A Dual-Prompting for Interpretable Mental Health Language Models

Hyolim Jeon; Dongje Yoo; Daeun Lee; Sejung Son; Seungbae Kim; Jinyoung Han

A Dual-Prompting for Interpretable Mental Health Language Models

Hyolim Jeon
, Dongje Yoo
, Daeun Lee
, Sejung Son
, Seungbae Kim
, Jinyoung Han

School of Convergence

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Scopus citations

Abstract

Despite the increasing demand for AI-based mental health monitoring tools, their practical utility for clinicians is limited by the lack of interpretability. The CLPsych 2024 Shared Task¹ aims to enhance the interpretability of Large Language Models (LLMs), particularly in mental health analysis, by providing evidence of suicidality through linguistic content. We propose a dual-prompting approach: (i) Knowledge-aware evidence extraction by leveraging the expert identity and a suicide dictionary with a mental health-specific LLM; and (ii) Evidence summarization by employing an LLM-based consistency evaluator. Comprehensive experiments demonstrate the effectiveness of combining domain-specific information, revealing performance improvements and the approach’s potential to aid clinicians in assessing mental state progression.

Original language	English
Title of host publication	CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop
Editors	Andrew Yates, Bart Desmet, Emily Prud�hommeaux, Ayah Zirikly, Steven Bedrick, Sean MacAvaney, Kfir Bar, Molly Ireland, Yaakov Ophir, Yaakov Ophir
Publisher	Association for Computational Linguistics (ACL)
Pages	247-255
Number of pages	9
ISBN (Electronic)	9798891760806
State	Published - 2024
Event	9th Workshop on Computational Linguistics and Clinical Psychology, CLPsych 2024 - St. Julian's, Malta Duration: 21 Mar 2024 → …

Publication series

Name	CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop

Conference

Conference	9th Workshop on Computational Linguistics and Clinical Psychology, CLPsych 2024
Country/Territory	Malta
City	St. Julian's
Period	21/03/24 → …

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Cite this

Jeon, H., Yoo, D., Lee, D., Son, S., Kim, S., & Han, J. (2024). A Dual-Prompting for Interpretable Mental Health Language Models. In A. Yates, B. Desmet, E. Prud�hommeaux, A. Zirikly, S. Bedrick, S. MacAvaney, K. Bar, M. Ireland, Y. Ophir, & Y. Ophir (Eds.), CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop (pp. 247-255). (CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop). Association for Computational Linguistics (ACL).

Jeon, Hyolim ; Yoo, Dongje ; Lee, Daeun et al. / A Dual-Prompting for Interpretable Mental Health Language Models. CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop. editor / Andrew Yates ; Bart Desmet ; Emily Prud�hommeaux ; Ayah Zirikly ; Steven Bedrick ; Sean MacAvaney ; Kfir Bar ; Molly Ireland ; Yaakov Ophir ; Yaakov Ophir. Association for Computational Linguistics (ACL), 2024. pp. 247-255 (CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop).

@inproceedings{461531e4f4fe48b2b80affb282a97038,

title = "A Dual-Prompting for Interpretable Mental Health Language Models",

abstract = "Despite the increasing demand for AI-based mental health monitoring tools, their practical utility for clinicians is limited by the lack of interpretability. The CLPsych 2024 Shared Task1 aims to enhance the interpretability of Large Language Models (LLMs), particularly in mental health analysis, by providing evidence of suicidality through linguistic content. We propose a dual-prompting approach: (i) Knowledge-aware evidence extraction by leveraging the expert identity and a suicide dictionary with a mental health-specific LLM; and (ii) Evidence summarization by employing an LLM-based consistency evaluator. Comprehensive experiments demonstrate the effectiveness of combining domain-specific information, revealing performance improvements and the approach{\textquoteright}s potential to aid clinicians in assessing mental state progression.",

author = "Hyolim Jeon and Dongje Yoo and Daeun Lee and Sejung Son and Seungbae Kim and Jinyoung Han",

note = "Publisher Copyright: {\textcopyright}2024 Association for Computational Linguistics.; 9th Workshop on Computational Linguistics and Clinical Psychology, CLPsych 2024 ; Conference date: 21-03-2024",

year = "2024",

language = "English",

series = "CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop",

publisher = "Association for Computational Linguistics (ACL)",

pages = "247--255",

editor = "Andrew Yates and Bart Desmet and Emily Prud�hommeaux and Ayah Zirikly and Steven Bedrick and Sean MacAvaney and Kfir Bar and Molly Ireland and Yaakov Ophir and Yaakov Ophir",

booktitle = "CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop",

}

Jeon, H, Yoo, D, Lee, D, Son, S, Kim, S & Han, J 2024, A Dual-Prompting for Interpretable Mental Health Language Models. in A Yates, B Desmet, E Prud�hommeaux, A Zirikly, S Bedrick, S MacAvaney, K Bar, M Ireland, Y Ophir & Y Ophir (eds), CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop. CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop, Association for Computational Linguistics (ACL), pp. 247-255, 9th Workshop on Computational Linguistics and Clinical Psychology, CLPsych 2024, St. Julian's, Malta, 21/03/24.

A Dual-Prompting for Interpretable Mental Health Language Models. / Jeon, Hyolim; Yoo, Dongje; Lee, Daeun et al.
CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop. ed. / Andrew Yates; Bart Desmet; Emily Prud�hommeaux; Ayah Zirikly; Steven Bedrick; Sean MacAvaney; Kfir Bar; Molly Ireland; Yaakov Ophir; Yaakov Ophir. Association for Computational Linguistics (ACL), 2024. p. 247-255 (CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A Dual-Prompting for Interpretable Mental Health Language Models

AU - Jeon, Hyolim

AU - Yoo, Dongje

AU - Lee, Daeun

AU - Son, Sejung

AU - Kim, Seungbae

AU - Han, Jinyoung

PY - 2024

Y1 - 2024

N2 - Despite the increasing demand for AI-based mental health monitoring tools, their practical utility for clinicians is limited by the lack of interpretability. The CLPsych 2024 Shared Task1 aims to enhance the interpretability of Large Language Models (LLMs), particularly in mental health analysis, by providing evidence of suicidality through linguistic content. We propose a dual-prompting approach: (i) Knowledge-aware evidence extraction by leveraging the expert identity and a suicide dictionary with a mental health-specific LLM; and (ii) Evidence summarization by employing an LLM-based consistency evaluator. Comprehensive experiments demonstrate the effectiveness of combining domain-specific information, revealing performance improvements and the approach’s potential to aid clinicians in assessing mental state progression.

AB - Despite the increasing demand for AI-based mental health monitoring tools, their practical utility for clinicians is limited by the lack of interpretability. The CLPsych 2024 Shared Task1 aims to enhance the interpretability of Large Language Models (LLMs), particularly in mental health analysis, by providing evidence of suicidality through linguistic content. We propose a dual-prompting approach: (i) Knowledge-aware evidence extraction by leveraging the expert identity and a suicide dictionary with a mental health-specific LLM; and (ii) Evidence summarization by employing an LLM-based consistency evaluator. Comprehensive experiments demonstrate the effectiveness of combining domain-specific information, revealing performance improvements and the approach’s potential to aid clinicians in assessing mental state progression.

UR - https://www.scopus.com/pages/publications/85189752743

M3 - Conference contribution

AN - SCOPUS:85189752743

T3 - CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop

SP - 247

EP - 255

BT - CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop

A2 - Yates, Andrew

A2 - Desmet, Bart

A2 - Prud�hommeaux, Emily

A2 - Zirikly, Ayah

A2 - Bedrick, Steven

A2 - MacAvaney, Sean

A2 - Bar, Kfir

A2 - Ireland, Molly

A2 - Ophir, Yaakov

PB - Association for Computational Linguistics (ACL)

T2 - 9th Workshop on Computational Linguistics and Clinical Psychology, CLPsych 2024

Y2 - 21 March 2024

ER -

Jeon H, Yoo D, Lee D, Son S, Kim S, Han J. A Dual-Prompting for Interpretable Mental Health Language Models. In Yates A, Desmet B, Prud�hommeaux E, Zirikly A, Bedrick S, MacAvaney S, Bar K, Ireland M, Ophir Y, Ophir Y, editors, CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop. Association for Computational Linguistics (ACL). 2024. p. 247-255. (CLPsych 2024 - 9th Workshop on Computational Linguistics and Clinical Psychology, Proceedings of the Workshop).

A Dual-Prompting for Interpretable Mental Health Language Models

Abstract

Publication series

Conference

UN SDGs

Other files and links

Fingerprint

Cite this