Search

Electrocardiogram (ECG) is the most commonly observed signal for monitoring cardiovascular activities. Recent advances in wearable technology resulted in easy, comfortable and continuous acquisition of ECG signals in non-clinical settings. One of the most important concerns with wearable ECG sensors is noise sensitivity, which affects the precision and dependability of the recorded signals. Recently, deep learning (DL)-based ECG noise detection is widely used to limit the impact of noise in ECG signal and provides better signal quality. However, existing studies did not investigate the deployability of the noise detection and classification model in different devices (e.g., resourceful, semi-resourceful and resource-constraint) and its effect on the performance of the deployed model. This study investigates the impact of model conversions on performances that were done for deploying developed DL-based ECG classifiers in resourceful, semi-resourceful and resource-constraint IoT-enabled devices. That said, we propose a DL-based ECG noise classifier framework to provide noise-free ECG signals on smart devices. Our results show that the VGG model gains accuracy and F1 score 97.31% and 94.18% for resourceful devices, whereas the converted DL-based model for semi-resourceful and resource-constraint devices shows 92.84% and 84.59% respectively. At the same time, the Resnet model improves the 1.34% F1 score from resourceful DL-based models. In addition, Alexnet and VGG model is not fit for resource-constraint devices due to memory limitation. The study provides a guideline for designing DL-based models based on the target IoT deployment environment, which can be followed for problems other than ECG noise classification.

Deep-learning-based QRS-detection algorithms often require essential post-processing to refine the output prediction-stream for R-peak localisation. The post-processing involves basic signal-processing tasks including the removal of random noise in the model's prediction stream using a basic Salt and Pepper filter, as well as, tasks that use domain-specific thresholds, including a minimum QRS size, and a minimum or maximum R-R distance. These thresholds were found to vary among QRS-detection studies and empirically determined for the target dataset, which may have implications if the target dataset differs such as the drop of performance in unknown test datasets. Moreover, these studies, in general, fail to identify the relative strengths of deep-learning models and the post-processing to weigh them appropriately. This study identifies the domain-specific post-processing, as found in the QRS-detection literature, as three steps based on the required domain knowledge. It was found that the use of minimal domain-specific post-processing is often sufficient for most of the cases and the use of additional domain-specific refinement ensures superior performance, however, it makes the process biased towards the training data and lacks generalisability. As a remedy, a domain-agnostic automated post-processing is introduced where a separate recurrent neural network (RNN)-based model learns required post-processing from the output generated from a QRS-segmenting deep learning model, which is, to the best of our knowledge, the first of its kind. The RNN-based post-processing shows superiority over the domain-specific post-processing for most of the cases (with shallow variants of the QRS-segmenting model and datasets like TWADB) and lags behind for others but with a small margin ($\leq$2%). The consistency of the RNN-based post-processor is an important characteristic which can be utilised in designing a stable and domain agnostic QRS detector.

Automatic sleep stage classification is vital for evaluating the quality of sleep. Conventionally, sleep is monitored using multiple physiological sensors that are uncomfortable for long-term monitoring and require expert intervention. In this study, we propose an automatic technique for multi-stage sleep classification using photoplethysmographic (PPG) signal. We have proposed a convolutional neural network (CNN) that learns directly from the PPG signal and classifies multiple sleep stages. We developed models for two- (Wake-Sleep), three- (Wake-NREM-REM) and four- (Wake-Light sleep-Deep sleep-REM) stages of sleep classification. Our proposed approach shows an average classification accuracy of 94.4%, 94.2%, and 92.9% for two, three, and four stages, respectively. Experimental results show that the proposed CNN model outperforms existing state-of-the-art models (classical and deep learning) in the literature.

Interpretability often seeks domain-specific facts, which is understandable to human, from deep-learning (DL) or other machine-learning (ML) models of black-box nature. This is particularly important to establish transparency in ML model’s inner-working and decision-making, so that a certain level of trust is achieved when a model is deployed in a sensitive and mission-critical context, such as health-care. Model-level transparency can be achieved when its components are transparent and are capable of explaining reason of a decision, for a given input, which can be linked to domain-knowledge. This article used convolutional neural network (CNN), with sinc-convolution as its constrained first-layer, to explore if such a model’s decision-making can be explained, for a given task, by observing the sinc-convolution’s sinc-kernels. These kernels work like band-pass filters, having only two parameters per kernel - lower and upper cutoff frequencies, and optimised through back-propagation. The optimised frequency-bands of sinc-kernels may provide domain-specific insights for a given task. For a given input instance, the effects of sinc-kernels was visualised by means of explanation vector, which may help to identify comparatively significant frequency-bands, that may provide domain-specific interpretation, for the given task. In addition, a CNN model was further optimised by considering the identified subset of prominent sinc frequency-bands as the constrained first-layer, which yielded comparable or better performance, as compared to its all sinc-bands counterpart, as well as, a classical CNN. A minimal CNN structure, achieved through such an optimisation process, may help design task-specific interpretable models. To the best of our knowledge, the idea of sinc-convolution layer’s task-specific significant sinc-kernel-based network optimisation is the first of its kind. Additionally, the idea of explanation-vector-based joint time-frequency representation to analyse time-series signals is rare in the literature. The above concept was validated for two tasks, ECG beat-classification (five-class classification task), and R-peak localisation (sample-wise segmentation task).

A multilayer watermarking system is a system that is able to embed watermarks to a host media signal repeatedly in an overlaying manner, without incurring troubles in extracting the watermarks in each layer. In this paper, we present a novel patchwork-based audio watermarking algorithm that can embed and extract watermark bits successfully in such a multilayer framework. In the proposed method, a new watermark embedding algorithm is designed to ensure that the embedded watermarks in a certain layer do not affect the detection of watermarks in other layers. Adding multiple layers of watermark bits inevitably reduces the perceptual quality. However, to minimize the perceptual quality degradation in multilayer watermarking, the audio fragments for watermark embedding are selected from a set of specially arranged discrete cosine transform coefficients of the host audio signal. Watermark embedding is achieved by modifying the mean values of selected sample fragments. With the use of an embedding error buffer, the proposed system can withstand a wide range of common attacks. To maintain the balance between the perceptual quality and robustness, watermark embedding strength is adjusted according to the specific layer used. The proposed multilayer scheme ensures the independence of the processing in different layers. The effectiveness of the proposed system is demonstrated and verified by extensive simulation results.

AbstractRecent studies have demonstrated that Multi-Disciplinary Meetings (MDM) practiced in some medical contexts can contribute to positive health care outcomes. The group reasoning and decision-making in MDMs has been found to be most effective when deliberations revolve around the patient’s needs, comprehensive information is available during the meeting, core members attend and the MDM is effectively facilitated. This article presents a case study of the MDMs in cancer care in a region of Australia. The case study draws on a group reasoning model called the Reasoning Community model to analyse MDM deliberations to illustrate that many factors are important to support group reasoning, not solely the provision of pertinent information. The case study has implications for the use of data analytics in any group reasoning context.

Electrocardiogram (ECG) signal quality indices (SQIs) are essential for improving diagnostic accuracy and reliability of ECG analysis systems. In various practical applications, the ECG signals are corrupted by different types of noise. These corrupted ECG signals often provide insufficient and incorrect information regarding a patient’s health. To solve this problem, signal quality measurements should be made before an ECG signal is used for decision-making. This paper investigates the robustness of existing popular statistical signal quality indices (SSQIs): relative power of QRS complex (SQIp), skewness (SQIskew), signal-to-noise ratio (SQIsnr), higher order statistics SQI (SQIhos) and peakedness of kurtosis (SQIkur). We analysed the robustness of these SSQIs against different window sizes across diverse datasets. Results showed that the performance of SSQIs considerably fluctuates against varying datasets, whereas the impact of varying window sizes was minimal. This fluctuation occurred due to the use of a static threshold value for classifying noise-free ECG signals from the raw ECG signals. Another drawback of these SSQIs is the bias towards noise-free ECG signals, that limits their usefulness in clinical settings. In summary, the fixed threshold-based SSQIs cannot be used as a robust noise detection system. In order to solve this fixed threshold problem, other techniques can be developed using adaptive thresholds and machine-learning mechanisms.

Abstract: Phishing activity has recently been focused on social networking sites as a more effective way of exploiting not only the technology but also the trust that may exist between members in a social network. In this paper, a novel method for profiling phishing activity from an analysis of phishing emails is proposed. Profiling is useful in determining the activity of an individual or a particular group of phishers. Work in the area of phishing is usually aimed at detection of phishing emails. In this paper, we concentrate on profiling as distinct from detection of phishing emails. We formulate the profiling problem as a multi-label classification problem using the hyperlinks in the phishing emails as features and structural properties of emails along with whois (i.e. DNS) information on hyperlinks as profile classes. Further, we generate profiles based on the classifier predictions. Thus, classes become elements of profiles. We employ a boosting algorithm (AdaBoost) as well as SVM to generate multi-label class predictions on three different datasets created from hyperlink information in phishing emails. These predictions are further utilized to generate complete profiles of these emails. Results show that profiling can be done with quite high accuracy using hyperlink information.

The increasing demand for technology-enabled public sector services drives state agencies to launch information and communication technology (ICT) projects. The Australian and German state agencies are taking a proactive role towards technological change by employing so-called ICT change agents. These ICT change agents introduce, diffuse, manage and implement ICT within projects. Despite the mobilisation of change agents, there is scant research on the formal and informal roles of these key individuals within public sector projects. This article bridges that gap by providing valuable insights into the activities of public sector ICT change agents. It is based on empirical research from six case studies in Australian and German state agencies. Findings from these studies indicate that public sector ICT change agents position organisations to take advantage of cutting edge technologies by performing a great variety of formal and informal roles. Formal roles are performed in order to accomplish set formal project tasks, while informal roles help to speed up rapid ICT adoption and innovation through the change agents’ informal networks. The findings are delineated in a framework for future research which shows that formal and informal roles impact on the outcomes of public sector ICT projects.

We reduce the classification problem to solving a global optimization problem and a method based on a combination of the cutting angle method and a local search is applied to the solution of this problem. The proposed method allows to solve classification problems for databases with an arbitrary number of classes. Numerical experiments have been carried out with databases of small to medium size. We present their results and provide comparisons of these results with those obtained by 29 different classification algorithms. The best performance overall was achieved with the global optimization method.

Argumentation concepts have been applied to numerous knowledge engineering endeavours in recent years. For example, a variety of logics have been developed to represent argumentation in the context of a dialectical situation such as a dialogue. In contrast to the dialectical approach, argumentation has also been used to structure knowledge. This can be seen as a non-dialectical approach. The Toulmin argument structure has often been used to structure knowledge non-dialectically yet most studies that apply the Toulmin structure do not use the original structure but vary one or more components. Variations to the Toulmin structure can be understood as different ways to integrate a dialectical perspective with a non-dialectical one. Drawing the dialectical/non-dialectical distinction enables the specification of a framework called the generic actual argument model that is expressly non-dialectical. The framework enables the development of knowledge-based systems that integrate a variety of inference procedures, combine information retrieval with reasoning and facilitate automated document drafting. Furthermore, the non-dialectical framework provides the foundation for simple dialectical models. Systems based on our approach have been developed in family law, refugee law, determining eligibility for government legal aid, copyright law and e-tourism.

Cooperative telemedicine environments are required for many situations such as consultations between residents and senior doctors, case correlations, and for teaching and research purposes. The mode of collaboration may vary with different situations, in terms of the synchronisation of tasks, the sharing of data and the extent of collaboration among participants. It is essential for participants to be able to remotely view and manipulate visual data (images, two-dimensional and three-dimensional graphics, animation, and video) as well as interactively run application programs that involve visual data in real-time. However, this is not possible with current network bandwidth limitations when large amount of visual data are involved. In this article, we first provide an analysis of functional requirements by participants in cooperative diagnosis in different types of situations, before discussing technical requirements, which form the basis for our system architecture design. A new approach is also presented for efficient handling of programs, which involve visual data in real time. This is achieved via the construction and transmission of small messages that encapsulate the operations in a pipelined or hierarchical fashion.

Many medical services are often not available to people living in remote areas because of the lack of medical specialists. This problem would be alleviated if a suitable environment was designed to allow physicians to collaborate and exchange ideas with centrally located medical specialists. This article describes an ongoing research project to design and implement a collaborative multimedia environment to allow medical specialists to cooperate in diagnosis. The environment will support remote database access for medical images, the retrieval of relevant medical cases to support diagnosis, and communication among participants through telepointers and image annotation by free-hand drawing.

Heart rate variability (HRV) has been analysed using linear and nonlinear methods. In the framework of a controlled neonatal stress model, we applied tone–entropy (T–E) analysis at multiple lags to understand the influence of external stressors on healthy term neonates. Forty term neonates were included in the study. HRV was analysed using multi-lag T–E at two resting and two stress phases (heel stimulation and a heel stick blood drawing phase). Higher mean entropy values and lower mean tone values when stressed showed a reduction in randomness with increased sympathetic and reduced parasympathetic activity. A ROC analysis was used to estimate the diagnostic performances of tone and entropy and combining both features. Comparing the resting and simulation phase separately, the performance of tone outperformed entropy, but combining the two in a quadratic linear regression model, neonates in resting as compared to stress phases could be distinguished with high accuracy. This raises the possibility that when applied across short time segments, multi-lag T–E becomes an additional tool for more objective assessment of neonatal stress.