EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Stress detection through prompt engineering with a general-purpose LLM

Nima Esmi, Asadollah Shahbahrami, Yasaman Nabati, Bita Rezaei, Georgi Gaydadjiev, Peter de Jonge

Acta Psychologica · 2025-10 · 260 · 105462

DOI: 10.1016/j.actpsy.2025.105462 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Acta Psychologica, 2025-10. Verfasst von Nima Esmi und 5 weiteren Autor:innen. Beteiligte Einrichtung: Bernoulli Institute, University of Groningen, Groningen, The Netherlands; ISRC, Khazar University, Baku, Azerbaijan. Electronic address: n.esmi.rudba…. Hauptthemen laut MEDLINE: Stress, Psychological, Social Media, Language. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Advancements in large language models (LLMs) have opened new avenues for mental health monitoring through social media analysis. In this study, we present an iterative prompt engineering framework that significantly enhances the performance of the general-purpose LLM, GPT-4, for stress detection in social media posts, leveraging psychologist-informed hints. This approach achieved a substantial 17% accuracy improvement from 72% to 89% for the January 2025 version of GPT-4, alongside an 80% reduction in false positives compared to baseline zero-shot prompting. Our method not only surpassed domain-specific models like Mental-RoBERTa by 5% but also uniquely generates human-readable rationales. These rationales are crucial for mental health professionals, assisting them in understanding and validating the model's outputs-a key benefit for sensitive mental health applications. These results highlight prompt engineering as a resource-efficient, transparent strategy to adapt general-purpose LLMs for specialized tasks, offering a scalable solution for mental health monitoring without the need for costly fine-tuning.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Stress, PsychologicalSocial MediaLanguageHumans

Stichwörter der Autor:innen

GPT-4Large language modelPrompt engineeringSocial mediaStress recognition

Autor:innen und Institutionen

  • Bernoulli Institute, University of Groningen, Groningen, The Netherlands; ISRC, Khazar University, Baku, Azerbaijan. Electronic address: n.esmi.rudbardeh@rug.nl.
  • ISRC, Khazar University, Baku, Azerbaijan; University of Guilan, Guilan, Iran. Electronic address: shahbahrami@guilan.ac.ir.
  • University of Guilan, Guilan, Iran. Electronic address: yas.nabauti@gmail.com.
  • University of Guilan, Guilan, Iran. Electronic address: bita.rezaeidogaheh@gmail.com.
  • Delft University of Technology, Delft, The Netherlands. Electronic address: g.n.gaydadjiev@tudelft.nl.
  • University of Groningen, Groningen, The Netherlands. Electronic address: peter.de.jonge@rug.nl.

Zitieren

APA

Esmi, N., Shahbahrami, A., Nabati, Y., Rezaei, B., Gaydadjiev, G., & de Jonge, P. (2025). Stress detection through prompt engineering with a general-purpose LLM. Acta Psychologica, 260, 105462. https://doi.org/10.1016/j.actpsy.2025.105462

Vancouver

Esmi N, Shahbahrami A, Nabati Y, Rezaei B, Gaydadjiev G, de Jonge P. Stress detection through prompt engineering with a general-purpose LLM. Acta Psychologica. 2025;260:105462. doi:10.1016/j.actpsy.2025.105462

BibTeX
@article{Esmi2025,
  author = {Esmi, Nima and Shahbahrami, Asadollah and Nabati, Yasaman and Rezaei, Bita and Gaydadjiev, Georgi and de Jonge, Peter},
  title = {Stress detection through prompt engineering with a general-purpose LLM},
  journal = {Acta Psychologica},
  year = {2025},
  volume = {260},
  pages = {105462},
  doi = {10.1016/j.actpsy.2025.105462}
}
Kostenlos starten →