Chen Chen, Lam Kok Tai, Yip Ka Man, So Hung Kwan, Lum Terry Yat Sang, Wong Ian Chi Kei, Yam Jason C, Chui Celine Sze Ling, Ip Patrick
Department of Paediatrics and Adolescent Medicine, School of Clinical Medicine, Li Ka Shing Faculty of Medicine, University of Hong Kong, Room 115, 1/F, New Clinical Building, Queen Mary Hospital, 102 Pokfulam Road, Hong Kong SAR, China (Hong Kong), 852 22554090, 852 28551523.
Department of Social Work Administration, University of Hong Kong, Hong Kong SAR, China (Hong Kong).
JMIR Hum Factors. 2025 Mar 6;12:e65785. doi: 10.2196/65785.
Artificial intelligence (AI) chatbots have been customized to deliver on-demand support for people with mental health problems. However, the effectiveness of AI chatbots in tackling mental health problems among the general public in Hong Kong remains unclear.
This study aimed to develop a local AI chatbot and compare the effectiveness of the AI chatbot with a conventional nurse hotline in reducing the level of anxiety and depression among individuals in Hong Kong.
This study was a pilot randomized controlled trial conducted from October 2022 to March 2023, involving 124 participants allocated randomly (1:1 ratio) into the AI chatbot and nurse hotline groups. Among these, 62 participants in the AI chatbot group and 41 in the nurse hotline group completed both the pre- and postquestionnaires, including the GAD-7 (Generalized Anxiety Disorder Scale-7), PHQ-9 (Patient Health Questionnaire-9), and satisfaction questionnaire. Comparisons were conducted using independent and paired sample t tests (2-tailed) and the χ2 test to analyze changes in anxiety and depression levels.
Compared to the mean baseline score of 5.13 (SD 4.623), the mean postdepression score in the chatbot group was 3.68 (SD 4.397), which was significantly lower (P=.008). Similarly, a reduced anxiety score was also observed after the chatbot test (pre vs post: mean 4.74, SD 4.742 vs mean 3.4, SD 3.748; P=.005), respectively. No significant differences were found in the pre-post scores for either depression (P=.38) or anxiety (P=.19). No statistically significant difference was observed in service satisfaction between the two platforms (P=.32).
The AI chatbot was comparable to the traditional nurse hotline in alleviating participants' anxiety and depression after responding to inquiries. Moreover, the AI chatbot has shown potential in alleviating short-term anxiety and depression compared to the nurse hotline. While the AI chatbot presents a promising solution for offering accessible strategies to the public, more extensive randomized controlled studies are necessary to further validate its effectiveness.
人工智能（AI）聊天机器人已被定制用于为有心理健康问题的人提供按需支持。然而，AI聊天机器人在解决香港普通公众心理健康问题方面的有效性仍不明确。
本研究旨在开发一个本地AI聊天机器人，并比较该AI聊天机器人与传统护士热线在降低香港个体焦虑和抑郁水平方面的有效性。
本研究是一项于2022年10月至2023年3月进行的试点随机对照试验，124名参与者被随机（1:1比例）分配到AI聊天机器人组和护士热线组。其中，AI聊天机器人组的62名参与者和护士热线组的41名参与者完成了前后问卷，包括广泛性焦虑障碍量表（GAD - 7）、患者健康问卷（PHQ - 9）和满意度问卷。使用独立样本t检验、配对样本t检验（双侧）和χ²检验进行比较，以分析焦虑和抑郁水平的变化。
与平均基线得分5.13（标准差4.623）相比，聊天机器人组的平均抑郁后得分是3.68（标准差4.397），显著更低（P = 0.008）。同样，在聊天机器人测试后也观察到焦虑得分降低（前测与后测：平均4.74，标准差4.742与平均3.4，标准差3.748；P = 0.005）。抑郁（P = 0.38）或焦虑（P = 0.19）的前后得分没有显著差异。两个平台在服务满意度方面没有观察到统计学上的显著差异（P = 0.32）。
在回复咨询后，AI聊天机器人在缓解参与者焦虑和抑郁方面与传统护士热线相当。此外，与护士热线相比，AI聊天机器人在缓解短期焦虑和抑郁方面显示出潜力。虽然AI聊天机器人为向公众提供可及的策略提供了一个有前景的解决方案，但需要更广泛的随机对照研究来进一步验证其有效性。
