Type 2 diabetes (T2D) is a common metabolic disorder that poses threat to human health. Blood glucose (BG) concentrations in T2D subjects can be influenced by various factors, such as stress, physical activity, and meal consumption. Obtaining personalized mathematical models for people with T2D can be beneficial for developing effective T2D management strategies, such as insulin dosing algorithms. However, fitting these models for T2D subjects can be challenging due to the limited data typically available during treatment, consisting of only Continuous Glucose Monitoring (CGM) readings and injected insulin amounts. To address this issue, we propose a stochastic jump diffusion model that incorporates the uncertainties from meal consumption behavior together with other disturbances. Additionally, we provide an inference strategy that enables us to estimate both physiological (e.g., insulin sensitivity) and behavioral (e.g., average number of meals per day) parameters using only CGM data and injected insulin amounts. We validate the proposed stochastic model and inference strategy with synthetic and clinical data, assessing their ability to estimate various parameters of the model and fit CGM data. The results demonstrate that our proposed method, along with the model, manages to estimates parameters with different interquartile range (IQR) values. Moreover, the model, in combination with the method, provide a good fit for CGM data, and demonstrates the ability to obtain estimates of unannounced meal times.

In the first part, uncompensated and unstable polynomials from 2nd order to 10th order are mentioned and it was also concluded that control loops are unstable, the change in coefficients could be seen in the list. In this part, rather taking such unpractical or impractical circuit (or novel B circuit or novel B-topology), a practical ‘buck-boost’ known circuit and its stability has been explored. The known working circuit is assumed as a black box and is declared as  equipment under test (EUT) with just input and output and the control loop stability would be investigated using compensators. The author examines the stability and analyses how the abyss of coefficients change with respect to different compensators when the MOSFET switching frequency is changed from 50 Hz to  1.999 MHz or 1999 kHz. Due to large number,this paper studies the abyss of stability and unstability, the game of coefficients, upto third order polynomials. The Bode plots of the compensators designed with P, I, PID, series PI, parallel PI, One pole one zero (1P1Z), multipoles, two pole two zero (2P2Z) and three pole three zero (3P3Z) are submitted. The known theory is also explained as summary. This paper is also the transcript of the proposed invited talk as speaker in some conference.

In this paper, we develop a hierarchical aerial computing framework composed of high altitude platform (HAP) and unmanned aerial vehicles (UAVs) to compute the fully offloaded tasks of terrestrial mobile users which are connected through an uplink non-orthogonal multiple access (UL-NOMA). To better assess the freshness of information in computationintensive applications the criterion of age of information (AoI) is considered. In particular, the problem is formulated to minimize the average AoI of users with elastic tasks, by adjusting UAVs’ trajectory and resource allocation on both UAVs and HAP, which is restricted by the channel state information (CSI) uncertainty and multiple resource constraints of UAVs and HAP. In order to solve this non-convex optimization problem, two methods of multi-agent deep deterministic policy gradient (MADDPG) and federated reinforcement learning (FRL) are proposed to design the UAVs’ trajectory, and obtain channel, power, and CPU allocations. It is shown that task scheduling significantly reduces the average AoI. This improvement is more pronounced for larger task sizes. On one hand, it is shown that power allocation has a marginal effect on the average AoI compared to using full transmission power for all users. Compared with traditional transmission schemes, the simulation results show our scheduling scheme results in a substantial improvement in average AoI.

To provide various intelligent services in Internet of Vehicles (IoVs), such as autonomous driving, data sharing technologies enable vehicles to overcome information barriers and provide a big data foundation. Federated Learning (FL), which shares models instead of raw data, has emerged as a popular solution to address privacy concerns. However, current approaches are limited scalability and security, which are not suitable for the dynamic network topology of IoV scenarios. In this letter, we propose a Multi-Region Asynchronous Swarm Learning (MASL) framework in IoVs, which is empowered by the hierarchical blockchain and executed parallel between multiple regions. The MASL integrates identity verification and asynchronous model training while ensuring secure aggregation as well as data privacy. Through intra-regional and cross-regional sharing, the security and efficiency during large-scale data sharing in IoVs are effectively improved while alleviating the Non-IID data problem. Finally, both the simulation and hardware testbed results demonstrate that the proposed MASL framework could achieve better performances in terms of efficiency and security compared with the existing algorithms.

A document by Zixuan Wang . Click on the document to view its contents.

Federated learning (FL) emerges as a distributed training method in the Internet of Vehicles (IoVs), allowing connected and automated vehicles (CAVs) to train a global model by exchanging models instead of raw data, protecting data privacy. Due to the limitation of model accuracy and communication overhead in FL, in this paper, we propose a directed acyclic graph (DAG) blockchain-based IoV that comprises a DAG layer and a CAV layer for model sharing and training, respectively. Furthermore, a DAG blockchain-based asynchronous federated mutual learning (DAFML) algorithm is introduced to improve the model performance, which utilizes a student-teacher model in the local training and a mutual distillation method that enables models to be partially updated using mutual transferred knowledge. Moreover, to further improve the model accuracy, the personalized weight is designed to adjust the mutual distillation in the model updating. Simulation results demonstrate that the proposed DAFML algorithm outperforms other benchmarks in terms of the model accuracy and distillation ratio.

REDCap, a widely used electronic data capture tool for clinical and translational research, receives crucial support from various committees and groups within the extensive network of institutional collaborators known as the REDCap Consortium. For example, the Regulatory and Software Validation Committee defines and releases manual validation test scripts intended to aid institutions in the testing and validation efforts of their institutional REDCap deployments. Although these resources are important and valuable, the manual testing process itself has certain limitations such as time-consuming test execution and potential for inconsistent database state. To address some of the limitations of manual testing, a group of software developers within the Regulatory and Software Validation Committee wrote and released automated feature test scripts using an automated testing framework called the REDCap Cypress Test Framework. This report provides an overview of the REDCap Cypress Test Framework and its potential to improve the testing and validation of REDCap through automated testing.

We investigate an economic readout for depth-of-interaction (DOI) and time-of-flight capable PET detector that consists of an N×N array of crystals whose light outputs at the front- and back-end surfaces are detected by using silicon photomultipliers (SiPM). The N×N SiPMs at the front-end (back-end) of the detector are read by a stripline configured to support discrimination of the row (column) position of the signal-producing crystal, producing only four outputs for the detector. To evaluate this design, we built 4×4 and 8×8 detector modules (DM) by using 3.0×3.0×20 mm3 lutetium-yttrium oxyorthosilicates. The outputs were sampled at 2 GHz and processed offline. For both DMs, crystal discrimination was successful. For the 4×4 (8×8) DM, we obtained a crystal-level energy resolution ranging from 11.3% to 19.3% with an average of 14.1% (9.5% to 21.6% with an average of 13.9%), an average DOI resolution of 2.5 mm (3.1 mm), and a best coincidence resolving time (CRT), measured in coincidence with a single-pixel reference detector with a 302 ps CRT, of 391 ps (603 ps). The CRT deteriorated with the (8×8) DM possibly due to intercrystal scattering.

Technique for improved USBL position accuracy, but using a fundamental trigonometric relation to allow higher accuraccy bearings to be obtained from a given operating frequency sent by the USBL beacon.

The rapid growth of renewable generation, driven by increasing concerns about climate change, sustainable energy sources, and energy independence, has presented significant challenges for distribution system operators (DSOs). Integrating intermittent sources while ensuring grid stability and reliability demands a robust evaluation of hosting capacity (HC) in power distribution networks. However, prevailing HC analysis methodologies predominantly rely on conservative assumptions or worst-case scenarios, often leading to impractical and unreliable outcomes. Commercial tools, though offer readily available solutions for hosting capacity analysis, suffer from limitations as they typically assess HC of individual nodes independently or focus on only a few distributed scenarios. More elaborate methods have been proposed, attempting to cope with the complexity and stochastic nature of the HC problem. However, no much consideration has been given to their scalability to large distribution systems. In response to the complexity and stochastic nature of the HC problem, this paper introduces an efficient methodology based on stochastic power system simulations and statistical analysis. The proposed approach undergoes comprehensive assessments to substantiate its efficacy, with a key focus on validating its accuracy for reliable HC estimates. Utilizing the bootstrap method, a resampling technique, multiple samples are generated to mimic possible Distributed Energy Resources (DERs) deployments, enabling estimation of confidence intervals for HC metrics. By applying statistical analysis to power system simulation results, insights into the expected variability and uncertainty of the problem are gained. These insights guide the selection of the minimum number of DER deployment scenarios necessary to ensure an efficient and well-informed HC assessment process. The proposed methodology effectively addresses the challenges of HC analysis, offering scalability to large distribution networks. Its efficiency and enhanced accuracy make it a valuable tool for DSOs in facilitating the integration of renewable generation in a dependable and practical manner.

This research paper explores the integration of AI technology and Brain-Based Learning Theories in education, aiming to address a significant research gap and highlight their potential impact on cognitive development. While both AI technology and Brain-Based Learning Theories have been extensively studied and applied independently in the field of education, there is a lack of comprehensive research exploring their combined potential. This study delves into the theoretical foundations and guiding principles of brain-based learning, investigating its application in personalizing and enhancing learning experiences. It evaluates the current state of AI technology in education and examines how AI-powered brain-based learning techniques can improve student engagement, knowledge retention, and critical thinking skills. Ethical considerations and challenges associated with integrating AI into brain-based learning methodologies are acknowledged, and practical guidelines are provided for educators and decision-makers to effectively leverage AI in implementation. The research also examines the long-term effects of AI-enabled brain-based learning on educational systems, workforce readiness, and lifelong learning opportunities. Drawing on case studies and best practices from successful academic institutions, valuable insights are presented regarding the synergistic relationship between brain-based learning and AI. The paper proposes strategies for scaling and implementing AI-based brain-based learning approaches across diverse educational settings, with the aim of driving future innovations and advancements in education. Ultimately, this research sheds light on the transformative potential of AI-enabled brain-based learning, opening new avenues for educational improvement and advancement.

The probing depth of transient electromagnetic method (TEM) refers to the depth  range at which changes in underground conductivity can be effectively detected. It typically  ranges from tens of meters to several kilometers and is influenced by factors such as instrument  parameters and the conductivity of the subsurface structure. Rapid and accurate calculating the  probing depth is beneficial for determining the feasibility of exploration engineering, setting appropriate inversion parameters and improving exploration accuracy. However, mainstream  methods suffer from issues such as low computational precision, large uncertainties, or high  computational requirements, making them unsuitable for processing massive airborne  electromagnetic data. In this study, we propose a prediction model based on deep learning that  can directly compute the probing depth from the TEM responses, and its effectiveness and  accuracy is validated through synthetic models and field measurements. Furthermore, we apply  this algorithm to deep learning-based ATEM inversion by constraining the one-dimensional  resistivity models in the training set above the probing depth, to reduce the non-uniqueness of  the inversion, accelerate the convergence, and improve its prediction accuracy.

How to correct quantum errors without encoding and parity checks? what forms of real-life quantum noise can actually be reversed on the run-time? and how can we engineer more noise-robust quantum circuits? learn more on these new findings in this PRA research article.

Sound event detection (SED) in the medical environment can be helpful in accomplishing different healthcare tasks. Due to the success of transformer encoder architectures for sound event detection, they seem to be a promising choice for detecting audio events in hospital settings. However, there are two main difficulties in detecting medical audio events with transformers. Firstly, the availability of medical audio data is extremely limited, making it difficult to effectively train a transformer model. Secondly, it is necessary for the SED model to be computationally efficient in order to be deployed in medical environments with limited resources. But, the transformer has high computational complexity because of the attention mechanism. To address these challenges, this paper introduces the Audio Spectrogram Fourier Network (ASFNet), a novel attention-free transformer encoder designed specifically for sound event detection in the medical environment. ASFNet replaces the attention operation with a simplified Fast Fourier Transform. By leveraging this approach, ASFNet outperforms the other methods, achieving a superior average mAP of 0.474 with a 16.76% relative improvement. ASFNet achieves this performance with fewer model parameters and smaller model size, making it an efficient and effective solution for medical audio event detection.

Integrated Thermal and Energy Management of Connected Hybrid Electric Vehicles Using Deep Reinforcement Learning

In this paper, we fuse data from an Inertial Measurement Unit (IMU) and a 2D Light Detection and Ranging (LiDAR) with the help of an Extended Kalman Filter (EKF) for producing a 3D map of an indoor environment. The IMU is mounted on top of the 2D-LiDAR, and this system is attached to the tilt bracket of the Pan-Tilt-Unit (PTU). Point cloud registration algorithms such as Iterative Closest Point (ICP) and Normal Distribution Transform (NDT) are used to estimate the pose using the current and the previous point cloud transformation. Due to certain drawbacks of the ICP and NDT registration algorithms, an optimized version that combines both two algorithms is used for the pose estimation of the setup.

A document by XIANPING MA . Click on the document to view its contents.

Sleep stage classification is pivotal for assessing sleep quality and diagnosing sleep disorders. While deep learning has shown promise in automating sleep stage classification, the dynamic nature of sleep research and data collection processes presents unique challenges. These include class-incremental learning, where models must adapt to newly emerging sleep patterns, and domain-incremental learning, necessitating generalization across diverse data collection conditions or varying EEG equipment. Moreover, the unavailability of old datasets, particularly between different institutions, and the lack of annotations in historical data hinder the effective training and adaptability of deep learning models. This study introduces a Dual-Incremental Learning Framework, amalgamating unsupervised domain adaptation, hierarchy-aware feature learning, and diffusion probabilistic model-based generative replay design to improve the generalizability of sleep stage classification models and enhance the re-analysis of historical data with updated classification schemes. We establish two benchmarks using two publicly available datasets to elucidate the challenges posed by evolving sleep datasets. The effectiveness and performance of the proposed framework are evaluated against the established benchmarks, showcasing enhanced generalizability and the potential for unearthing deeper insights from historical patient data. This framework also demonstrates compatibility with existing deep learning models, promoting a versatile solution for advancing sleep stage classification. To the best of our knowledge, this study is the first to address class and domain-incremental learning challenges in sleep stage classification, laying a solid foundation for future research in this domain. Additionally, the proposed Dual-Incremental Learning Framework holds potential for a wide range of healthcare applications.