Aprendizado de máquina para análise de recaída para depressão em pacientes com transtorno bipolar
Nenhuma Miniatura disponível
Data
2018-10-04
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Goiás
Resumo
Depression relapse in patients with Bipolar Disorder (BD) have 70% rate of recurrence in the first 4 years of treatment and may cause a severe loss of quality of life and even lead to suicide. BD is a mood disorder characterized by recurrent episodes of depression or mania. To study the disorder and find more efficient treatments, the Harvard Medical School created the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD). It is a widely used dataset that comprises data of 4,360 patients with BD, which can be considered one of the most complete databases in terms of scope nowadays. Several studies have been developed to discover more efficient treatments to prevent relapses in BD. However, most of them used only classical statistical methods, mainly aimed at measuring its correlation to specific features. This study presents an analysis of the use of machine learning algorithms to discover patterns related to depression relapse in BD with the use of longitudinal data provided by STEP-BD. This longitudinal data includes 148 features collected in 50,987 visits of patients spread across different weeks over the years. Thus, several experiments were conducted and the results show that the algorithms attained limited performance. We concluded that features related to depression and mania mood states, collected by the STEP-BD, cannot be used properly to predict the relapse to depression before it occurs, being suited only as an indicator that the patient is already in the state of depression.
Descrição
Palavras-chave
Citação
BORGES JÚNIOR, R. G. Aprendizado de máquina para análise de recaída para depressão em pacientes com transtorno bipolar. 2018. 83 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Goiás, Goiânia, 2018.