Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MindfulDiary: Harnessing Large Language Model to Support Psychiatric Patients' Journaling

Taewan Kim, Seolyeong Bae, Hyun Ah Kim, Su-woo Lee, Hwajung Hong, Chanmo Yang, Young-Ho Kim

TL;DR

This paper presents MindfulDiary, an LLM-driven journaling app designed for psychiatric patients, developed in collaboration with mental health professionals to ensure safety and clinical relevance. It combines a patient-facing conversational interface with a clinician dashboard, using a state-based prompting framework to guide interactions and safety checks. Through a four-week field deployment with 28 patients and five psychiatrists, MindfulDiary yielded enriched daily records, enabling deeper patient insight and greater clinician empathy while highlighting risks around inaccuracies and potential misuse. The work offers design principles, safety mechanisms, and practical evidence for integrating LLM-driven journaling into clinical mental health settings, suggesting a pathway to more nuanced patient data and improved patient-provider communication.

Abstract

In the mental health domain, Large Language Models (LLMs) offer promising new opportunities, though their inherent complexity and low controllability have raised questions about their suitability in clinical settings. We present MindfulDiary, a mobile journaling app incorporating an LLM to help psychiatric patients document daily experiences through conversation. Designed in collaboration with mental health professionals (MHPs), MindfulDiary takes a state-based approach to safely comply with the experts' guidelines while carrying on free-form conversations. Through a four-week field study involving 28 patients with major depressive disorder and five psychiatrists, we found that MindfulDiary supported patients in consistently enriching their daily records and helped psychiatrists better empathize with their patients through an understanding of their thoughts and daily contexts. Drawing on these findings, we discuss the implications of leveraging LLMs in the mental health domain, bridging the technical feasibility and their integration into clinical settings.

MindfulDiary: Harnessing Large Language Model to Support Psychiatric Patients' Journaling

TL;DR

This paper presents MindfulDiary, an LLM-driven journaling app designed for psychiatric patients, developed in collaboration with mental health professionals to ensure safety and clinical relevance. It combines a patient-facing conversational interface with a clinician dashboard, using a state-based prompting framework to guide interactions and safety checks. Through a four-week field deployment with 28 patients and five psychiatrists, MindfulDiary yielded enriched daily records, enabling deeper patient insight and greater clinician empathy while highlighting risks around inaccuracies and potential misuse. The work offers design principles, safety mechanisms, and practical evidence for integrating LLM-driven journaling into clinical mental health settings, suggesting a pathway to more nuanced patient data and improved patient-provider communication.

Abstract

In the mental health domain, Large Language Models (LLMs) offer promising new opportunities, though their inherent complexity and low controllability have raised questions about their suitability in clinical settings. We present MindfulDiary, a mobile journaling app incorporating an LLM to help psychiatric patients document daily experiences through conversation. Designed in collaboration with mental health professionals (MHPs), MindfulDiary takes a state-based approach to safely comply with the experts' guidelines while carrying on free-form conversations. Through a four-week field study involving 28 patients with major depressive disorder and five psychiatrists, we found that MindfulDiary supported patients in consistently enriching their daily records and helped psychiatrists better empathize with their patients through an understanding of their thoughts and daily contexts. Drawing on these findings, we discuss the implications of leveraging LLMs in the mental health domain, bridging the technical feasibility and their integration into clinical settings.
Paper Structure (48 sections, 5 figures, 9 tables)

This paper contains 48 sections, 5 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Journaling for Self-Reflection and Mental Health
  4. Journaling as Patient-Generated Health Data in Clinical Settings
  5. Conversational Agents for Mental Health
  6. Formative Study: Focus Group Interview
  7. Procedure and Analysis
  8. Findings from the Interviews
  9. Challenges in Eliciting Responses from Patients with Depression
  10. LLMs as a Bridge for Enhanced Patient Communication
  11. LLMs for Analytical Insights and Personalized Mental Health Support
  12. Improvements after the Interviews
  13. MindfulDiary
  14. Jane, diagnosed with chronic anxiety, frequently grapples with panic attacks. To keep track of her daily experiences, her psychiatrist recommends trying MindfulDiary as part of her treatment plan.
  15. Every evening, Jane converses with the MindfulDiary app regarding her daily activities, emotions, and thoughts. The AI leads the conversation with Jane by asking prompted and follow-up questions about her day. After a session, the app summarizes the dialogue into a journal-style essay, on which she can revisit and reflect later. She can explore the summarized essays whenever she wants to reflect on past events or thoughts.
  16. ...and 33 more sections

Figures (5)

  • Figure 1: Main screens of the MindfulDiary app. (a) The main screen, (b) the journaling screen, (c) the summary screen shown when the user submitted the journal dialogue, and (d) the review screen displaying the user's past journal.
  • Figure 2: Use flow of MindfulDiary's journaling session: (1) Pre-Journaling Assessment: Users undergo a mental health survey using the modified PHQ-9 kroenke2001phq before using MindfulDiary; (2) Users converse with MindfulDiary, documenting their day; (3) Summary Presentation: After three turns, MindfulDiary presents a diary-styled summary of the conversation so far, which can also be edited by the user. Users can continue the conversation as they want. (4) Session Closure: Once all processes are completed, MindfulDiary displays today's mental health and diary content, concluding the journaling session.
  • Figure 3: Structure of MindfulDiary's conversational pipeline. (1) Users respond to MindfulDiary's messages. (2) The recent turn count, current state, and whole user-MindfulDiary dialogue are fed into Dialogue Analyzer. (3) Using the output of Dialogue Analyzer, a designated state prompt, a summary of dialogue (containing overall dialogue context), and the latest three conversation turns are fed into Response Generator. The resulting response is then displayed to the user. Both the Dialogue Analyzer and Response Generator operate based on the GPT-4 LLM.
  • Figure 4: Procedure of the four-week field deployment study: A four-week exploration into the utilization of MindfulDiary by outpatient patients, encompassing daily use, and its integration into clinical decision-making. We note that some participants did not have a follow-up visit during the experimental period. The surveys are outside the scope of this work's investigation.
  • Figure 5: Overview of daily engagement of participants with MindfulDiary. The colored squares denote the days that participants conversed with MindfulDiary, with darker colors indicating weekend days. The bar charts on the right visualize the total number of days with interaction against the four-week study period. Participants are sorted by the number of days with interaction.