Leveraging large language models for automated depression screening
Einsatz großer Sprachmodelle für ein automatisiertes Depressionsscreening
PLOS Digital Health · 28.07.2025 · 4 (7) · e0000943
DOI: 10.1371/journal.pdig.0000943 ↗Auf einen Blick
In Daten aus klinischen Interviews und Fragebögen sagte GPT-4o die PHQ-8-Items besser voraus als andere getestete LLMs. Die beste mittlere Genauigkeit lag bei 75,9 %.
Fragestellung
Wie gut können Zero-Shot-LLMs Depressionssymptome und PHQ-8-Items aus Textdaten vorhersagen?
Methode
Vergleich verschiedener LLMs auf dem öffentlich verfügbaren Datensatz DAIC-WOZ mit klinischen Interviewtexten und Selbstbericht-Fragebögen; Bewertung anhand der PHQ-8-Items mit dem RISEN-Prompt-Framework.
Ergebnisse
GPT-Modelle, besonders GPT-4o, waren über alle 8 PHQ-8-Items hinweg besser als Llama3_8B, Cohere und Gemini. Berichtet wurden eine Genauigkeit von M = 75,9 % und ein F1-Score von 0,74.
Einschränkungen
Keine weiteren Datensätze; kein Fine-Tuning; weitere Forschung wurde von den Autor:innen gefordert.
Schlussfolgerung der Autor:innen
Die Ergebnisse deuten laut den Autor:innen darauf hin, dass LLMs Selbstberichtswerte aus Interviewtexten vorhersagen und das Depressionsscreening unterstützen könnten.
Kernaussagen
- GPT-4o schnitt über die 8 PHQ-8-Items am besten ab
- Berichtet: Genauigkeit M = 75,9 % und F1-Score 0,74
- LLMs könnten das Depressionsscreening unterstützen
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Fachartikel in der Zeitschrift PLOS Digital Health, online erschienen am 28.07.2025. Verfasst von Bazen Gashaw Teferra und 14 weiteren Autor:innen. Beteiligte Einrichtung: Interventional Psychiatry Program, St. Michael's Hospital, Unity Health Toronto, Toronto, Ontario, Canada. Die Arbeit wurde bisher 25-mal zitiert (Crossref) und nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Themen: Depression
Abstract (Originalsprache)
Mental health diagnoses possess unique challenges that often lead to nuanced difficulties in managing an individual’s well-being and daily functioning. Self-report questionnaires are a common practice in clinical settings to help mitigate the challenges involved in mental health disorder screening. However, these questionnaires rely on an individual’s subjective response which can be influenced by various factors. Despite the advancements of Large Language Models (LLMs), quantifying self-reported experiences with natural language processing has resulted in imperfect accuracy. This project aims to demonstrate the effectiveness of zero-shot learning LLMs for screening and assessing item scales for depression using LLMs. The DAIC-WOZ is a publicly available mental health dataset that contains textual data from clinical interviews and self-report questionnaires with relevant mental health disorder labels. The RISEN prompt engineering framework was utilized to evaluate LLMs’ effectiveness in predicting depression symptoms based on individual PHQ-8 items. Various LLMs, including GPT models, Llama3_8B, Cohere, and Gemini were assessed based on performance. The GPT models, especially GPT-4o, were consistently better than other LLMs (Llama3_8B, Cohere, Gemini) across all eight items of the PHQ-8 scale in accuracy (M = 75.9%), and F1 score (0.74). GPT models were able to predict PHQ-8 items related to emotional and cognitive states. Llama 3_8B demonstrated superior detection of anhedonia-related symptoms and the Cohere LLM’s strength was identifying and predicting psychomotor activity symptoms. This study provides a novel outlook on the potential of LLMs for predicting self-reported questionnaire scores from textual interview data. The promising preliminary performance of the various models indicates there is potential that these models could effectively assist in the screening of depression. Further research is needed to establish a framework for which LLM can be used for specific mental health symptoms and other disorders. As well, analysis of additional datasets while fine-tuning models should be explored.
Autor:innen und Institutionen
- Interventional Psychiatry Program, St. Michael's Hospital, Unity Health Toronto, Toronto, Ontario, Canada.
- Electrical and Computer Engineering, University of Toronto, Toronto, Ontario, Canada.
- Department of Mathematical and Computational Sciences, University of Toronto, Toronto, Ontario, Canada.
- Electrical and Computer Engineering, University of Alberta, Edmonton, Alberta, Canada.
- Department of Computer Science, University of Waterloo, Waterloo, Ontario, Canada.
- Electrical, Computer, and Biomedical Engineering, Toronto Metropolitan University, Toronto, Ontario, Canada.
ORCID-Profile: Bazen Gashaw Teferra, Alice Rueda, Yuqi Wu, Achint Soni, Rakesh Jetly, Sri Krishnan, Venkat Bhat
Zitieren
APA
Teferra, B. G., Perivolaris, A., Hsiang, W. N., Sidharta, C. K., Rueda, A., Parkington, K., Wu, Y., Soni, A., Samavi, R., Jetly, R., Zhang, Y., Cao, B., Rambhatla, S., Krishnan, S., & Bhat, V. (2025). Leveraging large language models for automated depression screening. PLOS Digital Health, 4(7), e0000943. https://doi.org/10.1371/journal.pdig.0000943
Vancouver
Teferra BG, Perivolaris A, Hsiang WN, Sidharta CK, Rueda A, Parkington K, et al. Leveraging large language models for automated depression screening. PLOS Digital Health. 2025;4(7):e0000943. doi:10.1371/journal.pdig.0000943
BibTeX
@article{Teferra2025,
author = {Teferra, Bazen Gashaw and Perivolaris, Argyrios and Hsiang, Wei-Ni and Sidharta, Christian Kevin and Rueda, Alice and Parkington, Karisa and Wu, Yuqi and Soni, Achint and Samavi, Reza and Jetly, Rakesh and Zhang, Yanbo and Cao, Bo and Rambhatla, Sirisha and Krishnan, Sri and Bhat, Venkat},
title = {Leveraging large language models for automated depression screening},
journal = {PLOS Digital Health},
year = {2025},
volume = {4},
number = {7},
pages = {e0000943},
doi = {10.1371/journal.pdig.0000943}
}
EUVIMED