Abstract
This study presents the process of developing and adapting WAY2AGE-2 for the Spanish context. WAY2AGE -2 is a voice-based virtual assistant (VA) designed to screen for cognitive deficits in the elderly using natural language processing (NLP). Language plays a central role among the cognitive domains that can reveal early signs of cognitive decline. Spontaneous speech analysis is garnering increasing interest in neuropsychological research for the early detection of cognitive decline, owing to the high complexity of tasks that require not only lexical-semantic skills but also memory and executive functions. WAY2AGE-2 consists of various tasks assessing cognitive functions such as attention, temporal and spatial orientation, verbal episodic memory, verbal fluency, and language. The application works by the user (healthcare professional) logging in and identifying themselves in the system. Credentials are stored in a database for security reasons. Once logged in, healthcare professionals can create new sessions or check results and recordings. The healthcare professional accesses the application, which interacts with the elderly via voice. The elderly person’s words are recorded and then transcribed using an audio transcription template. A large language model (LLM), like GPT, analyzes the transcription without human interpretation. Based on the results, the elderly can be classified into one of three options: healthy, mild cognitive impairment, or dementia.
Presenters
Luciana Jornada LourençoStudent, eDoctoral School, Pontifícia Universidade Católica do Rio Grande do Sul, Catholic University of Valencia San Vicente Mártir, Rio Grande do Sul, Brazil Cristiane Boff
Dalton Breno Costa
Carmen Moret-Tatay
Professor, Universitat Politècnica de València, Spain Jose María Tormos Muñoz
Vice Chancelor of Research, Medical School, Universidad Católica de Valencia, Valencia, Spain Tatiana Quarti Irigaray
Details
Presentation Type
Theme
KEYWORDS
ARTIFICIAL INTELLIGENCE, ELDERLY, NEUROPSYCHOLOGICAL ASSESSMENT, WAY2AGE-2