Estimating the family bias to autism: a bayesian approach

• Ano de defesa: 2022
• Autor(a) principal: CARVALHO, Emerson Assis de
• Orientador(a): BASTOS, Guilherme Sousa
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Itajubá
• Programa de Pós-Graduação: Programa de Pós-Graduação: Doutorado - Engenharia Elétrica
• Departamento: IESTI - Instituto de Engenharia de Sistemas e Tecnologia da Informação
• País: Brasil
• Palavras-chave em Português: Autism spectrum disorder prevalence; Autism spectrum disorder etiology; Probabilistic graphical models; Bayesian networks; Markov models
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS::ENGENHARIA ELÉTRICA
• Link de acesso: https://repositorio.unifei.edu.br/jspui/handle/123456789/3193

Autism is an age- and sex-related lifelong neurodevelopmental condition characterized pri marily by persistent deficits in core domains such as social communication. It is estimated that ≈ 2% of children have some ASD trait. The autism etiology is mainly due to inherited genetic factors (>80%). The importance of early diagnosis and interventions motivated several studies involving groups at high risk for ASD, those with a greater predisposition to the disorder. Such studies are characterized by evaluating some characteristics of the individual itself or the family members of diagnosed individuals, mainly aiming to predict a future diagnosis or recurrence rates. One of the primary goals of Artificial Intelligence is to create artificial agents capable of intelligent behaviors, such as prediction problems. Prediction problems usually involve reasoning with uncertainty due to some information deficiency, in which the data may be imprecise or incorrect. Such solutions may seek the application of probabilistic methods to construct inference models. In this thesis, we will discuss the development of probabilistic networks capable of estimating the risk of autism among the family members given some evidence (e.g., other family members with ASD). In particular, the main novel contributions of this thesis are as follows: the proposal of some estimates regarding parents with ASD generating children with ASD; the highlight ing regarding the decrease in the ASD prevalence sex ratio among males and females when genetic factors are taken into account; the corroboration and quantification of past evidence that the clustering of ASD in families is primarily due to genetic factors; the computation of some estimates regarding the risk of ASD for parents, grandparents, and siblings; an estimate regarding the number of ASD cases in a family sufficient to attribute the ASD occurrences to the genetic inheritance; the assessment of some estimates for males and females individuals given evidence in grandparents, aunts-or-uncles, nieces-or nephews and cousins; and the proposition of some estimates indicating risk ranges for ASD by genetic similarity.