Matthias Struck

User-friendly system for recognition of activities with an accelerometer

Monitoring of a persons daily activities can provide valuable information for health care and prevention and can be an important supportive application in the field of ambient assisted living (AAL). The goals of this study are the classification of postures and activities using knowledge-based methods as well as the evaluation of the performance of these methods. The acceleration data are gained by a single tri-axial accelerometer, which is mounted on a specific position on the test subject. A data set for training and testing was gained by collecting data from subjects, who performed varying postures and activities. For these purposes, three different knowledge-based (decision tree and neural network) classification methods and a hybrid classifier were implemented, tested and evaluated. The results of the tests illustrated that the hybrid classifier performed best with an overall accuracy of 98.99%. The advantages of knowledge-based methods are the exchangeable knowledge base, which can be developed for different types of sensor positions and the state of health of the subject.

A new real-time fall detection approach using fuzzy logic and a neural network

A real-time fall detection system monitors the daily activity of especially elderly people to enlist someones help as fast as possible in case of emergency. This paper presents a new real-time fall detection algorithm using a single commercial accelerometer. After transforming the acceleration data from Cartesian coordinates to spherical coordinates, the main part of the algorithm is based on a fuzzy logic inference system and a neural network. These methods allow both the integration of specific expert knowledge about typical falls as well as generalization ability. In order to compare the achieved performance of the method to those of literature, four fall scenarios (forward, backward, sideward and collapse) were performed and evaluated in a laboratory trial with, in the first instance, 5 test subjects. The average sensitivity of those four fall scenarios reached 94% and the false positive rate was about 0.35%. These results show that one single accelerometer is completely sufficient to implement a reliable fall detection system and, furthermore, that knowledge based methods are a suitable alternative to standard pattern recognition methods.

Sampling rate impact on energy consumption of biomedical signal processing systems

Long battery runtime is one of the most wanted properties of wearable sensor systems. The sampling rate has an high impact on the power consumption. However, defining a sufficient sampling rate, especially for cutting edge mobile sensors is difficult. Often, a high sampling rate, up to four times higher than necessary, is chosen as a precaution. Especially for biomedical sensor applications many contradictory recommendations exist, how to select the appropriate sample rate. They all are motivated from one point of view - the signal quality. In this paper we motivate to keep the sampling rate as low as possible. Therefore we reviewed common algorithms for biomedical signal processing. For each algorithm the number of operations depending on the data rate has been estimated. The Bachmann-Landau notation has been used to evaluate the computational complexity in dependency of the sampling rate. We found linear, logarithmic, quadratic and cubic dependencies.

Method and system for standardized and platform independent medical data information persistence in telemedicine

Wireless communication between sensors monitoring patient vital signs has become more and more important in the past few years. Essential requirements to integrate sensors of different manufacturers into a clinical network are standardized communication protocols and a unique data representation of the vital signs. Both issues are realized by CEN ENV 13734/35 ldquoVital Signs Information Representationrdquo (VITAL) [1]. The standard was implemented and integrated into a generic framework with different interfaces allowing integration of extensions [2]. In order to guarantee readability of vital signs communicated with the VITAL framework in the future, standardized storing methods are indispensable. That is why a new user interface was created that allows standardized persistence of medical data information in real time. Focussing on our implementation, we evaluated three relevant file formats: the ldquoEuropean Data Formatrdquo (EDF/EDF+), the ldquoFile Exchange Formatrdquo (FEF) and SCIPHOX.

Feasibility of a wireless health monitoring system for prevention and health assessment of elderly people

The work presented in this paper comprises the methodology and results of a pilot study on the feasibility of a wireless health monitoring system designed under main EU challenges for the promotion of healthy and active ageing. The system is focused on health assessment, prevention and lifestyle promotion of elderly people. Over a hundred participants including elderly users and caregivers tested the system in four pilot sites across Europe. Tests covered several scenarios in senior centers and real home environments, including performance and usability assessment. Results indicated strong satisfactoriness on usability, usefulness and user friendliness, and the acceptable level of reliability obtained supports future investigation on the same direction for further improvement and transfer of conclusions to the real world in the healthcare delivery.

Evaluation of QRS detection algorithm implemented for mobile applications based on ECG data acquired from sensorized garments

People can greatly benefit from mobile technologies that continuously monitor their vital signs, in medicine as well as in home environments and sports. In order to meet the requirements of mobile systems the algorithms have to be robust, reliable, take the limited resources into account and overcome the drawback of motion artefacts. This paper presents the evaluation of an algorithm for QRS detection based on ECG signals from a sensorized garment. The system saves the ECG data, measured via two textile electrodes sewed into the shirt, on a microSD card using the EDF+-format. The raw data is processed on a desktop PC using a modified state-of-the-art algorithm. QRS complexes and R-peaks of electrocardiographic signals are detected using the technique of zero crossings. Hereby, main focus has to be placed on the proper specification of the band pass filter, which is the basis for high accuracy. For the evaluation a well-defined test protocol has been specified. Six activities respectively postures were defined: Sitting, standing, walking, running, cycling and rowing. Each activity was performed by 10 test persons for a fixed time interval. Various parameters, where the temporal location of the R-peak is of importance, can be derived from the recorded ECG raw data, such as heart rate, heart rate variability or ECG classification. This method is robust and provides high accuracy even in case of noisy signals. Motion artefacts could be compensated on a high level. The performed study illustrates that even validated state-of-the-art R-peak detection algorithms have to be adapted and optimized for the mobile and daily usage. Due to its computational efficiency it is suitable for mobile applications in real-time.

Filter and processing method to improve R-peak detection for ECG data with motion artefacts from wearable systems

The electrocardiogram (ECG) is one of the most reliable information sources for assessing cardiovascular health and training success. Since the early 1990s, the heart rate variability (HRV), namely the variation from beat to beat, has become the focus of investigations as it provides insight into the complex interplay of body circulation and the influence of the autonomic nervous system on heartbeats. However, HRV parameters during physical activity are poorly understood, mostly due to the challenging signal processing in the presence of motion artefacts. To derive HRV parameters in time (heart rate (HR)) and frequency domains (high frequency (HF), low frequency (LF)), it is crucial to reliably detect the exact position of the R-peaks. We introduce a full algorithm chain where a sophisticated filtering technique is combined with an enhanced R-peak detection that can cope with motion artefacts in ECG data originating from physical activity.

Method for daily-life movement classification of elderly people

Activity monitoring using accelerometers is of growing interest. Acceleration contains information about intensity and frequency of activities. An additional barometer provides information on altitude. This paper presents an evaluation of several features with respect to their suitability for recognizing resting, walking, cycling, going upand downstairs. Two classifiers are proposed: one using a barometer and the other neglecting the barometer. The classifiers were trained with reference data from 25 older adults, which performed a predefined set of activities of daily living. Both classifiers correctly recognized 82.7% of resting and 83.8% of cycling. Using a barometer, classification rates of 87.3% for walking, 82.9% for going stairs up and 77.4% for going stairs down were achieved. In contrast, neglecting the barometer resulted in 70.8% for normal walking, 79.7% for going stairs up and 71.8% for going stairs down.

Battery runtime optimization toolbox for wearable biomedical sensors

Battery runtime is a critical concern for practical usage of wearable biomedical sensor systems. A long runtime requires an interdisciplinary low-power knowledge and appropriate design tools. We addressed this issue designing a toolbox in three parts: (1) Modular evaluation kit for development of wearable ultra-low-power biomedical sensors; (2) Miniaturized, wearable, and code compatible sensor system with the same properties as the development kit; (3) Web-based battery runtime calculator for our sensor systems. The purpose of the development kit is optimization of the power consumption. Once optimization is finished, the same embedded software can be transferred to the miniaturized body worn sensor. The web-based application supports development quantifying the effects of use case and design decisions on battery runtime. A sensor developer can select sensor modules, configure sensor parameters, enter use case specific requirements, and select a battery to predict the battery runtime for a specific application. Our concept adds value to development of ultra-low-power biomedical wearable sensors. The concept is effective for professional work and educational purposes.

Human authentication implemented for mobile applications based on ECG-data acquired from sensorized garments

In recent years biometric systems gain more and more importance. Studies showed, that authentication with a clinical electrocardiogram (ECG) is principally possible and hence could be used as a biometric feature. In this work an algorithm was implemented. which is capable of segmenting single heartbeats of a mobile recorded single-channel-ECG. Based on these heartbeats, fiducial features, features from the combination of autocorrelation and discrete cosine transform, and wavelet features were extracted and considered for the classification process. They were evaluated concerning distinctiveness and stability over time. In order to reduce the feature space, sequential forward selection was used to eliminate unstable and non-distinctive features. A sensorized garment was used to derive ECG-signals from ten persons in order to evaluate the performance of the proposed methods. The wavelet-transform provides the best features since it is focusing on the characteristics of the QRS-complex of a human heartbeat, which provides the most stable information over time. Using the wavelet coefficients as features the developed authentication algorithm produced an equal error rate of 12.53 %.