Título: Detecção Precoce de Alzheimer Usando Machine Learning
Autores: Nathalia Paiva and Tatiana Escovedo.
Resumo:
Alzheimer’s disease is a neurodegenerative disease,
responsible for approximately 60% to 80% of cases of
dementia in the world and, as it still has no cure, it can end the
life or family of the individual who presents its symptoms. Its
diagnosis is made through the results of laboratory exams,
cognitive tests and image exams, and MRIs are the most
accurate method. As it is a disease without cure, the ideal is
that it is diagnosed as soon as possible, so that doctors can slow
down its evolution and the patient can maintain a life without
depending on other people. Knowing that data science has
been shown to be effective in supporting diagnosis in several
areas of medicine, that is why in the present work we tested
the performance of intelligent algorithms in the classification
of characteristics extracted from magnetic resonance images.
We achieved an accuracy of approximately 90% by predicting
cognitive decline based on characteristics extracted from the
OASIS-1 dataset, made available by the Open Access Series
of Imaging Studies, using the logistic regression algorithm.
Different classifier configurations were used in order to
evaluate them, based on maximizing performance and
minimizing computational cost. We concluded that it is
possible to predict the disease using characteristics such as
age, socioeconomic status and level of education.
Palavras-chave:
Alzheimer, Dementia, OASIS-1, Machine Learning, Data Science.
Páginas: 7
Código DOI: 10.21528/CBIC2021-2
Artigo em pdf: CBIC_2021_paper_2.pdf
Arquivo BibTeX: CBIC_2021_2.bib