Using Chat GPT to Analyze Language in Depressed Adolescents
Author Information
Author(s): Jarvers Irina, Ecker Angelika, Donabauer Pia, Kampa Katharina, Weißenbacher Maximilian, Schleicher Daniel, Kandsperger Stephanie, Brunner Romuald, Ludwig Bernd
Primary Institution: University of Regensburg
Hypothesis
Can linguistic patterns in adolescents' verbal responses help identify depressive disorders?
Conclusion
The study found that linguistic patterns can effectively indicate depression in adolescents, and Chat GPT can generate useful synthetic data for training classifiers.
Supporting Evidence
- The classifier based on BERT achieved accuracies up to 97%.
- Chat GPT was found to generate realistic utterances for training purposes.
- Participants included 40 adolescents with depressive disorders and 13 healthy controls.
Takeaway
The researchers looked at how teenagers talk about their feelings to see if it can help find out if they are depressed, and they used a computer program to help create more examples.
Methodology
The study involved analyzing interviews with adolescents, using machine learning models to classify responses as depressive or non-depressive.
Potential Biases
Potential biases in the sample and the reliance on synthetic data generation methods.
Limitations
The study's sample was predominantly female, which may limit generalizability.
Participant Demographics
75% of participants had depressive disorders; 79.2% were female, with a mean age of 15.5 years.
Statistical Information
P-Value
p<0.001
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website