Article

An automated hierarchical gait pattern identification tool employing cross-correlation-based feature extraction and recurrent neural network based classification.

Expert Systems (Impact Factor: 0.75). 05/2009; 26:202-217. DOI: 10.1111/j.1468-0394.2009.00479.x
Source: DBLP

ABSTRACT In this paper Elman's recurrent neural network (ERNN) is employed for automatic identification of healthy and pathological gait and subsequent diagnosis of the neurological disorder in pathological gaits from the respective gait patterns. Stance, swing and double support intervals (expressed as percentages of stride) of 63 subjects were analysed for a period of approximately 300 s. The relevant gait features are extracted from cross-correlograms of these signals with corresponding signals of a reference subject. These gait features are used to train modular ERNNs performing binary and tertiary classifications. The average accuracy of binary classifiers is obtained as 90.6%–97.8% and that of tertiary classifiers is 89.8%. Hence, two hierarchical schemes are developed each of which uses more than one modular ERNN to segregate healthy, Parkinson's disease, Huntington's disease and amyotrophic lateral sclerosis subjects. The average testing performances of the schemes are 83.8% and 87.1%.

0 Followers
 · 
70 Views
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Human activity recognition in everyday environments is a critical, but challenging task in Ambient Intelligence applications to achieve proper Ambient Assisted Living, and key challenges still remain to be dealt with to realize robust methods. One of the major limitations of the Ambient Intelligence systems today is the lack of semantic models of those activities on the environment, so that the system can recognize the specific activity being performed by the user(s) and act accordingly. In this context, this thesis addresses the general problem of knowledge representation in Smart Spaces. The main objective is to develop knowledge-based models, equipped with semantics to learn, infer and monitor human behaviours in Smart Spaces. Moreover, it is easy to recognize that some aspects of this problem have a high degree of uncertainty, and therefore, the developed models must be equipped with mechanisms to manage this type of information. A fuzzy ontology and a semantic hybrid system are presented to allow modelling and recognition of a set of complex real-life scenarios where vagueness and uncertainty are inherent to the human nature of the users that perform it. The handling of uncertain, incomplete and vague data (i.e., missing sensor readings and activity execution variations, since human behaviour is non-deterministic) is approached for the first time through a fuzzy ontology validated on real-time settings within a hybrid data-driven and knowledge-based architecture. The semantics of activities, sub-activities and real-time object interaction are taken into consideration. The proposed framework consists of two main modules: the low-level sub-activity recognizer and the high-level activity recognizer. The rst module detects sub-activities (i.e., actions or basic activities) that take input data directly from a depth sensor (Kinect). The main contribution of this thesis tackles the second component of the hybrid system, which lays on top of the previous one, in a superior level of abstraction, and acquires the input data from the first module's output, and executes ontological inference to provide users, activities and their influence in the environment, with semantics. This component is thus knowledge-based, and a fuzzy ontology was designed to model the high-level activities. Since activity recognition requires context-awareness and the ability to discriminate among activities in different environments, the semantic framework allows for modelling common-sense knowledge in the form of a rule-based system that supports expressions close to natural language in the form of fuzzy linguistic labels. The framework advantages have been evaluated with a challenging and new public dataset, CAD-120, achieving an accuracy of 90.1% and 91.1% respectively for low and high-level activities. This entails an improvement over both, entirely data-driven approaches, and merely ontology-based approaches. As an added value, for the system to be sufficiently simple and flexible to be managed by non-expert users, and thus, facilitate the transfer of research to industry, a development framework composed by a programming toolbox, a hybrid crisp and fuzzy architecture, and graphical models to represent and con gure human behaviour in Smart Spaces, were developed in order to provide the framework with more usability in the final application. As a result, human behaviour recognition can help assisting people with special needs such as in healthcare, independent elderly living, in remote rehabilitation monitoring, industrial process guideline control, and many other cases. This thesis shows use cases in these areas.
    04/2015, Degree: Artificial Intelligence and Embedded systems, Supervisor: Johan Lilius and Miguel Delgado Calvo-Flores
  • [Show abstract] [Hide abstract]
    ABSTRACT: For the purpose of realizing an intelligent and highly accurate diagnosis system for neuro-degenerative diseases (NDD), such as amyotrophic lateral sclerosis (ALS), Parkinson's disease (PD) and Huntington's disease (HD), the present study investigated the classification capability of different gait statistical features extracted from gait rhythm signals. Nine statistical measures, including several seldom-used variability measures for these signals, were calculated for each time series. Next, after an evaluation of four popular machine learning methods, the optimal feature subset was generated with a hill-climbing feature selection method. Experiments were performed on a data set with 16 healthy control (CO) subjects, 13 ALS, 15 PD and 20 HD patients. When evaluated with the leave-one-out cross-validation (LOOCV) method, the highest accuracy rate for discriminating between groups of NDD patients and healthy control subjects was 96.83%. The best classification accuracy (100%) was obtained with two subtype binary classifiers (PD vs. CO and HD vs. CO).
    Biomedical Signal Processing and Control 04/2015; 18. DOI:10.1016/j.bspc.2015.02.002 · 1.53 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Huntington's disease (HD) is one of the neural diseases with movement disorders like chorea, ballism, and athetosis. These symptoms play an important role on gait asymmetry, thus gait recordings are considered important resources for studying this disease. Diagnosing HD in its early stages is important and critical. We aimed to discover differences between HD and normal behavior. As the gait is semi-periodic, analyzing frequency could reveal disorders. We made an attempt to extract some proper features using power spectra density. Studies revealed that HD adds high frequencies to the spectrum on all gait phases. Statistical analysis of the gait features showed significant differences between normal and HD groups. At the end, we tried to separate the patients and healthy individuals using a new intelligent mathematical system. An artificial neural network classifier was used for this reason,and our best separation accuracy was 96.6%. This study could be the basis of designing a practical decision support system. This system can diagnose patients at the first stages of the disease, and it also can recommend suspected persons to the specialist.
    Journal of Mechanics in Medicine and Biology 02/2014; 14(01). DOI:10.1142/S0219519414500018 · 0.80 Impact Factor