Abstract:Despite the global mental health crisis, access to screenings, professionals, and treatments remains high. In collaboration with licensed psychotherapists, we propose a Conversational AI Therapist with psychotherapeutic Interventions (CaiTI), a platform that leverages large language models (LLM)s and smart devices to enable better mental health self-care. CaiTI can screen the day-to-day functioning using natural and psychotherapeutic conversations. CaiTI leverages reinforcement learning to provide personalized conversation flow. CaiTI can accurately understand and interpret user responses. When the user needs further attention during the conversation, CaiTI can provide conversational psychotherapeutic interventions, including cognitive behavioral therapy (CBT) and motivational interviewing (MI). Leveraging the datasets prepared by the licensed psychotherapists, we experiment and microbenchmark various LLMs' performance in tasks along CaiTI's conversation flow and discuss their strengths and weaknesses. With the psychotherapists, we implement CaiTI and conduct 14-day and 24-week studies. The study results, validated by therapists, demonstrate that CaiTI can converse with users naturally, accurately understand and interpret user responses, and provide psychotherapeutic interventions appropriately and effectively. We showcase the potential of CaiTI LLMs to assist the mental therapy diagnosis and treatment and improve day-to-day functioning screening and precautionary psychotherapeutic intervention systems.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address the issues of difficulty in accessing psychological screening, professional treatment, and therapeutic interventions amidst the global mental health crisis. Specifically, the authors propose a conversational AI therapist based on large language models (LLM), named CaiTI, which facilitates daily function screening and psychological therapeutic interventions through everyday smart devices. ### Background and Motivation 1. **Global Mental Health Crisis**: The COVID-19 pandemic has exacerbated global mental health issues, yet the barriers to accessing mental health screening and treatment remain high. 2. **Limitations of Existing Technologies**: - Most existing research focuses primarily on understanding emotional or affective states, neglecting the importance of patients' daily activities and behaviors. - Current psychological therapy systems lack personalization and user understanding capabilities. - AI chatbots like ChatGPT have limitations in psychological therapy and are rarely used in clinical settings. - Tools like smartphones may not be suitable for elderly individuals with vision or memory impairments. ### Solution CaiTI is a conversational AI therapist based on large language models that addresses the aforementioned issues through the following methods: 1. **Comprehensive Daily Function Screening**: CaiTI evaluates users' daily functions from 37 dimensions through natural conversations and open-ended questions. 2. **Personalized Conversation Flow**: Using reinforcement learning techniques, CaiTI can personalize and adjust the conversation flow based on users' responses. 3. **Effective Psychological Therapeutic Interventions**: CaiTI combines psychological therapy methods such as Cognitive Behavioral Therapy (CBT) and Motivational Interviewing (MI) to provide personalized psychological therapeutic interventions. 4. **Multimodal Data Processing**: CaiTI can understand and analyze users' voice and text inputs, ensuring privacy while providing high-quality services. 5. **Usability and Accessibility**: CaiTI offers services through common devices like smartphones, computers, and smart speakers, making it convenient for users to use both indoors and outdoors. ### Main Contributions 1. **CaiTI System**: A conversational AI therapist based on LLM that can comprehensively screen users' daily functions and provide appropriate empathetic validation and psychological therapy based on users' mental and physical health conditions. 2. **Personalized Conversation Experience**: Utilizing reinforcement learning techniques, CaiTI can dynamically adjust the conversation flow based on users' responses, prioritizing the dimensions that users need to focus on. 3. **Integration of Psychological Therapy Techniques**: CaiTI's conversational architecture integrates Motivational Interviewing (MI) and Cognitive Behavioral Therapy (CBT), providing natural and practical psychological therapeutic dialogue interventions. 4. **Quality Control and Effectiveness Validation**: CaiTI integrates multiple task-specific LLM foundational models in the psychological therapy process to ensure therapy quality and prevent the propagation of biases in AI algorithms. 5. **Empirical Research**: In collaboration with licensed psychotherapists, a proof-of-concept prototype of CaiTI was designed, implemented, and deployed, followed by a 24-week practical application test, demonstrating CaiTI's effectiveness and user acceptance. ### Conclusion CaiTI is the first conversational AI therapist system that utilizes smart home devices and LLM technology to simulate clinical psychological therapy processes, providing continuous monitoring and interaction. By integrating psychological therapy techniques (MI and CBT), it offers comprehensive mental health support to users.

LLM-based Conversational AI Therapist for Daily Functioning Screening and Psychotherapeutic Intervention via Everyday Smart Devices

Can AI Relate: Testing Large Language Model Response for Mental Health Support

Advancing Conversational Psychotherapy: Integrating Privacy, Dual-Memory, and Domain Expertise with Large Language Models

Interactive Agents: Simulating Counselor-Client Psychological Counseling via Role-Playing LLM-to-LLM Interactions

Supporting the Demand on Mental Health Services with AI-Based Conversational Large Language Models (LLMs)

CBT-Bench: Evaluating Large Language Models on Assisting Cognitive Behavior Therapy

AI-Enhanced Cognitive Behavioral Therapy: Deep Learning and Large Language Models for Extracting Cognitive Pathways from Social Media Texts

Can an LLM-Powered Socially Assistive Robot Effectively and Safely Deliver Cognitive Behavioral Therapy? A Study With University Students

Are Large Language Models Possible to Conduct Cognitive Behavioral Therapy?

Using Conversational AI to Facilitate Mental Health Assessments and Improve Clinical Efficiency Within Psychotherapy Services: Real-World Observational Study

Evaluating the Efficacy of Interactive Language Therapy Based on LLM for High-Functioning Autistic Adolescent Psychological Counseling

Psy-LLM: Scaling up Global Mental Health Psychological Services with AI-based Large Language Models

ChatCounselor: A Large Language Models for Mental Health Support

Therapy as an NLP Task: Psychologists' Comparison of LLMs and Human Peers in CBT

LLM-Mini-CEX: Automatic Evaluation of Large Language Model for Diagnostic Conversation

From Classification to Clinical Insights: Towards Analyzing and Reasoning About Mobile and Behavioral Health Data With Large Language Models

Designing Human-centered AI for Mental Health: Developing Clinically Relevant Applications for Online CBT Treatment

HealMe: Harnessing Cognitive Reframing in Large Language Models for Psychotherapy

The Role of Humanization and Robustness of Large Language Models in Conversational Artificial Intelligence for Individuals With Depression: A Critical Analysis

Enhancing AI-Driven Psychological Consultation: Layered Prompts with Large Language Models

A Novel Nuanced Conversation Evaluation Framework for Large Language Models in Mental Health