The World Health Organization (WHO) in 2016 considered m-health as: “the use of mobile wireless technologies including smart devices such as smartphones and smartwatches for public health”. WHO emphasizes the potential of this technology to increase its use in accessing health information and services as well as promoting positive changes in health behaviours and overall management of diseases. In this regard, the capability of smartphones and smartwatches for m-health monitoring through the collection of patient data remotely, has become an important component in m-health system. It is important that the integrity of the data collected is verified continuously through data authentication before storage. In this research work, we extracted heart rate variability (HRV) and decomposed the signals into sub-bands of detail and approximation coefficients. A comparison analysis is done after the classification of the extracted features to select the best sub-bands. An architectural framework and a used case for m-health data authentication is carried out using two sub-bands with the best performance from the HRV decomposition using 30 subjects’ data. The best sub-band achieved an equal error rate (EER) of 12.42%.

No abstract

With the widespread use of embedded sensing capabilities of mobile devices, there has been unprecedented development of context-aware solutions. This allows the proliferation of various intelligent applications, such as those for remote health and lifestyle monitoring, intelligent personalized services, etc. However, activity context recognition based on multivariate time series signals obtained from mobile devices in unconstrained conditions is naturally prone to imbalance class problems. This means that recognition models tend to predict classes with the majority number of samples whilst ignoring classes with the least number of samples, resulting in poor generalization. To address this problem, we propose augmentation of the time series signals from inertial sensors with signals from ambient sensing to train Deep Convolutional Neural Network (DCNNs) models. DCNNs provide the characteristics that capture local dependency and scale invariance of these combined sensor signals. Consequently, we developed a DCNN model using only inertial sensor signals and then developed another model that combined signals from both inertial and ambient sensors aiming to investigate the class imbalance problem by improving the performance of the recognition model. Evaluation and analysis of the proposed system using data with imbalanced classes show that the system achieved better recognition accuracy when data from inertial sensors are combined with those from ambient sensors, such as environmental noise level and illumination, with an overall improvement of 5.3% accuracy.

There has been exponential growth in the use of wearable technologies in the last decade with smart watches having a large share of the market. Smart watches were primarily used for health and fitness purposes but recent years have seen a rise in their deployment in other areas. Recent smart watches are fitted with sensors with enhanced functionality and capabilities. For example, some function as standalone device with the ability to create activity logs and transmit data to a secondary device. The capability has contributed to their increased usage in recent years with researchers focusing on their potential. This paper explores the ability to extract physiological data from smart watch technology to achieve user authentication. The approach is suitable not only because of the capacity for data capture but also easy connectivity with other devicesprincipally the Smartphone. For the purpose of this study, heart rate data is captured and extracted from 30 subjects continually over an hour. While security is the ultimate goal, usability should also be key consideration. Most bioelectrical signals like heart rate are nonstationary time-dependent signals therefore Discrete Wavelet Transform (DWT) is employed. DWT decomposes the bioelectrical signal into n level sub-bands of detail coefficients and approximation coefficients. Biorthogonal Wavelet (bior 4.4) is applied to extract features from the four levels of detail coefficents. Ten statistical features are extracted from each level of the coffecient sub-band. Classification of each sub-band levels are done using a Feedforward neural Network (FF-NN). The 1 st , 2 nd , 3 rd and 4 th levels had an Equal Error Rate (EER) of 17.20%, 18.17%, 20.93% and 21.83% respectively. To improve the EER, fusion of the four level sub-band is applied at the feature level. The proposed fusion showed an improved result over the initial result with an EER of 11.25%. As a one-off authentication decision, an 11% EER is not ideal, its use on a continuous basis makes this more than feasible in practice.

No abstract