ICFICE 2018
Current Result Document :
| 한글제목(Korean Title) |
Different Feature Selection Techniques Based on Gait Parameters |
| 영문제목(English Title) |
Different Feature Selection Techniques Based on Gait Parameters |
| 저자(Author) |
Satyabrata Aich
Kiwon Choi
Kim Hee-Cheol
|
| 원문수록처(Citation) |
VOL 10 NO. 01 PP. 0355 ~ 0358 (2018. 06) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
The aim of the paper is to discriminate the Parkinson’s disease(PD) from other nerological diseases such as Amyotrophic Lateral Sclerosis (ALS), Huntington’s disease (HD),using different feature selction technique such as Prinicipal Componet Analysis and Recursive Feature Elimination, and also different classification approaches that includes linear classifiers,nonlinear classifiers,and Probabilistic classifier. Mostly the neurological disease affect the brain cells and spinal cords and that lead to the abnormality in the gait behaviour.Recetly the gait analysis caught much attention for predciting the neurological diseases such as PD, ALS and HD.With the advent of intelligent machine learning technique an attempt has been made in this paper to discriminate the PD from other neurological patients. In this paper the gait dynamics data of pateints suffered with PD,ALS and HD has been collected from the public database and a binary classification approach has been used by taking PD as one group and ALS HD as the other group.Our results founds highest accuracy of 96.78% using SVM with Radial Basis Function(RBF) combined with reduced feature sets obtained from RFE compared to the other classifiers with feature sets from PCA.The result provided enough evidence that can asssit medical practitioners to distinguish PD from other neurological diseases by using the gait dynamics data.
|
| 키워드(Keyword) |
|
| 파일첨부 |
PDF 다운로드
|
