Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 16, 2026

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Published on: May 15, 2020

Exploring GPT-based models for depression detection in adolescents using naturalistic speech: A proof-of-concept

Anna Viduani¹, Leonardo Z Ferreira², Claudia Buchweitz¹

¹Prodia - Child & Adolescent Depression Program, Hospital de Clínicas de Porto Alegre (HCPA), Porto Alegre, Brazil; Graduate Program in Psychiatry and Behavioral Sciences, Universidade Federal do Rio Grande do Sul (UFRGS), Porto Alegre, Brazil; Instituto de Pesquisa, Hospital Moinhos de Vento, Porto Alegre, RS, Brazil.

Journal of Affective Disorders

|May 14, 2026

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

The science of child and adolescent mental health in Brazil: a nationwide systematic review and compendium of evidence-based resources.

BMJ global health·2026

Same author

Community-based tuberculosis screening with computer-aided detection technology alone and in combination with point-of-care C-reactive protein testing: a paired screen-positive trial.

The Lancet. Infectious diseases·2026

Same author

Combining self-regulation and poverty reduction to prevent depression and anxiety among adolescents experiencing multi-dimensional poverty in Colombia, Nepal and South Africa: study protocol of a pilot 4-arm cluster randomised controlled trial.

Pilot and feasibility studies·2026

Same author

Contextually salient barriers and facilitators to implementing community-based mental health interventions delivered by non-specialists in Senegal: A qualitative study.

Global mental health (Cambridge, England)·2026

Same author

Adapted Early Adolescent Skills for Emotions (EASE) to decrease psychological distress among urban adolescents in New York City: a feasibility study protocol for a cluster single arm trial.

Pilot and feasibility studies·2026

Same author

Couple-based intervention for HIV prevention, care and treatment in South Africa: a study protocol for a randomised controlled trial of Simunye.

BMJ open·2026

Large Language Models (LLMs) like GPT-4o show promise in detecting depression from natural speech. This study found GPT-4o achieved 86.6% accuracy in identifying depression in adolescents from daily conversations.

Area of Science:

Artificial Intelligence in Mental Health
Computational Psychiatry
Natural Language Processing

Background:

Depression detection often relies on structured clinical assessments.
Naturalistic speech data offers a more ecologically valid window into mental state.
Large Language Models (LLMs) are increasingly explored for clinical applications.

Purpose of the Study:

To evaluate GPT-4o's ability to identify Major Depressive Disorder (MDD) from daily speech transcripts of adolescents.
To assess the impact of input volume on LLM performance for depression detection.
To compare GPT-4o's efficacy against other LLMs in this task.

Main Methods:

Thirty Brazilian adolescents (16-19 years) participated in a 14-day remote data collection via a WhatsApp chatbot.

Keywords:

Depression screening Large Language Models Natural language processing

More Related Videos

Closed-Loop Neurostimulation for Biomarker-Driven, Personalized Treatment of Major Depressive Disorder

Closed-Loop Neurostimulation for Biomarker-Driven, Personalized Treatment of Major Depressive Disorder

Published on: July 7, 2023

Related Experiment Videos

Last Updated: May 16, 2026

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Published on: May 15, 2020

Closed-Loop Neurostimulation for Biomarker-Driven, Personalized Treatment of Major Depressive Disorder

Closed-Loop Neurostimulation for Biomarker-Driven, Personalized Treatment of Major Depressive Disorder

Published on: July 7, 2023

Fifteen participants had a confirmed MDD diagnosis.

GPT-4o classified participants as depressed or non-depressed based on transcribed daily audio messages in Brazilian Portuguese, with performance analyzed across varying input lengths.

Main Results:

GPT-4o achieved 86.6% overall accuracy, with 100% sensitivity and 73.3% specificity for depression detection.
Classification performance remained stable across different input lengths, showing only modest improvements with increased data.
No statistically significant differences were found between GPT-4o and 12 other GPT-based models, though GPT-4o demonstrated numerically superior performance.

Conclusions:

LLMs, exemplified by GPT-4o, show potential for supporting depression detection using naturalistic conversational text.
This approach extends AI's utility beyond structured assessments to more ecologically valid contexts.
Findings are preliminary but highlight the promise and limitations of AI for mental health screening.