A-e-i!: Herramienta para monitorear el deterioro del habla en pacientes con enfermedades neurodegenerativas

Abstract

In some neurodegenerative dementias, people may have decreased language skills that are difficult to quantify with standardized tests. Monitoring and analysis of speech in these patients can provide valuable information about the evolution of their spoken speech. Reading aloud is included in tests and treatments, but it is rarely applied systematically as an integral part of therapy and little is known (although it is highly prescribed) regarding its efficacy. This paper presents the development of a web application, called "A-E-I!", which consumes functions of services hosted on cloud servers to transcribe audio readings from patients that suffer primary progressive aphasia. The objective is to monitor the changes in the speech over time of these patients. To carry out this monitoring, metrics were designed to measure different aspects in relation to reading fluency, to calculate them it is necessary to analyze every transcription obtained before. This is useful for a later evaluation based on indicators that will interpret the values ​​measured over time. From a technological point of view, speech to text and recognition of parts of a speech services were combined.