INVESTIGADORES
GARCIA adolfo Martin
congresos y reuniones científicas
Título:
Dependency analysis of spoken language for assessment of neurological disorders
Autor/es:
EYIGOZ, ELIF; PIETROWICZ, MARY; AGURTO, CARLA; OROZCO-ARROYAVE, JUAN RAFAEL; GARCÍA, ADOLFO M.; SKODDA, SABINE; RUSZ, JAN; NÖTH, ELMAR; CECCHI, GUILLERMO
Reunión:
Conferencia; Language Resources and Evaluation Conference 2020; 2020
Resumen:
Spoken language carries a greater amount of information about the speakers' cognitive state compared to written language, when analysis of disfluencies and expressivity is considered. For the same reason, however, spoken language presents challenges for automated syntactic analysis. This study presents comparative performance of different preprocessing methods applied to spoken language data for syntactic analysis. Furthermore, this study presents a novel dependency tree analysis for assessment of neurological disorders. The methods described in this paper were compared against a baseline model consisting of features typically considered to provide for the largest amount of information within the context of neurological disorders. Significant improvements were obtained across multiple languages and multiple neurological disorders: The method improved detection of Alzheimer's disease patients in English-speaking subjects and detection of Parkinson's disease patients in German-speaking subjects.