A research team has designed and implemented a group motivational interviewing system using multiple chatbots to support premenstrual syndrome (PMS), a common disorder among women. The system consists of chatbots serving as a facilitator or peers, which simulate a group counseling environment for PMS management. The study could provide valuable insights into the use of chatbots for group therapy to support women's health management and to address mental health issues.
The current findings are presented at the 2025 CHI Conference on Human Factors in Computing Systems held in Yokohama, Japan.
PMS is a disorder characterized by a variety of physical, emotional and behavioral symptoms that occur prior to menstruation, which experts estimate affects most women of reproductive age. For about 3-8% of them — which translates to tens of millions globally —it is severe enough that it is comparable to a chronic mood disorder characterized by mild depression (and only slightly less severe than major depressive disorder) that it impairs the way they function interpersonally or in the workplace.
In the current study, the research team, led by Associate Professor Koji Yatani at the University of Tokyo's Graduate School of Engineering, drew inspiration from the established benefits of group therapy to develop their group motivational interviewing system using multiple chatbots. A chatbot is a computer program that simulates human conversation so that humans can interact with digital devices similar to ways they communicate with a real person.
The challenges that women with PMS experience are often made worse by a lack of peer support because of the stigma attached to it. Even though the negative impacts of PMS can mirror that of other mental or physical disorders, research has shown women often cope with PMS alone instead of seeking peer support, which has proven to be beneficial for various disorders, including PMS.
Nowadays, women can use technology, such as smartphone applications, to track their menstrual cycles and predict their symptoms. However, these digital tools have their limitations. "Although formal clinical interventions and therapies exist to enhance awareness and coping strategies for PMS, there remains a lack of easily accessible and self-administered digital interventions designed to support women in understanding and coping with both the symptoms and the emotional and social challenges of PMS, leaving them to navigate these difficulties through self-exploration," said Shixian Geng, the study's lead author and doctoral student in Yatani's Interactive Intelligent Systems Laboratory.
Studies show the usefulness of chatbots in providing therapeutic support with mental health conditions like anxiety and depression. However, most chatbots used for mental health support are designed for one-on-one settings, where the chatbot functions as either a therapist or a friend. Existing chatbot systems typically either offer knowledge or provide companionship. However, an issue like PMS requires different types of assistance that include simultaneous support for understanding, coping and connections with peers. Up to this point, systems where multiple chatbots serve different functions remained an underexplored area of research.
Based on the ways traditional group therapy works, the team designed a system consisting of a facilitator bot and two peer bots, computer programs designed to mimic the actions of a person.
Group therapy has been widely used in treating mental health disorders such as depression, anxiety, substance use disorders and PMS. In group therapy settings, clients can reflect on their own symptoms with guidance from the facilitator, while also receiving peer support from group members. So the research team designed their multichatbot system to simulate a group therapy environment that provides simultaneous symptom tracking and social support for PMS management.
The researchers conducted their study with 63 participants and divided them into three conditions — no intervention, one-on-one chatbot or group chatbots. They evaluated the participants over two menstrual cycles. They discovered that participants in the group chat condition exhibited higher levels of engagement and language convergence with the chatbots. These participants were also able to engage in social learning and demonstrated motivation in coping through interactions with the chatbots.
Through their qualitative analysis of the interview data, the team also gained insights into participants' perceived sense of support, including a sense of belonging and social learning, as well as social comparison while interacting with peer bots.
"Our results showed that participants in the group chat demonstrated higher engagement, as well as linguistic and cognitive convergence with the chatbots when discussing PMS-related topics," said Yatani. "Additionally, we identified both potential benefits and risks of multichatbot interaction in managing PMS. These findings provide valuable insights into the integration of multiple chatbots or agents for addressing mental health issues."
In the future, the team would like to expand on its findings beyond the cultural context of Japan, where it carried out the study, and conduct longer-term studies to assess long-term effects and important aspects, such as attrition rates, to make the research more robust.
