In the Internet of Medical Things (IoMT), securing patient data access is critical, but must be achieved without overwhelming the limited computational resources of edge devices. While cryptographic methods and access policies are widely applied to secure medical data, existing solutions often assume high computational capacity, centralized infrastructure, or predefined key structures, which are not ideal for the heterogeneous and resource-constrained environments found in IoMT. In addition to security, patient-centricity is becoming an essential design principle, where patients must have control over who accesses their data and under what conditions. Similarly, edge computing has emerged as a means to reduce latency, but edge devices are often semi-trusted, exposed to physical threats, and limited in processing power, making them unsuitable for heavyweight integrity verification or outsourced computation. Therefore, this paper presents a lightweight, patient-centric architecture that unifies attribute-based access control, dynamic edge data integrity verification, and consent-driven sharing in a secure and scalable architecture. The system minimizes communication and computational overhead by eliminating predefined keys, restricting edge computation, and guaranteeing verifiable data delivery. The experimental results demonstrate efficient handling of tampered and untampered cases, achieving fast, verifiable, and secure access with minimal resource consumption. This paper offers a practical and integrative solution to ensure security without sacrificing lightweight performance in real-world IoMT deployments.

Analyzing high-resolution data from smart meters (SMs) enables the identification of internal household activities and even personal user information, posing a significant privacy threat to energy consumers. Thus, preserving SM’s privacy is paramount. While data tampering methods offer a solution, they often lead to issues like inaccurate state estimation and complex billing. To address this, we propose demand-side energy management leveraging rechargeable batteries (RB) and solar panels. This is because larger RBs are costly and environmentally unfriendly. Thus, we prioritize the use of green energy sources, such as solar panels, minimizing RB’s reliance on enhanced consumer privacy. Strategically managing loads with solar panels and grid energy sales achieves cost-effective privacy. Considering realistic off-peak and peak energy consumption periods renders our approach practical. Theoretical analysis and simulations demonstrate the superiority of our method over existing state-of-the-art approaches.

Machine learning (ML) has become integral to cybersecurity applications, e.g., phishing detection, intrusion detection systems, malware analysis, and botnet identification. However, the integration of ML also exposes novel attack surfaces that can be exploited through adversarial machine learning (AML). While prior surveys have examined individual threats or defenses, they often focus narrowly on specific stages, e.g., training or testing. In contrast, in this paper, we provide the first comprehensive survey of adversarial attacks and defenses across the entire ML development life cycle within the cybersecurity domain. Using a structured methodology, we categorize vulnerabilities and countermeasures at each stage, data gathering, model training, testing, deployment, and maintenance, highlighting cross-stage interactions and emerging distributed threat models. Our study addresses key gaps in current defenses, including their limited generalizability and lack of standardized evaluation practices, and identifies promising directions, e.g., lifecycle-aware robustness, distributed resilience, and the integration of statistical with generative methods. Consolidating fragmented research into an end-to-end perspective, this study advances the understanding of AML in cybersecurity and outlines a roadmap for building more trustworthy, and resilient ML-driven security systems.

Generative artificial intelligence (AI), an advancing frontier, uses machine learning to autonomously create media content, though it faces challenges with bias and fair representation. This research investigates demographic bias within the LAION dataset, specifically focusing on the representation of Indigenous Australians. Large image generative models are typically trained on extensive datasets scraped from the Internet, which often reflect the demographics of the most active online communities rather than accurately representing local populations. While existing studies have examined hate speech, gender balance, and broad ethnic diversity within LAION, limited research addresses representation relative to specific national demographics or minority subgroups. Auditing the dataset with state-of-the-art facial recognition and sentiment analysis models, we assessed the age, gender, and ethnicity of images in LAION and compared these distributions against census data for Australia and the United States. We also conducted a focused analysis of Indigenous Australian representation by identifying relevant images using keyword searches and applying the aforementioned models. Our findings reveal that the dataset demographic composition poorly aligns with the actual population of Australia, with regional subgroups under-represented, potentially leading to inaccurate portrayals of Indigenous Australians. These results underscore the need for either strengthened safeguards on globally developed generative models or the development of locally trained models to ensure responsible and inclusive cultural representation in AI-generated imagery.

The Internet of Medical Things (IoMT) systems enable the continuous monitoring and collection of healthcare data from various medical devices and sensors, facilitating real-time analysis and timely interventions. One such example of the IoMT system is the early and accurate detection of arrhythmia using electrocardiogram (ECG) signals, which plays a crucial role in improving patient health. However, healthcare data contains sensitive information that raises privacy concerns for users. In recent years, Federated Learning (FL) offers a promising solution by enabling collaborative model training on distributed ECG data at the device level while preserving data privacy. However, FL is computationally expensive and suffers from data heterogeneity, which slower convergence, reduces performance, and hinders generalization across diverse client datasets. In this work, we propose a lightweight FL-based model designed explicitly for arrhythmia detection with data heterogeneity. We evaluate our model's performance on two publicly available ECG datasets (PTBD and MIT-BIH arrhythmia). The proposed model achieves high accuracy (between 0.95 and 0.98) while maintaining robustness against heterogeneous data distributions. Furthermore, the experimental results demonstrate significant efficiency gains compared to a baseline model (ResNet). Our proposed lightweight FL-based model requires substantially less mega floating point operations (MFLOPS) (0.07 vs. 2.14 for ResNet) and communication cost (1000 Mb vs. 7500 Mb for ResNet) to achieve convergence. These results indicate the potential of our proposed approach for practical and privacy-preserving arrhythmia detection in resource-constrained IoMT settings.

We propose a blockchain-enabled lightweight federated learning (BlockFL-Med) framework tailored for smart medical spaces, e.g., the Internet of Medical Things (IoMT), addressing key challenges, e.g., privacy preservation, trust management, and scalability. The framework ensures the privacy of sensitive patient data by employing federated learning, where only model updates are shared instead of raw data. To enhance trust, the framework integrates blockchain technology, creating a decentralized and tamper-proof network that verifies client contributions and mitigates risks from malicious participants. Experimental results demonstrate the scalability and efficiency issues by optimizing communication costs, e.g., transmitting lightweight kilobyte-sized model updates instead of larger megabyte-sized models, making it well-suited for heterogeneous and resource-constrained IoMT environments.

The Internet of Medical Things (IoMT) represents a significant technological advancement with exceptional capabilities across various domains, particularly in healthcare. IoMT integrates medical devices, software applications, and healthcare systems, enabling seamless communication and data exchange over the Internet. As deep learning (DL) continues to evolve, applications within IoMT are increasingly dominant. However, these DL applications face new reliability challenges, particularly due to the security threat posed by adversarial attacks. These attacks introduce subtle and often imperceptible perturbations that can lead to significantly erroneous predictions by classifiers. To address these reliability concerns, we propose a novel security mechanism using an attack detector specifically designed to counter adversarial attacks within IoMT environments. This approach leverages a transformer model to enhance resistance against such attacks. We validate our method through experiments using datasets for skin cancer, retina damage, and chest X-rays, testing against both white-box attacks (e.g., Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD)) and black-box attacks (e.g., Additive Gaussian Noise (AGN) and Additive Uniform Noise (AUN)). Our proposed attack detector exhibited F1 and accuracy 0.91 and 0.94. Following the successful application of our attack detector, the disease classification model achieved an average F1 and accuracy of 0.97 and 0.98 compared to the attack model performance (F1 and accuracy of 0.64 and 0.60, respectively) across the three datasets.

Adversarial attacks pose a significant threat to the integrity and reliability of electrocardiogram (ECG) signals, compromising their use in critical applications, e.g., arrhythmia detection and classification. In this paper, we propose an attack-data-independent defence mechanism to effectively mitigate adversarial attacks on ECG signals. Unlike existing defence mechanisms that rely on learning from adversarial samples, our proposed approach operates as a ‘gatekeeper,’ selectively discarding noisy and attack signals while allowing only clean and non-attack ECG signals to be stored in the data layer. This ensures the availability of reliable and high-quality ECG data for subsequent analysis. The proposed defence mechanism not only detects and filters out the attack and noisy ECG signals but also provides robust protection against adversarial attacks, enhancing the integrity and trustworthiness of ECG data for critical applications. To evaluate the effectiveness of our proposal, we conduct experiments using physiologic and synthetic ECG datasets against two well-known attacks: a white-box attack (Fast Gradient Signed Method (FGSM) and Projected Gradient Descent (PGD)) and a black-box attack (HopSkipJump and Boundary). Our experimental results demonstrate the superiority and effectiveness of our approach in defending against adversarial attacks on ECG signals, making it a promising solution for ensuring the security and reliability of ECG-based diagnosis in smart healthcare applications.

Fog computing plays a prominent role in offloading computational tasks in heterogeneous environments since it provides less service delay than traditional cloud computing. The Internet of Things (IoT) devices cannot handle complex tasks due to less battery power, storage and computational capability. Full offloading has issues in providing efficient computation delay due to more response time and transmission cost. A suitable solution to overcome this problem is to partition the tasks into splittable subtasks. Considering multi-criteria decision parameters like processing efficiency and deadline helps to achieve efficient resource allocation and task assignment. The matching theory is applied to map task nodes to heterogeneous fog nodes and VMs for stability. Compared to baseline algorithms, proposed algorithms like Resource Allocation based on Processing Efficiency (RABP) and Task Assignment Based on Completion Time (TAC) are efficient enough to provide reasonable service delay and discard the non-beneficial tasks, i.e., tasks that do not execute within the deadline.

Given the escalating diversity, sophistication, and frequency of cyber attacks, it is imperative for critical infrastructure entities, e.g. smart grids, to recognize the inherent risks of operating in isolation. Sharing cyber threat intelligence (CTI) helps them stand together and build a collective cyber defense by knowledge, skills, and experience encompassing information related to identifying and evaluating cyber and physical threats. The present studies lack on robust CTI sharing strategies in smart grid systems. To address the critical need for secure and effective CTI sharing in smart grid systems, this article proposes a novel approach. Our solution leverages encrypted federated learning (FL) with integrated malicious client detection mechanisms. This approach facilitates collaborative learning of a threat detection model while preserving the privacy of raw CTI data. Employing real-world, heterogeneous smart grid datasets, we rigorously evaluated our approach under two distinct attack scenarios. The results demonstrate resilience against both man-in-the-middle attacks and malicious clients, exceeding the performance typically observed in traditional FL models.

Major depressive disorder (MDD) is a chronic mental illness which affects people's well-being and is often detected at a later stage of depression with a likelihood of suicidal ideation. Early detection of MDD is thus necessary to reduce the impact, however, it requires monitoring vitals in daily living conditions. EEG is generally multi-channel and due to difficulty in signal acquisition, it is unsuitable for home-based monitoring, whereas, wearable sensors can collect single-channel ECG. Classical machine-learning based MDD detection studies commonly use various heart rate variability features. Feature generation, which requires domain knowledge, is often challenging, and requires computation power, often unsuitable for real time processing, MDDBranchNet is a proposed parallel-branch deep learning model for MDD binary classification from a single channel ECG which uses additional ECG-derived signals such as R-R signal and degree distribution time series of horizontal visibility graph. The use of derived branches was able to increase the model's accuracy by around 7%. An optimal 20-second overlapped segmentation of ECG recording was found to be beneficial with a 70% prediction threshold for maximum MDD detection with a minimum false positive rate. The proposed model evaluated MDD prediction from signal excerpts, irrespective of location (first, middle or last one-third of the recording), instead of considering the entire ECG signal with minimal performance variation stressing the idea that MDD phenomena are likely to manifest uniformly throughout the recording.

Software exploitation is an ever-growing problem. Signature-based exploitation detection techniques have not been effective as malicious actors continuously develop circumvention techniques. Current ML-based (signature-less) exploitation detection research is limited in quantity and use cases. Key to the success of any ML model is the characteristics used to depict program behaviour (i.e., features). Current work on using ML for software exploitation is focused on novelty ML algorithms while neglecting program characterization and under-reporting the approach for data preparation. There are two main competing program characterization techniques, micro-architecture independent (MAI) and micro-architecture dependent (MAD) techniques. This study evaluates MAI program characterization techniques for use with ML-based exploitation detection. A publicly available runtime-based traces of 11 Windows applications under buffer-overflow exploitation is used to replicate the feature engineering work found in research that uses MAI for ML-based exploitation detection. The performance and feature importance are evaluated with two different ensemble ML models (Random Forests and XGBoost). The results demonstrate that, although 0% FPR has been achieved in all datasets, MAI features that are purely fine-grained in nature can achieve a maximum recall value of 100% and an average recall of 40%, respectively. While features that contain a higher coarse-grained to fine-grained features ratio can achieve a maximum recall of 100% with an average value of 62%. The study provides a detailed discussion of the feature importance and reveals that the most important features relate to memory traffic characteristics.

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).

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 ( ≤ 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.

Hand gesture recognition is the process of detecting the hand movements via sensor measurements for detecting an activity, such as writing a letter or a number. Recognising the handwritten characters using wearable devices enables machine-human interaction to occur without the need for a communication method. An intelligent automated framework is required to accurately detect the handwritten characters using wrist worn sensor signals, in particular, with minimal number of sensors. Moreover, the system developed needs to have the capacity to recognise the characters written in different sizes. In order to address these, we analyse performance of several machine learning models using single/multiple sensors namely, accelerometer or/and gyroscope, for recognising hand gesture characters including alphabet and numbers of varying sizes. We formulate a set of features that enable robust and accurate detection of the characters.We performed novel data collection using an off-the-shelf wrist-worn sensor based device, and evaluated our framework to detect the different characters effectively. The maximum accuracy (90.40%) was achieved using both sensors and Random Forest (RF) model. This was dropped to 82.51% for the same model using accelerometer sensor alone. Using the gyroscope sensor, an overall average accuracy of 80.16% was achieved with the Forward Neural Network (FNN) model. Although the model based on both sensors showed the best performance, our evaluation reveals that it is feasible to develop a machine learning model using single sensor to detect hand gesture characters of varying sizes with reasonable (≥ 80%) accuracy.

The evolution of ubiquitous sensors has led to the generation of copious amounts of waveform data. Human motion waveform analysis has found significance in clinical and home-based activity monitoring. Exploration of cluster structure in such waveform data prior to developing learning models is an important pattern recognition problem. A prominent category of algorithms in this direction, known as Visual Assessment of (cluster) Tendency (VAT), employs visual approaches to study cluster evolution through heat maps. This paper proposes shape-iVAT, a new relative of an improved VAT model, that captures local time-series characteristics through representative subsequences, known as shapelets, to identify interesting patterns in motion data. We propose an unsupervised method for shapelet extraction using maximin shape sampling and shape-based distance computation for selecting key shapelets representing characteristic motion patterns. These shapelets are used to transform waveform data into a dissimilarity matrix for VAT evaluation. We demonstrate that the proposed method outperforms standard VAT with global distance measures for identifying complex upper limb motion captured using a camera-based motion sensing device. We also show that our method has significance in efficient and interpretable cluster tendency assessment for anomaly detection and continuous motion monitoring.

With the development of information technology, thousands of devices are connected to the Internet, various types of data are accessed and transmitted through the network, which pose huge security threats while bringing convenience to people. In order to deal with security issues, many effective solutions have been given based on traditional machine learning. However, due to the characteristics of big data in cyber security, there exists a bottleneck for methods of traditional machine learning in improving security. Owning to the advantages of processing big data and high-dimensional data, new solutions for cyber security are provided based on deep learning. In this paper, the applications of deep learning are classified, analyzed and summarized in the field of cyber security, and the applications are compared between deep learning and traditional machine learning in the security field. The challenges and problems faced by deep learning in cyber security are analyzed and presented. The findings illustrate that deep learning has a better effect on some aspects of cyber security and should be considered as the first option.

Stroke survivors are often characterized by hemiparesis, i.e., paralysis in one half of the body, severely affecting upper limb movements. Monitoring the progression of hemiparesis requires manual observation of limb movements at regular intervals, and hence is a labour intensive process. In this work, we use wrist-worn accelerometers for automated assessment of hemiparesis in acute stroke. We propose novel measures of similarity and asymmetry in hand activities through bivariate Poincaré analysis between two-hand accelerometer data for quantifying hemiparetic severity. The proposed descriptors characterize the distribution of activity surrogates derived from acceleration of the two hands, on a 2D bivariate Poincaré Plot. Experiments show that while the descriptors CSD1 and CSD2 can identify hemiparetic patients from control subjects, their normalized difference CSDR and the descriptors Complex Cross-Correlation Measure ( C3M) and Activity Asymmetry Index ( AAI) can distinguish between mild, moderate and severe hemiparesis. These measures are compared with traditional measures of cross-correlation and evaluated against the National Institutes of Health Stroke Scale (NIHSS), the clinical gold standard for hemiparetic severity estimation. This study, undertaken on 40 acute stroke patients with varying levels of hemiparesis and 15 healthy controls, validates the use of short length ( 5 minutes) wearable accelerometry data for identifying hemiparesis with greater clinical sensitivity. Results show that the proposed descriptors with a hierarchical classification model outperform state-of-the-art methods with overall accuracy of 0.78 and 0.85 for 4-class and 3-class hemiparesis identification respectively.

Breathing rate (BR) is a significant bio marker that provides both prognostic and diagnostic information for monitoring physiological condition. In addition to vital bio markers, such as blood oxygen saturation and pulse rate, BR can be extracted from non-invasive and wearable pulse oximeter based photoplethysmogram (PPG). Empirical mode decomposition (EMD) and its noise-assisted variants are widely used for decomposing non-linear and non-stationary signals. In this work, the effect of all variants of EMD in extracting BR from PPG has been investigated. We have used an EMD family PCA based hybrid model in extracting BR from PPG, which is a natural extension of our previously developed ensemble EMD (EEMD) PCA hybrid model. The performance of each model has been tested using two different datasets: MIMIC and Capnobase. Median absolute error varied from 0 to 5.03 and from 2.47 to 10.55 breaths/min for MIMIC and Capnobase dataset, respectively. Among all the EMD variants, EEMD-PCA and improved complete EEMD with adaptive noise (ICEEMDAN) PCA hybrid model present better performance for both datasets. This is the first study to compare EMD variants performance for decomposing real world signal and determine that from current methods, ICEEMDAN and EEMD are optimal for estimating BR from PPG.

The photoplethysmographic (PPG) signal measures the local variations of blood volume in tissues, reflecting the peripheral pulse modulated by cardiac activity, respiration, and other physiological effects. Therefore, PPG can be used to extract the vital cardiorespiratory signals like heart rate (HR), and respiratory rate (RR) and this will reduce the number of sensors connected to the patient's body for recording these vital signs. In this paper, we propose an algorithm based on ensemble empirical mode decomposition with principal component analysis (EEMD-PCA) as a novel approach to estimate HR and RR simultaneously from PPG signal. To examine the performance of the proposed algorithm, we used 310 (from 35 subjects) and 632 (from 42 subjects) epochs of simultaneously recorded electrocardiogram, PPG, and respiratory signal extracted from MIMIC (Physionet ATM data bank) and Capnobase database, respectively. Results of EEMD-PCA-based extraction of HR and RR from PPG signal showed that the median RMS error (1st and 3rd quartiles) obtained in MIMIC data set for RR was 0.89 (0, 1.78) breaths/min, for HR was 0.57 (0.30, 0.71) beats/min and in Capnobase data set it was 2.77 (0.50, 5.9) breaths/min and 0.69 (0.54, 1.10) beats/min for RR and HR, respectively. These results illustrated that the proposed EEMD-PCA approach is more accurate in estimating HR and RR than other existing methods. Efficient and reliable extraction of HR and RR from the pulse oximeter's PPG signal will help patients for monitoring HR and RR with low cost and less discomfort.

In this study, a linear parametric modeling technique was applied to model ventricular repolarization (VR) dynamics. Three features were selected from the surface ECG recordings to investigate the changes in VR dynamics in healthy and cardiac autonomic neuropathy (CAN) participants with diabetes including heart rate variability (calculated from RR intervals), repolarization variability (calculated from QT intervals), and respiration [calculated by ECG-derived respiration (EDR)]. Surface ECGs were recorded in a supine resting position from 80 age-matched participants (40 with no cardiac autonomic neuropathy (NCAN) and 40 with CAN). In the CAN group, 25 participants had early/subclinical CAN (ECAN) and 15 participants were identified with definite/clinical CAN (DCAN). Detecting subclinical CAN is crucial for designing an effective treatment plan to prevent further cardiovascular complications. For CAN diagnosis, VR dynamics was analyzed using linear parametric autoregressive bivariate (ARXAR) and trivariate (ARXXAR) models, which were estimated using 250 beats of derived QT, RR, and EDR time series extracted from the first 5 min of the recorded ECG signal. Results showed that the EDR-based models gave a significantly higher fitting value (p < 0.0001) than models without EDR, which indicates that QT-RR dynamics is better explained by respiratory-information-based models. Moreover, the QT-RR-EDR model fitting values gradually decreased from the NCAN group to ECAN and DCAN groups, which indicate a decoupling of QT from RR and the respiration signal with the increase in severity of CAN. In this study, only the EDR-based model significantly distinguished ECAN and DCAN groups from the NCAN group (p < 0.05) with large effect sizes (Cohen's d > 0.75) showing the effectiveness of this modeling technique in detecting subclinical CAN. In conclusion, the EDR-based trivariate QT-RR-EDR model was found to be better in detecting the presence and severity of CAN than the bivariate QT-RR model. This finding also establishes the importance of adding respiratory information for analyzing the gradual deterioration of normal VR dynamics in pathological conditions, such as diabetic CAN.

Angiotensin converting enzyme (ACE) polymorphism has been shown to be important in hypertension progression and also in diabetes complications, especially associated with heart disease. Heart rate variability (HRV) is an established measure for classification of autonomic function regulating heart rate, based on the interbeat interval time series derived from a raw ECG recording. Results of this paper show that the length (number of interbeat intervals) and preprocessing of the tachogram affect the HRV analysis outcome. The comparison was based on tachogram lengths of 250, 300, 350, and 400 RR-intervals and five preprocessing approaches. An automated adaptive preprocessing method for the heart rate biosignal and tachogram length of 400 interbeat intervals provided the best classification. HRV results differed for the Type 2 Diabetes Mellitus (T2DM) group between the I/I genotype and the I/D and D/D genotypes, whereas for controls there was no significant difference in HRV between genotypes. Selecting an appropriate length of recording and automated preprocessing has confirmed that there is an effect of ACE polymorphism including the I/I genotype and that I/I should not be combined with I/D genotype in determining the extent of autonomic modulation of the heart rate.

Respiratory events during sleep induce cortical arousals and manifest changes in autonomic markers in sleep disorder breathing (SDB). Finger photoplethysmography (PPG) has been shown to be a reliable method of determining sympathetic activation. We hypothesize that changes in PPG signals are sufficient to predict the occurrence of respiratory-event-related cortical arousal. In this study, we develop a respiratory arousal detection model in SDB subjects by using PPG features. PPG signals from 10 SDB subjects (9 male, 1 female) with age range 43-75 years were used in this study. Time domain features of PPG signals, such as 1) PWA--pulse wave amplitude, 2) PPI--peak-to-peak interval, and 3) Area--area under peak, were used to detect arousal events. In this study, PWA and Area have shown better performance (higher accuracy and lower false rate) compared to PPI features. After investigating possible groupings of these features, combination of PWA and Area (PWA + Area) was shown to provide better accuracy with a lower false detection rate in arousal detection. PPG-based arousal indexes agreed well across a wide range of decision thresholds, resulting in a receiver operating characteristic with an area under the curve of 0.91. For the decision threshold (PC(thresh) = 25%) chosen for the final analyses, a sensitivity of 68.1% and a specificity of 95.2% were obtained. The results showed an accuracy of 84.68%, 85.15%, 86.93%, and 50.79% with a false rate of 21.80%, 55.41%, 64.78%, and 50.79% at PC(thresh) = 25% or PPI, PWA, Area , and PWA + Area features, respectively. This indicates that combining PWA and Area features reduced the false positive rate without much affecting the sensitivity of the arousal detection system. In conclusion, the PPG-based respiratory arousal detection model is a simple and promising alternative to the conventional electroencephalogram (EEG)-based respiratory arousal detection system.

Obstructive sleep apnea syndrome (OSAS) is associated with cardiovascular morbidity as well as excessive daytime sleepiness and poor quality of life. In this study, we apply a machine learning technique [support vector machines (SVMs)] for automated recognition of OSAS types from their nocturnal ECG recordings. A total of 125 sets of nocturnal ECG recordings acquired from normal subjects (OSAS - ) and subjects with OSAS (OSAS +), each of approximately 8 h in duration, were analyzed. Features extracted from successive wavelet coefficient levels after wavelet decomposition of signals due to heart rate variability (HRV) from RR intervals and ECG-derived respiration (EDR) from R waves of QRS amplitudes were used as inputs to the SVMs to recognize OSAS +/- subjects. Using leave-one-out technique, the maximum accuracy of classification for 83 training sets was found to be 100% for SVMs using a subset of selected combination of HRV and EDR features. Independent test results on 42 subjects showed that it correctly recognized 24 out of 26 OSAS + subjects and 15 out of 16 OSAS - subjects (accuracy = 92.85%; Cohen's kappa value of 0.85). For estimating the relative severity of OSAS, the posterior probabilities of SVM outputs were calculated and compared with respective apnea/hypopnea index. These results suggest superior performance of SVMs in OSAS recognition supported by wavelet-based features of ECG. The results demonstrate considerable potential in applying SVMs in an ECG-based screening device that can aid a sleep specialist in the initial assessment of patients with suspected OSAS.