The number of people diagnosed with dementia is expected to rise in the coming years. Given that there currently no definite cure for and cost care this condition soars dramatically, slowing decline maintaining independent living are important goals supporting dementia. This paper discusses a study called Technology Integrated Health Management (TIHM). TIHM technology assisted monitoring system uses Internet Things (IoT) enabled solutions continuous their own homes. We have developed machine...

With the proliferation of sensors and IoT technologies, stream data are increasingly stored analyzed, but rarely combined, due to heterogeneity sources technologies. Semantics used share sensory data, not so much for annotating data. Semantic models annotation scarce, as generally semantics heavy process ideal Internet things (IoT) environments, where frequently updated. We present a light model semantically annotate streams, IoT-Stream. It takes advantage common knowledge sharing semantics,...

Dementia is a neurological and cognitive condition that affects millions of people around the world. At any given time in United Kingdom, 1 4 hospital beds are occupied by person with dementia, while about 22% these admissions due to preventable causes. In this paper we discuss using Internet Things (IoT) technologies in-home sensory devices combination machine learning techniques monitor health well-being dementia. This will allow us provide more effective preventative care reduce...

In this article, we discuss a technical design and an ongoing trial that is being conducted in the UK, called Technology Integrated Health Management (TIHM). TIHM uses Internet of Things-enabled solutions provided by various companies collaborative project. The Things (IoT) devices are integrated common platform supports interoperable open standards. A set machine-learning data analytics algorithms generate notifications regarding well-being patients. information monitored around clock group...

A novel quaternion-valued singular spectrum analysis (SSA) is introduced for multichannel of electroencephalogram (EEG). The EEG typically requires the decomposition data channels into meaningful components despite notoriously noisy nature EEG-which aim SSA. However, value involved in SSA implies strict orthogonality decomposed components, which may not reflect accurately sources exhibit similar neural activities. To allow modelling such co-channel coupling, quaternion domain considered...

Continual learning models allow them to learn and adapt new changes tasks over time. However, in continual sequential scenarios, which the are trained using different data with various distributions, neural networks (NNs) tend forget previously learned knowledge. This phenomenon is often referred as catastrophic forgetting. The forgetting an inevitable problem for dynamic environments. To address this issue, we propose a method, called Bayesian (CBLNs), enables allocate additional resources...

Sleep quality is an important determinant of human health and wellbeing. Novel technologies that can quantify sleep at scale are required to enable the diagnosis epidemiology poor sleep. One indicator body posture. In this paper, we present design implementation a non-contact monitoring system analyses posture movement. Supervised machine learning strategies applied noncontact vision-based infrared camera data using transfer approach, successfully quantified poses participants covered by...

The rapid growth of Internet Things (IoT) and sensing technologies has led to an increasing interest in time-series data analysis. In many domains, detecting patterns IoT interpreting these are challenging issues. There several methods analysis that deal with issues such as volume velocity streams. However, analysing the content streams extracting insights from dynamic is still a task. this paper, we propose pattern representation method which represents frames vectors by first applying...

In this study, a single-channel electroencephalography (EEG) analysis method has been proposed for automated 3-state-sleep classification to discriminate Awake, NREM (non-rapid eye movement) and REM (rapid movement). For purpose, singular spectrum (SSA) is applied automatically extract four brain rhythms: delta, theta, alpha, beta. These subbands are then used generate the appropriate features sleep using multi class support vector machine (M-SVM). The provided 0.79 agreement between manual...

The current Web and data indexing search mechanisms are mainly tailored to process text-based limited in addressing the intrinsic characteristics of distributed, large-scale dynamic Internet Things (IoT) networks. IoT demands novel solutions for create an ecosystem system; however, often numerical, multi-modal heterogeneous. We propose a distributed adaptable mechanism that allows discovery real-world Comparing state-of-the-art approaches, our model does not require any prior knowledge about...

The Internet of Things (IoT) offers an incredible innovation potential for developing smarter applications and services. However, today we see solutions in the development vertical services reflecting what used to be early days Web, leading fragmentation intra-nets Things. To achieve open IoT ecosystem systems platforms, several key enablers are needed effective, adaptive scalable mechanisms exploring discovering resources their data/capabilities. This paper discusses our work EU H2020...

Activity recognition using deep learning and sensor data can help monitor activities health conditions of people who need assistance in their daily lives. Deep Neural Network (DNN) models to infer the require collected by in-home sensory devices. These are often sent a centralised cloud be used for training model. Centralising introduces privacy risks. The contain sensitive information about subjects. cloud-based approach increases risk that stored reused other purposes without owner's...

In recent years, the development and deployment of Internet Things (IoT) devices has led to generation large volumes real world data. Analytical models can be used extract meaningful insights from this However, most IoT data is not fully utilised, which mainly due interoperability issues difficulties analyse collected by heterogeneous resources. To overcome heterogeneity, semantic technologies are create common share various originated sources. semantics add further overhead delivery,...

We describe a digital platform developed in collaboration with clinicians and user groups to provide remote healthcare monitoring support dementia care application. The uses data from sensory devices that are deployed participants' homes utilises set machine learning analytical algorithms identify risks of adverse health conditions such as Urinary Tract Infections (UTIs) hypertension people dementia. includes clinical interface is used by team view alerts notifications generated the also...

The advent of Internet Things, has resulted in the development infrastructure for capturing and storing data from domains ranging smart devices (e.g. smartphones) to cities. This is often available publicly enabled a wider range consumers utilise such sets applications scientific experimentation enhancing commercial activity businesses. Accordingly this need analysis tools that are both simple use provide most effective given set. To end, we introduce as web service, enables consumer make...

Background The move of cancer treatment into the outpatient setting can impact patient experience. Understanding how service delivery change impacts different people requires feedback to inform future development. use experience questionnaires often generates large amount free-text data that are difficult analyze. Objective aim this study was describe a proof-of-concept exploring experiences and perceptions undergoing treatment, using novel analysis techniques provide rapid analysis. Method...

A robust constrained complex singular spectrum analysis approach for the assessment of Parkinson's tremor by separation electromyograms (EMGs) is presented in this paper. This exploits expected EMG characteristics within a subspace single channel surface signal measured during prescribed hand movement including flexion and extension decomposed using analysis. The results are validated signals simultaneously recorded motion sensors.

Nocturnal disturbance is frequently observed in dementia and a major contributor to institutionalisation. Unobtrusive technology that can quantify sleep/wake determine bed occupancy during the nocturnal sleep episode may be beneficial for long-term clinical monitoring carer. Such technologies have, however, not been validated older people. Here we assessed performance of Withings Sleep Mattress (WSM) heterogenous population ensure external validity.Eighteen participants (65 - 80 years,...

The recent introduction of η-Hermitian matrices A = <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ηH</sup> has opened a new avenue research in quaternion signal processing. However, the exploitation this matrix structure been limited, perhaps due to lack joint diagonalisation methodologies these matrices. As such, we propose novel decompositions address shortcoming literature. an application, consider blind source separation problem form...

A novel quaternion-valued common spatial patterns (QCSP) algorithm is introduced to model co-channel coupling of multi-dimensional processes. To cater for the generality non-circular data, we propose a generalized QCSP (G-QCSP) which incorporates information on power difference between real and imaginary parts data channels. As an application, demonstrate how G-QCSP can be used provide high classification rates, even at signal-to-noise ratio (SNR) as low -10 dB. illustrate usefulness our...

Learning and adapting to new distributions or learning tasks sequentially without forgetting the previously learned knowledge is a challenging phenomenon in continual models. Most of conventional deep models are not capable one model ones. We address this issue by using Kalman Optimiser. The Optimiser divides neural network into two parts: long-term short-term memory units. unit used remember adapt task. have evaluated our method on MNIST, CIFAR10, CIFAR100 datasets compare results with...

Forecasting one step ahead is generally straightforward. two steps a little more challenging. further into the horizon may require prior forecasted samples, as availability of historical data not be adequate to do so. It in this motivational context that we proposed an eigen-based approach for complex-valued multiple-step forecasting. Here establish augmented complex-domain singular spectrum analysis framework perform prediction beyond 50 ahead. shown other algorithms such least mean square,...

In this work, we introduce a complex-valued singular spectrum analysis for the of electroencephalogram (EEG), which typically exhibits noncircular probability distribution. To exploit such prior knowledge, our technique makes use recent advances in statistics to power difference or correlation between data channels, contrast current methods cater only restrictive class circular data. particular, principal component analysis-like was employed detect onset P300, and tracked event-related...