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

**Authors:** Taewan Kim, Seolyeong Bae, Hyun Ah Kim, Su-woo Lee, Hwajung Hong, Chanmo Yang, Young-Ho Kim  
**arXiv:** [2310.05231](https://arxiv.org/abs/2310.05231)  
**Date:** 2023-10-08  
**Summary page:** https://www.summarizepaper.com/en/paper/2310.05231v1/

## Key points

- Large Language Models (LLMs) offer new opportunities for supporting patients in the field of mental health
- MindfulDiary is a mobile journaling app that incorporates an LLM to help psychiatric patients document their daily experiences through conversation
- The app was designed in collaboration with mental health professionals (MHPs) and takes a state-based approach to ensure compliance with guidelines while allowing for free-form conversations
- A four-week field study involving 28 patients with major depressive disorder and five psychiatrists evaluated the effectiveness of MindfulDiary
- Results showed that the app supported patients in consistently enriching their daily records and helped psychiatrists better empathize with their patients by gaining insights into their thoughts and daily contexts
- The use of LLMs in MindfulDiary allowed for dynamic responses on diverse topics, enhancing the versatility and narrative-building capability of the app
- The findings highlight the potential benefits of leveraging LLMs in the mental health domain, improving patient record-keeping and treatment quality
- Caution must be exercised when using LLMs in mental health contexts due to challenges related to control and generating unintended or inaccurate responses
- MindfulDiary demonstrates how LLMs can support psychiatric patients' journaling efforts through naturalistic conversations, providing valuable insights for MHPs and overcoming challenges related to detailed record-keeping and expression
- Responsible integration of LLMs into clinical mental health settings has the potential to enhance patient care while ensuring safety and privacy.

## Summary

In the field of mental health, Large Language Models (LLMs) offer new opportunities for supporting patients. However, their complexity and lack of control have raised concerns about their suitability in clinical settings. To address this, researchers developed MindfulDiary - a mobile journaling app that incorporates an LLM to help psychiatric patients document their daily experiences through conversation. The app was designed in collaboration with mental health professionals (MHPs) and takes a state-based approach to ensure compliance with guidelines while allowing for free-form conversations. To evaluate the effectiveness of MindfulDiary, a four-week field study was conducted involving 28 patients with major depressive disorder and five psychiatrists. The results showed that the app supported patients in consistently enriching their daily records and helped psychiatrists better empathize with their patients by gaining insights into their thoughts and daily contexts. The use of LLMs in MindfulDiary allowed for dynamic responses on diverse topics, enhancing the versatility and narrative-building capability of the app. The findings from the study highlight the potential benefits of leveraging LLMs in the mental health domain. By bridging technical feasibility with integration into clinical settings, LLM-driven conversational tools like MindfulDiary can aid in patient record-keeping and improve treatment quality. However, caution must be exercised when using LLMs in mental health contexts due to challenges related to control and generating unintended or inaccurate responses. Overall, MindfulDiary demonstrates how LLMs can be harnessed to support psychiatric patients' journaling efforts through naturalistic conversations. The app provides valuable insights for MHPs and helps overcome challenges related to detailed record-keeping and expression. Responsible integration of LLMs into clinical mental health settings has the potential to enhance patient care while ensuring patient safety and privacy.