Medically Driven Data Mining Application: Recognition of Health Problems from Gait Patterns of Elderly

Abstract

We propose a medically driven data mining application: system for diagnosing of gait patterns related to health problems of elderly to support their independent living. Gait of elderly is captured with motion capture system, which consists of tags attached to the body and sensors situated in the apartment. Position of the tags is acquired by the sensors and the resulting time series of position coordinates are analyzed with machine learning algorithms in order to recognize the specific health problem. We propose novel features for training a machine learning classifier that classifies the user's gait into: i) normal, ii) with hemiplegia, iii) with Parkinson's disease, iv) with pain in the back and v) with pain in the leg. Experimental results showed that decision tree classifier was able to reach only 95 % of classification accuracy, using 7 tags and 5 mm standard deviation of noise. On the other hand, k-nearest neighbor was much more accurate since it reached classification accuracy of over 99 %, using only 8 tags with 0-20 mm noise.

A Brief Literature Review of some Efficient Human Gait Analysis Based Gender Classification Techniques

“…Gait based gender recognition has become an interesting choice for researchers as it can be detected at a distance, is non-contact and non-invasive as opposed to the other two techniques. There has been some astounding work in this area in the past [30] [31]. In some attempts for gait-based gender classification, motion trackers motion tracking equipment are fixed at the major joints of the subject's body, and they were made to walk while wearing swimsuits [8].…”

Ambient media today and tomorrow

“…This recognition is based on the observation and analysis of the characteristics of their movement. The first approach presented in the study uses medically defined attributes and supports vector machine classification into the chosen five health states [69]. The second approach makes a classification of the same five health states using a more general data mining approach.…”

“…The paper of [70] presents the system as an example of the semantic ambient media for ambient assisted living. The motivation for the paper is the fact that the percentage of elderly people is increasing worldwide and this is particularly so in Europe [69], where the elderly tend to lead an isolated life away from their children. As a result, they may fear not being able to obtain help if they are injured or ill.…”

Human Gait Gender Classification in Spatial and Temporal Reasoning

“…There is some temporal information such as stride and cadence parameter per frame, angle of every knee per frame, and hip angle per frame. Some other classification might provide a better result to the recognition of human gait gender such as a k-nearest neighbor [14].…”