Dados do Trabalho

Título

Appraxi®: The artificial intelligence-based speech training app

Resumo

Verbal apraxia is the inability to program verbal movements voluntarily. However, rehabilitation should aim at active learning. Many applications with auxiliary subjects in the health area are being increasingly explored. However, very few speeches and language rehabilitation exist, and they are in Other languages (English and Russian), whose phonology is entirely different. It should be noted that public access to speech rehabilitation and the number of professionals are restricted, and the therapeutic process is lengthy. OBJECTIVE: Develop an application to assist in the speech rehabilitation of people with verbal apraxia. METHODOLOGY: A team of speech therapists with experience in the rehabilitation of acquired neurological disorders created the stimulus network with varying degrees of complexity. This network of stimuli was computaionally absorbed through Artificial Intelligence techniques for IOS® and Android® technologies. The program was developed for minimally literate individuals with minimally functioning hearing and vision, as it has written, acoustic and visual stimuli that must be repeated and their degree of accuracy of the articulatory precision captured. The program has four interaction modules: sustained vowels, phonemes, sequences, and phrases. Each one will have three levels of complexity. In each module, there will also be subgroups of skills worked on. Among the various modules, the 'report' stands out, which informs the professional quantitatively and qualitatively about the speaker's errors. The research was approved by the CEP/IFPI/CONEP under the opinion number 5,111,083 and is currently in the implementation stage. The registration filing with the National Institute of Industrial Property is still on request. RESULTS: During the 13 months after the research was installed, the conceptual design was awarded two research funding notices (CATALISA-ICT/Sebrae® and IFES). For validation of the prototype, 127 adults native to the Portuguese language participated, distributed in the states of São Paulo, Rio de Janeiro, Maranhão, Ceará, and Piauí. In this second stage, people with apraxia are invited to participate. CONCLUSION: It is expected that daily speech training will reduce the number of sessions for monitoring the Speech-Language Pathologist and increase the turnover of patients who will be seen, allowing better access to health at the tertiary level (rehabilitation).

Palavras Chave

Speech Acoustics; Speech Therapy; Apraxias; Machine Learning; Artificial Intelligence.