sentence structure, word choice, concision, and readability. It was not used to generate scientific content or substantive responses to the reviewers, search for or synthesize literature, generate or verify references, generate or modify study data, conduct statistical analyses, interpret the results, or draw conclusions. The authors independently developed all scientific content and substantive responses, reviewed and approved every AI-assisted language edit, and take full responsibility for the accuracy, integrity, and originality of the manuscript and the point-by-point response.

Supported by the Beijing Nova Programme Interdisciplinary Cooperation Project, No. 20240484674; Capital’s Funds for Health Improvement and Research, No. CFH2024-2-5024; and the Tongji University Medicine-X Interdisciplinary Research Initiative, No. 2025-0708-YB-02.

Institutional review board statement: This study was approved by the Ethics Committee of Chinese PLA General Hospital (Approval No. S2024-849-01).

Clinical trial registration statement: This study was registered at the Chinese Clinical Trial Registry (https://www.chictr.org.cn/). The registration identification number is ChiCTR2500100806.

Informed consent statement: All participants and their guardians provided informed consent before participation.

Conflict-of-interest statement: The authors declare potential competing interests related to professional affiliations and platform development. Ming-Hao Wang and Tang Li are employees of Beijing Situ Wellbeing Technology Co., Ltd., which contributed to the development of the Duoduo system. Hao Du, Yao-Xuan Wang, Yu-Jie Wang, and Jing-Wen Zhang are employees of Beijing Ruihong Embodied Intelligence Robot Technology Co., Ltd., and Li-Wei Huang is an employee of Beijing Gaidi Intelligent Technology Co., Ltd.; these authors contributed to technical operations, platform implementation, and/or study support activities. To reduce the risk of interpretive or commercial bias, role separation was implemented throughout the study. Company-affiliated authors contributed to system development, technical deployment, troubleshooting, and platform-related implementation support, but they did not have sole authority over participant eligibility criteria, outcome selection, inferential statistical analysis, interpretation of clinical outcome findings, or manuscript conclusions. Clinical outcome analyses were conducted by the named analysis team, and analytic outputs were checked by the named verification team. The guarantor authors take responsibility for the integrity of the data, the accuracy of the analyses, and the decision to submit the manuscript. Access to identifiable participant information, raw interaction logs, multimodal records, and monitoring dashboards was restricted according to role-based permissions required for technical operation, safety monitoring, and school follow-up. Company-affiliated authors did not independently use raw interaction logs or multimodal records to define clinical outcomes, adjudicate treatment effects, or determine the interpretation of the reported findings. The Duoduo system was used in this study as a supervised school-based research and implementation tool. No participant fees were charged for use of the system during the study, and the reported clinical outcome analyses were not linked to sales, marketing, or participant-level commercial decisions.

CONSORT 2010 statement: The authors have read the CONSORT 2010 Statement, and the manuscript was prepared and revised according to the CONSORT 2010 Statement.

Data sharing statement: Raw interaction logs and multimodal records will not be shared publicly because they may contain identifiable or sensitive information from minors. Any secondary use of de-identified analytic data or supporting materials will require appropriate ethics approval, privacy protection, and institutional data-governance review, and will not include access to identifiable raw audiovisual or interaction records.

Corresponding author: Shi-Jun Li, Department of Radiology, The First Medical Center of Chinese PLA General Hospital, No. 28 Fuxing Road, Haidian District, Beijing 100853, China. shijunli07@yeah.net

Received: June 2, 2026
Revised: July 7, 2026
Accepted: September 8, 2026
Published online: October 19, 2026
Processing time: 132 Days and 0.2 Hours

Abstract

BACKGROUND

Schools often identify signs of emotional distress in adolescents before they access formal mental health services; however, many school counseling units lack the capacity to provide individualized support. Generative artificial intelligence (GenAI) chatbots may offer an accessible, low-threshold means of providing support. Nevertheless, their use among minors requires staff supervision, strong privacy protections, and clearly defined protocols for identifying and escalating risk-related content.

AIM

To evaluate the feasibility, user engagement, working alliance, safety monitoring workflow, and exploratory short-term outcome indicators of Duoduo, an acceptance and commitment therapy-informed multimodal GenAI chatbot developed for adolescents with elevated depressive symptoms.

METHODS

We conducted a prospective, non-randomized, controlled pilot study at a junior high school in Zhejiang Province, China. Students with elevated depressive symptoms were assigned to either an intervention group (n = 30) or a waitlist control group (n = 32). The intervention comprised eight supervised sessions delivered over two weeks. Feasibility outcomes included completion, engagement, working alliance, and predefined safety-monitoring indicators. Exploratory clinical outcomes were evaluated using baseline-adjusted analysis of covariance, with false discovery rate (FDR) correction applied for multiple comparisons.

RESULTS

Among the 30 students who initiated the intervention, 26 (86.7%) completed all scheduled sessions. Participants used the system for a mean total of 101.3 minutes. The mean Working Alliance Inventory-Short Revised score was 3.56 (SD = 0.86). Nine students triggered dangerous-behavior alerts, and school counselors conducted offline follow-up according to the established safety protocol. In the unadjusted analyses, nominal between-group differences were observed for anxiety and optimism. However, these differences were no longer evident after adjustment for baseline symptom severity and did not remain significant after FDR correction for multiple comparisons. No clinical outcome demonstrated a statistically significant between-group difference after adjustment.

CONCLUSION

Supervised school-based implementation of the chatbot was feasible and supports further evaluation in randomized controlled trials to determine its clinical effectiveness and safety.

Keywords: Generative artificial intelligence; Adolescent mental health; Chatbot; Feasibility study; Depression; Nonrandomized pilot study; Quasi-experimental study; Acceptance and commitment therapy; Safety monitoring; Human-in-the-loop

Core Tip: This pilot study evaluated the supervised school-based implementation of Duoduo, an acceptance and commitment therapy-informed multimodal generative artificial intelligence chatbot developed for adolescents with elevated depressive symptoms. Most participants completed the 2-week intervention, engaged with the system for an average total duration of approximately 101 minutes, and reported a moderate working alliance. Risk-related conversation signals were referred for staff review through a predefined human-in-the-loop safety workflow. The findings primarily support the feasibility of implementation and provide guidance for future program planning. Clinical outcome findings were considered exploratory because participant allocation was pragmatic, baseline symptom severity differed between groups, multiple outcomes were evaluated, and the follow-up period was brief.