Journal for Research | Volume 02 | Issue 12 | February 2017 ISSN: 2395-7549 All rights reserved by www.journalforresearch.org 1 Assessment of Degree of Severity of Parkinson’s Disease using Gait Signal Envelope H K Shreedhar Dr. Anandthirtha B Gudi Associate Professor Professor & PG Coordinator Department of Electrical & Computer Engineering Department of Electrical & Computer Engineering Global Academy of Technology, Bengaluru Global Academy of Technology, Bengaluru Abstract Neurological disorders disrupt the daily activity of many human beings. Recent survey has shown that Parkinson’s disease is more predominant in the present scenario in the age group of 65 and above. Using appropriate sensors gait signals are obtained from both normal and pathological subjects. Such signals are processed to determine start time of a stride and envelope of gait signal. Further two different methods are implemented using gait envelope to determine the degree of severity of Parkinson’s disease in pathological subjects in comparison with normal. Keywords: Envelope, Gait Signal, Parkinson’s disease, Severity, Start Time _______________________________________________________________________________________________________ I. INTRODUCTION Present worldwide survey shows that around 1% of people older than 60 years and above suffer from Parkinson's disease and majority of them develop several speech impairments [1]. Parkinson’s disease is one of the chronic neural disorders which aff ect the people over the worldwide. PD symptoms include tremor, rigidity and loss of muscle control in general, as well as cognitive impairment [2]-[3].This is a progressive disease with a slow progressive rate and gradually results in severe disability in terms of walking ability, speech and in handling of day to day routine activities [4]. Computer-aided diagnosis system for the automatic evaluation of Parkinson disease may provide useful information to the medical practioner to make more accurate diagnosis and monitoring of PD patients and it can also help them to evaluate the severity level of disease periodically [5]. Gait is one of the behavioral biometrics and is defined as the manner of walking [6]. It may be considered as an effective tool to identify a person. Freezing of Gait (FOG) occurs quite commonly among persons with advanced Parkinson’s disease [7].Few researches have discussed how to distinguish normal pressure hydrocephalus and Parkinson’s disease [8] and other researchers have proposed methods to determine the contribution of abnormal leg muscle activation to freezing in patients with Parkinson's disease [9]. II. METHODOLOGY Normal and pathological subjects are allowed to walk on level ground at their usual pace for two minutes. Under each foot, eight sensors are placed. These sensors measure vertical ground reaction force as a function of time. The output of total sixteen sensors are digitized and recorded at 100 samples per second. The sum of eight sensor output for each foot is calculated and included in the data. This signal database is available in physionet and is used for the current study. Initially, ten numbers of healthy and normal persons are identified and they are treated as control subjects. The data of these subjects is used to create the reference values. The analysis is repeated for pathological subjects and compared with that of normal subjects to identify the level of severity of Parkinson’s disease. In each case of normal/ pathological subjects, one hundred and twenty seconds of walking/gait data is chosen. This consists of approximately 12000 samples, with sampling rate of 100 samples per second. In this data, first one thousand samples and last one thousand samples are removed. Starting thousand samples are eliminated since those samples may be appearing in the transition phase of rest to normal walking. The last thousand samples are eliminated since they represent transition period between normal walking to rest. After removal of these two thousand samples, intermediate ten thousand samples are considered for the data analysis. Such ten thousand samples are further divided into ten groups, each consisting of one thousand samples. The analysis is carried out using two methods. In the first method, start time of each step is calculated and using this information envelope is determined for both normal and pathological subjects. Comparing the normal and pathological subjects’ envelope, degree of severity is assessed. In the second method, after calculating envelope, standard deviation is obtained for each subject. Depending upon the values of standard deviation, severity is assessed for pathological subjects in comparison with the normal. Method 1 For start time detection, either left foot or on right foot samples may be chosen. Since both of them yield similar results, in the present work, analysis is carried out using left foot samples. For each subject, data representing sum of the 8 sensor outputs of left foot is considered and is normalized by dividing it by maximum value. One group of thousand samples is considered. Such