Procesamiento del Lenguaje Natural (PLN) y Salud Mental: Afinando un Modelo BERT para Evaluar Ansiedad
Year: 2023
Type: dissertation
Abstract: Language is an important access tool into the human mind, an objective which has been \ntheoretically grouped under psycholinguistics. The development of computational methods around \nartificial intelligence, specifically Natural Language Processing, has allowed the development of \nobjective methods to model speech and writing. The applications towards the psychological field \ncould be important to both theory and clinical practice. The most advanced NLP systems are built in \nthe Transformers architecture (BERT or GPT). In this study, a BERT Spanish model is fine-tuned to \ndifferentiate between interview transcriptions according to anxiety levels, in a supervised \nclassification task, using the STAI measure as reference. In a corpus generated by 11 women \nbetween 20 and 24 years-old, a 60% accuracy was reached when distinguishing two groups. \nAlthough a low accuracy, it is shown the high impact potential of these tools for automated \ndiagnosis in mental health.
Author: Vidal-Marciel, Javier
Language: English
Topic: Mental Health via Writing
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7061568528