ICA
  contato   links   
 

COMPARISON OF A GENETIC ALGORITHM NEURAL NETWORK WITH LOGISTIC

 
Disciplina: Computação Evolucionária (ELE2395)
Projeto: -
 
Link Original: http://www.asbweb.org/conferences/1990s/1998/162/index.html
 
Abstract
Ground reaction forces are often used to assess gait pattern. However, the simple description of the relationships between the parameters of ground reaction forces and gait patterns is not available yet. The changes in gait pattern will exert a significant effect on the magnitude or phase of ground reaction forces. So far, we do not have a mathematical model which gives a satisfactory prediction of the gait pattern by assessing the numerous factors that may influence a person's gait pattern. Genetic algorithm neural network (GANN) is often able to provide good answers to questions in the biomedical sciences (Miles F.J. et al., 1997, Harry P.S. et al., 1995, Narayanan M.N., 1993). It is a procedure that may be used to search among sets of clinical variables for those that are the best predictors. When a neural network uses a genetic algorithm for training there is an increase in computational time, but compared with simple gradient descent, optimization does not fall into local minimum and be more accurate in prediction. In this study, we compared the efficacy of a custom developed GANN method to predict gait pattern on ankle arthrodesis and compared with logistic regression method. Predictive methods were compared by examining accuracy of classifying target outcome of patients or controls. Achievement of this objective may help the therapists to design the therapeutic plan.
 



Palavras-Chave
Nenhuma palavra chave relacionada.



Arquivos
Nenhum arquivo cadastrado.