Kirupa Ganapathy

Video based automatic fall detection in indoor environment

An increase in population of elderly people living in isolated environments, leads to the need for an automated monitoring system. Fall is one of the major reasons for death of elderly people. So fall detection is an essential part of an automated indoor monitoring system. Video based automatic fall detection is robust and more reliable than other fall detection methods. Most of the available video based fall detection mechanisms are based on the extraction of motion dynamic features like velocity and intensity gradient of the person in video. These methods are computationally intensive and less accurate. This paper presents an accurate and computationally less intensive approach to detect fall, using only the static features of the person such as aspect ratio and inclination angle. Experimental results show that this method is robust in detecting all kinds of fall.

Hierarchical Particle Swarm Optimization with Ortho-Cyclic Circles

Designed for real world problems in order to handle changing environment.Significant out performance in comparison with the various fitness function and other existing PSO algorithms.Near optimal solution is obtained for any number of swarms and swarm particles.Maximum of customers are served with reduced service waiting time.Avoids the necessity of rediversification, thereby reducing the convergence problem. Cloud computing is an emerging technology which deals with real world problems that changes dynamically. The users of dynamically changing applications in cloud demand for rapid and efficient service at any instance of time. To deal with this paper proposes a new modified Particle Swarm Optimization (PSO) algorithm that work efficiently in dynamic environments. The proposed Hierarchical Particle Swarm Optimization with Ortho Cyclic Circles (HPSO-OCC) receives the request in cloud from various resources, employs multiple swarm interaction and implements cyclic and orthogonal properties in a hierarchical manner to provide the near optimal solution. HPSO-OCC is tested and analysed in both static and dynamic environments using seven benchmark optimization functions. The proposed algorithm gives the best solution and outperforms in terms of accuracy and convergence speed when compared with the performance of existing PSO algorithms in dynamic scenarios. As a case study, HPSO-OCC is implemented in remote health monitoring application for optimal service scheduling in cloud. The near optimal solution from HPSO-OCC and Dynamic Round Robin Scheduling algorithm is implemented to schedule the services in healthcare.

Cloud-enabled remote health monitoring system

Sensor Web Enablement (SWE) for health care allows the access of sensor data anytime, anywhere using standard protocol and Application Program Interface (API). In this paper Open Geo-Spatial Consortium (OGC) standard based remote health monitoring system is proposed that allows integration of sensor and web using standard web based interface. The aim is to provide the data in an open & interoperable manner, and reduce data redundancy. Fixed specification is used for exchange of sensor data globally for all sensor networks. OGC SWEis applicable to different sensor systems including medical sensor networks. A standard format is used to document sensor descriptions and encapsulate data. Sensor data is ported on to cloud which provides scalability, centralized user access, persistent data storage and no infrastructure maintenance cost for heavy volumes of sensitive health data. Decision tree pruning algorithm with high confidence factor is proposed for automatic decision making.

Multi-sensor based in-home health monitoring using Complex Event Processing

Web enablement of sensor data is an important mission in providing anywhere anytime access. Particularly, it is essential in Tele-health monitoring of geriatric patient who are alone at home. For continuous monitoring, it requires the patient to wear wireless body sensors which give information about his vital parameters. Sensor Web Enablement (SWE) provides a platform for making the raw sensor data available on the web so that it becomes accessible to doctors for making clinical diagnosis. The challenges involved are effective collection of sensor data and bringing them to web using Service Oriented Architecture (SOA), complexity in finding relationship between raw events, developing rules for identifying patterns that pose a threat and generating alerts to patient, caregivers and doctors, dealing storage and privacy issues in accessing data on web and developing algorithm for fast and accurate fall detection. This talk addresses those challenges by providing possible solutions.

Optimum steepest descent higher level learning radial basis function network

We model a neural network with adaptive structure and dynamic learning.Higher level learning components helps the network to think before learning.Number of hidden neurons are decreased and redundant samples are removed.Increases classification accuracy, reduces detection time & architecture complexity.Various abnormality levels in vital parameters of multiple patients are classified. Dynamically changing real world applications, demands for rapid and accurate machine learning algorithm. In neural network based machine learning algorithms, radial basis function (RBF) network is a simple supervised learning feed forward network. With its simplicity, this network is highly suitable to model and control the nonlinear systems. Existing RBF networks in literature are applied to static applications and also faces challenges such as increased model size, neuron removal, improper center selection etc leading to erroneous output. To overcome the challenges and handle complex real world problems, this paper proposes a new optimum steepest descent based higher level learning radial basis function network (OSDHL-RBFN). The proposed OSDHL-RBFN implements major components inspired from the human brain for efficient learning, adaptive structure and accurate classification. Higher level learning and thinking components of the proposed network are sample deletion, neuron addition, neuron migration, sample navigation and neuroplasticity. These components helps the classifier to think before learning the samples and regulates the learning strategy. The knowledge gained from the trained samples are used by the network to identify the incomplete sample, optimal center and bond strength of hidden & output neurons. Adaptive network structure is employed to minimize classification error. The proposed work also uses optimum steepest descent method for weight parameter update to minimize the sum square error. OSDHL-RBFN is tested and evaluated in both static and dynamic environments on nine benchmark classification (binary and multiclass) problems for balanced, unbalanced, small, large, low dimensional and high dimensional datasets. The overall and class wise efficiency of OSDHL-RBFN is improved when compared to other RBFNs in the literature. The performance results clearly show that the proposed OSDHL-RBFN reduces the architecture complexity and computation time compared to other RBFNs. Overall, the proposed OSDHL-RBFN is efficient and suitable for dynamic real world applications in terms of detection time and accuracy. As a case study, OSDHL-RBFN is implemented in real time remote health monitoring application for classifying the various abnormality levels in vital parameters.

Sensor based decision making inference system for remote health monitoring

Fall occurring in older people is of major concern in medical environment as they are more prone to unexpected and unpredictable falls. Such falls often leads to injury and death in elderly. Hence, an automated fall detection mechanism using multiple sensing and event detection methods are required to analyze the activities of elderly. This paper presents a system considering 3-axis accelerometer sensor to detect the fall in home environment. Intelligent modeling technique such as ANFIS (Adaptive Neuro-Fuzzy Inference System) classifier is employed in this paper to detect the fall with reduced computational complexity and more accuracy. Using ANFIS, the data obtained from 3-axis accelerometer is classified under one of the five states (standing, sitting, walking, falling and lying) and backpropogation method is used for weight updation. Weighted average method which provides crisp value is used for de-fuzzification process. Features such as mean, median and standard deviation are considered for training the neural network. When the activity is recognized as fall, the patient heart rate and ECG are examined to detect abnormality and alarm is raised.

Virtualization of healthcare sensors in cloud

Web enablement of sensors is found to improve healthcare applications. However usage of too many physical sensors are limiting the mobility of the patients and has serious drawbacks like huge power consumption, heavy weight, health hazards and high degree of discomfort. To overcome these problems, this paper proposed a virtual sensor that computes the blood pressure of patients using a limited number of physical sensors. The sensor data are ported to cloud for achieving scalability, elasticity, on-demand service provision and availability. The developed virtual sensor has shown an accuracy of about 95% when tested on patients with normal vital sign parameters.

Dynamic higher level learning Radial Basis Function for healthcare application

Neural Network making use of Radial Basis Function (RBF) in the hidden layer maps the input of a lower dimension to a higher dimensional space in order to make the input linearly separable. The traditional RBF model is normally referred as cognitive component. The major issues in the traditional model are large number of fixed neurons, use of complete training set, prior center selection etc,. These issues increase the computation time and architecture complexity. To overcome these issues, this paper proposes a novel Dynamic Higher Level Learning RBF (DHLRBF) architecture suitable for dynamic environment. The learning process of the cognitive component is controlled by the Higher Level Learning component such as Neuron addition and Sample deletion. The proposed work is applied for Health parameters to classify normal and abnormal category. The proposed DHLRBF is implemented and the results show that the model is efficient in terms of detection accuracy and time.

Medical intelligence for quality improvement in Service Oriented Architecture

In remote healthcare process, patient monitoring system is used to collect health data at home and in outdoor scenarios to facilitate disease management. Service Oriented Architecture (SOA) is used to build a comfortable web based management system for integrating distributed wireless sensor network (WSN) and internet. This paper proposes SOA based data architecture for healthcare. As a case study, patient data monitoring system is taken for implementation. It is not feasible for the caregiver to monitor, analyze voluminous data continuously and generate alerts. Data from the physiological sensors and the history data about the patient are used for analysis and alert generation. It is observed that the proposed architecture supports efficient information retrieval. The aggregated data in the server are sliced and diced by customized hand coding ETL (Extract Transform and Load) process. OBIEE (oracle business intelligence enterprise edition) tool is used to develop this application. This novel approach demonstrates data analysis subsystem of SOA for pattern identification and alert generation.

A genetic approach for personalized healthcare

Remote Health monitoring involves continuous monitoring of vital signs and transmission of alert signals to the physician when vital sign values fluctuate above or below the threshold. Existing healthcare systems obtain the vital data of a patient periodically and require the intervention of a doctor to detect the severity of abnormality which is time consuming. Hence, there is a need for an intelligent, personalized and efficient healthcare system to detect the abnormality. In a multipatient environment when several patients have abnormalities, existing scheduling schemes do not consider the degree of severity in order to schedule the most critical patient who has to be served first. To overcome these issues, this paper proposes a Genetic Algorithm (GA) based Personalized Healthcare System (GAPHS). This system represents the abnormality levels of the vital parameters of the patient as a chromosome and determines the Severity Index of the chromosome to identify the severity. The proposed system outperforms in terms of speed and accuracy when compared to traditional GA in dynamic scenarios.