Avaliação fonológica personalizada com estrutura de dados otimizada

• Ano de defesa: 2024
• Autor(a) principal: Marques, João Víctor Bolsson
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Ciência da Computação; UFSM; Programa de Pós-Graduação em Ciência da Computação; Centro de Tecnologia
• Palavras-chave em Português: Aplicativo móvel; Árvore de decisão; Teste de triagem fonológica; Avaliação dinâmica personalizada; Distúrbios dos sons da fala; Phonological screening test; Data structure; Customized dynamic assessment; Speech sound disorders; Mobile App; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/33184

The development and evaluation of technologies for phonological screening have attracted increasing interest, particularly in the fields of information technology and speech therapy. This study presents an innovative approach using Decision Tree (DT) for screening and dynamic assessment in speech therapy. The phonological evaluation process mainly occurs through the assessment of target phonemes present in words familiar to the children’s vocabulary. But how can we develop a method to create small phonological assessments capable of testing all phonemes from a given set, with predefined redundancy, while also defining a screening process that provides an indication of phonological deviation? To answer this question, the present research aimed to modernize and optimize the phonological evaluation process, traditionally manual and paper­based, through the implementation of a digital approach using advanced data structures.In this work, a phonological screening system was proposed, exemplifying its functionality in a prototype mobile application, using DT to adapt the assessment to each child’s difficulty level. The choice of the next word to be pronounced in the assessment is based on human validation of children’s speech, ensuring that each word used in the screening process is appropriate and relevant for phonological evaluation. Thus, we introduced a customized screening process with a maximum of 7 words to be evaluated, designed to quickly identify children with phonological deviations, optimizing the speech therapists’ time and prioritizing cases with suspected phonological deviations. Human validation ensures that no accuracy is lost compared to traditional methods, maintaining the quality of phonological assessment. The results indicate that the proposed approach not only maintains accuracy close to that of traditional assessments but also offers significant benefits in terms of time and resources, as we have shown that by using our method, there was about a 30% reduction in the number of words actually required in a phonological evaluation. Furthermore, the research highlights the potential future integration of automatic classification services for continuous comparison and improvement of the system. Finally, the implementation of a DT­based phonological screening system represents a significant contribution to the field of speech therapy through computing. This method does not aim to replace full phonological evaluations but rather to complement and enhance the existing process, providing an effective tool for initial screening. The developed prototype demonstrates the feasibility and benefits of the proposed technology, encouraging the adoption of technological methods in speech therapy practice.