Search

Heart rate variability (HRV) is an important metric with a variety of applications in clinical situations such as cardiovascular diseases, diabetes mellitus, and mental health. HRV data can be potentially obtained from electrocardiography and photoplethysmography signals, then computational techniques such as signal filtering and data segmentation are used to process the sampled data for calculating HRV measures. However, uncertainties arising from data acquisition, computational models, and physiological factors can lead to degraded signal quality and affect HRV analysis. Therefore, it is crucial to address these uncertainties and develop advanced models for HRV analysis. Although several reviews of HRV analysis exist, they primarily focus on clinical applications, trends in HRV methods, or specific aspects of uncertainties such as measurement noise. This paper provides a comprehensive review of uncertainties in HRV analysis, quantifies their impacts, and outlines potential solutions. To the best of our knowledge, this is the first study that presents a holistic review of uncertainties in HRV methods and quantifies their impacts on HRV measures from an engineer's perspective. This review is essential for developing robust and reliable models, and could serve as a valuable future reference in the field, particularly for dealing with uncertainties in HRV analysis.

Heart rate variability (HRV) is an important non-invasive parameter to assess the cardiac autonomic nervous system. In particular, spectrum matrices of HRV data have been widely used for physical and mental health monitoring. However, measurement uncertainties from data acquisition and physiological factors can easily affect the HRV spectrum and degrade outcomes of health monitoring. In this paper, we propose a new model for incomplete spectrum estimation of the HRV data based on matrix completion (MC). We show that our model performs efficiently when estimating missing entries for HRV spectra. Moreover, a refined model of matrix completion (RMC) is proposed that can be derived from correlation analysis of the HRV spectra. Two benchmark electrocardiography (ECG) datasets are retrieved and used to derive the HRV data, which are employed to evaluate the performance of our RMC method on the estimation of missing entries in the spectra. Furthermore, four different types of deep recurrent neural networks and the traditional MC method are used for a comparison study, and our RMC method obtains the least estimation error with different masking ratios. The experimental studies and comparison results demonstrate the advantages and robustness of our developed method for the estimation of incomplete HRV spectra.

Optimization within a layer of a deep-net has emerged as a new direction for deep-net layer design. However, there are two main challenges when applying these layers to computer vision tasks: (a) which optimization problem within a layer is useful?; (b) how to ensure that computation within a layer remains efficient? To study question (a), in this work, we propose total variation (TV) minimization as a layer for computer vision. Motivated by the success of total variation in image processing, we hypothesize that TV as a layer provides useful inductive bias for deep-nets too. We study this hypothesis on five computer vision tasks: image classification, weakly supervised object localization, edge-preserving smoothing, edge detection, and image denoising, improving over existing baselines. To achieve these results we had to address question (b): we developed a GPU-based projected-Newton method which is $37\times$ faster than existing solutions., CVPR 2022

Blood pressure (BP) is an important clinical vital sign that varies from beat-to-beat. Nevertheless, these variations cannot be captured by the conventional cuff-based BP monitors. This study proposes and evaluates novel cuffless frameworks to continuously estimate the 10-beat averaged systolic BP (SBP) and diastolic BP (DBP) during dynamic exercise by fusing information from multiple biosensors using five machine learning algorithms. Over 100 thousand beats of data were collected from 62 subjects (aged 59 ± 10 years), each underwent a maximal exercise stress test. The average length of recording for each subject was 35 minutes. The BP ranges were 75-280 mmHg for SBP and 36-157 mmHg for DBP respectively. Multiple physiological parameters were measured continuously and used as inputs to five machine learning algorithms for estimating the 10-beat SBP and DBP averages before, during and after the cycling exercise. The mean absolute error (MAE) of Gaussian process regression (GPR) model was 4.8 mmHg and 3.4 mmHg for SBP and DBP, respectively. The MAE of multiple linear regression (MLR), regression tree (RT), ensemble of trees (ETs), and support vector machine (SVM) models varied from 6.1 mmHg to 17.6 mmHg and from 4.0 mmHg to 9.7 mmHg for SBP and DBP, respectively. The GPR model outperformed the other four models and showed promising results in estimating the 10-beat averages of both SBP and DBP without a cuff in a general elderly population under dynamic conditions.

Different from the traditional methods of assessing the cardiac activities through heart rhythm statistics or P-QRS-T complexes separately, this study demonstrates their interactive effects on the power density spectrum (PDS) of ECG signal with applications for the diagnosis of ST-segment elevation myocardial infarction (STEMI) diseases. Firstly, a mathematical model of the PDS of ECG signal with a random pacing pulse train (PPT) mimicking S-A node firings was derived. Secondly, an experimental PDS analysis was performed on clinical ECG signals from 49 STEMI patients and 42 healthy subjects in PTB Diagnostic Database. It was found that besides the interactive effects which are consistent between theoretical and experimental results, the ECG PDSs of STEMI patients exhibited consistently significant power shift towards lower frequency range in ST-elevated leads in comparison with those of reference leads and leads of health subjects with the highest median frequency shift ratios at 51.39 ± 12.94% found in anterior MI. Thirdly, the results of ECG simulation with systematic changes in PPT firing statistics over various lengths of ECG data ranging from 10 s to 60 mins revealed that the mean and median frequency parameters were less affected by the heart rhythm statistics and the data length but more depended on the alterations of P-QRS-T complexes, which were further confirmed on 33 more STEMI patients in European ST-T Database, demonstrating that the frequency indexes could be potentially used as alternative indicators for STEMI diagnosis even with ultra-short-term ECG recordings suitable for wearable and mobile health applications in living-free environments.

Deep learning (DL) models have demonstrated great potential in cuffless blood pressure (BP) estimation under static conditions, while the performance under dynamic conditions was still not fully validated. This study developed a DL model using population data for training and followed by individualized fine-tuning to directly learn features from multisensory signals including electrocardiogram (ECG), photoplethysmogram (PPG) and PPG derivatives for beat-to-beat BP estimation under water drinking. 25 healthy subjects were recruited, and the leave-one-subject-out approach was used to evaluate the model performance. The results showed that individualized fine-tuning using a small amount of individual baseline data did not change the tracking capability of the model, while can largely reduce the individual bias in dynamic BP estimation, with the mean absolute errors decreased from 13.43 to 9.49 mmHg and 8.48 to 5.54 mmHg for systolic BP and diastolic BP, respectively. It was also found that the model presented better results around the baseline BP levels than that at larger deviations from the baseline, indicating that future work should incorporate individual dynamic data in the fine-tuning to improve dynamic BP estimation further.

Chip-on-Wafer-on-Substrate with Si interposer (CoWoS-S) is a TSV-based multi-chip integration technology that is widely used in high performance computing (HPC) and artificial intelligence (AI) accelerator area due to its flexibility to accommodate multiple chips of SoC, chiplet, and 3D stacks such as high bandwidth memory (HBM). The interposer size increases steadily over the past few years, from one full reticle size (∼830 mm2) to two reticle size (∼1700 mm2). The growth of interposer size offers more integration power to accommodate more active silicon in a package to satisfy the HPC/AI needs. In this paper, we report the new 5th generation CoWoS-S (CoWoS-S5) based on a Si interposer as large as three full reticle size (∼2500 mm2) by a novel 2-way lithography stitching approach. This will accommodate a multiple of logic chips at a total area of 1200 mm2 (with chiplets) together with eight HBM stacks. Besides the dimensional increase of the Si interposer, new features are incorporated to further enhance the electrical and thermal performances of CoWoS-S5 compared with the previous CoWoS-S portfolio. These include an integrated deep trench capacitor (iCap) for enhanced power integrity, 5 layers of sub-micron Cu interconnect with reduced sheet resistance to satisfy high speed die to die interconnect, new TSV structure interposer for both return and insertion loss reduction, and a higher thermal conductivity thermal interface material (TIM) to achieve a lower thermal resistance. Component level reliability with excellent electrical and physical results are also discussed.

One of the prominent challenges for widespread adoption of silicon photonics (SiPh) technology is the availability of an integration platform that can simultaneously meet a wide range of power, performance, and cost criteria in different applications. As a result, there is a diversity of SiPh integrated solutions proposed or demonstrated, but none is considered as a common solution. In this paper, we will first survey industry proposed photonic engine structures in monolithic and heterogeneous integration on their strengths and weaknesses. We will then propose a compact and universal PE structure - COUPE (COmpact Universal Photonic Engine) that could consolidate different requirements onto the same integration platform. COUPE has the electrical IC – photonic IC integration with the electrical interface designed to minimize the EIC-PIC coupling loss. Compared with industry proposed PE technology, COUPE can provide low insertion loss for both grating coupler (GC) and edge coupler (EC). For either GC or EC, the COUPE is a solid structure without cavities or mechanically weak parts, thus enabling low insertion loss without contamination or mechanical concerns. COUPE also has the flexibility to be integrated easily with host ASIC to form a co-package structure. The COUPE integration scheme can meet the most demanding system requirements and pave the way for SiPh-based wafer level system integration (WLSI) for high performance computing applications.

Cuffless blood pressure monitoring is critical for the prevention and early diagnosis of a wide variety of cardiovascular diseases. However, cuffless blood pressure estimation devices suitable for home healthcare applications still suffer from numerous technological limitations. In this paper, we propose a novel method to continuously estimate blood pressure from the pulse morphology of the photoplethysmographic signals, acquired by a commercial oximeter. We conducted cold-pressor tests for 70 healthy subjects, simultaneously acquiring the photoplethysmograms from the oximeter and the continuous blood pressure from a commercial sphygmomanometer as a reference. The shape of each pulse was extracted, normalized, and regressed with the reference blood pressure using support vector machines. The performance of the regression model for each subject was evaluated using 10-fold cross-validation. Results showed that the predicted values from the regression model were highly correlated with the reference measurements from the commercial device. The mean correlation coefficients were 0.748, 0.754, and 0.775 for systolic, diastolic, and mean blood pressures, respectively. The mean errors were 0.043 mmHg, 0.011 mmHg, and 0.008 mmHg, with standard deviations of 5.001 mmHg, 3.689 mmHg, and 4.140 mmHg for systolic, diastolic, and mean blood pressures, respectively. These results suggest high feasibility of the proposed method to be employed in wearable medical and home healthcare devices.

Self-driving vehicle systems have become mainstream as major international research and development companies compete for road detection and mapping projects such as localization. The latter plays an important role in self-driving systems because an accurate determination of the vehicle's position improves trajectory and offset results. And with a good localization function, a high-precision map can be established. This paper proposes using pole maps to enhance vehicle localization while correcting previous vehicle localization trajectory offsets. We have established a pole feature map with pole-shaped characteristic point clouds along the road that are marked with spatial coordinate information. We use three-dimensional light detection and ranging (3D Lidar) as a sensor for environmental scanning, dynamic segment thresholds, and point cloud projections to the image to extract poles, which avoids identification errors typically caused by the scattering effect of 3D Lidar. The resulting pole map uses the occupancy grid map to filter out noise. In terms of localization, we continue to detect pole features while the vehicle is moving, and compare these with the map. When the results match, we update the current localized information. The localization point corrects the information of the trajectory of the undetected pole, which improves vehicle localization accuracy.

Robots are widely used as unmanned vehicles in smart factories. In order to operate the robot in a known environment, the robot must be able to plan the path. The planned path must enable the robot to reach a destination from the starting point and the robot must avoid all obstacles during movement.This study uses a four-parameter logic curve path planning method that uses a closed formula solution and a curve with minimal design parameters to quickly generate an ideal path. The characteristics of the curve are used to derive the S and half-S curves as the solution path. The shortest path is used as the selection reference target to select parameters B and C for the four-parameter logic curve. For the S path, the best solution is chosen. The half -S curve is limited by the elastic end-point heading angle, which gives a unique set of parameter solutions. The generated path does not allow the robot to completely avoid obstacles so a hierarchical half-S curve path planning mechanism is used. A via point is generated at the collision point using the gradient vector for the edge of the obstacle. The planner continues to perform half-S path planning until the path planning ends.

A logic-HBM2E power delivery system on a chip-on-wafer-on-substrate (CoWoS) platform with a deep trench capacitor (DTC) has been designed and analyzed for high performance computing (HPC) applications. The DTC integrated in the silicon interposer of the CoWoS provides the capacitance density of 300 nF/mm2 and low leakage current of

We demonstrate two heterogeneous solutions to improve power delivery to High-Performance Computing (HPC) processors. The scalable HPC vehicle integrates two 7nm CMOS processor chiplets, each with four ARM® Cortex®-A72 cores, that are mounted on a Chip-on-Wafer-on-Substrate (CoWoS®) silicon interposer [1]. In the first solution, Integrated Passive Device (IPD) capacitors are placed directly beneath the interposer to provide more accessible and effective supply noise decoupling. The result is 3.9% higher maximum clock frequency at a core supply of 1.135V. In the second solution, the processor is powered by a laterally mounted in-Package Voltage Regulator (PVR) built in 28nm CMOS augmented with high-per-meability on-die inductors. The processor performance provided by the buck converter-based PVR matches that by an off-package External Voltage Regulator (EVR). As processor power increases with higher core counts, PVRs with on-die inductors will be increasingly compelling for efficient system power delivery.

To accommodate the exceedingly demanding power integrity (PI) requirements for the advanced artificial intelligence (AI) and high performance computing (HPC) components, high-K (HK) based deep trench capacitors (DTC) have been integrated the first time in the silicon interposer with through silicon via (TSV) and fine-pitch interconnects for chip-on-wafer-on-substrate (CoWoS) integration. A specific capacitance density (C s ) of up to 340 nF/mm2 is achieved over a large capacitor array, providing a total capacitance (C t ) of up to 68 μF per interposer die. The HK dielectric has intrinsic time-dependent dielectric breakdown (TDDB) lifetime of > 1,000 years at an operation voltage (V cc ) of 1.35V, and a normalized leakage current (I LK ) density LK & V bd tailing) were observed. The high capacitance, low leakage, large area and reliability-proven Si-interposer integrated DTC, or iCap, provides superior PI performance and therefore greatly enhances the merit of using CoWoS for the next-generation heterogeneous wafer level system integration (WLSI).

Nowadays, unmanned vehicles for navigation applications, such as factory unmanned vehicles and delivery dining systems, are becoming more and more extensive and gradually playing an indispensable role in our lives. Basically, navigation can be divided into three fields, the mapping, the localization and the path planning, in which the path planning is to use a pre-built map to plan a feasible path given the starting and the ending points. Hence, the path planning is the core part of the navigation operation, which is very important at the robot application level, such as the automatic driving and the driverless driving. For aircraft and space exploration, the path planning algorithms can be roughly divided into two types, the graph-based searching and the sampling-based searching. Among the two, the path planning based on random sampling provides with fast operation speed, high success rate on high-dimensional and complex problems, and disuse of extra considerations. The constraint of non-holonomic constraints, and the fast search for random trees are an algorithm based on random sampling. This paper mainly focuses on improving the path divergence in fixed iteration and on leading the direction from root to goal in fixed iteration condition. From the comparison of the experimental results, our approach is approximately 1.5 times better than the RRT on average path length.

Automated driving has been developed rapidly in the past five years. Many automakers have popularized SAE Level 2 on self-driving cars, some even on regular vehicles. To implement Level 3 or Level 4, it’s necessary to rely on a good localization system when GPS signal is not available. For this need, this paper proposes two effective methods to increase the accuracy of the real-time SLAM: The first method: When the original point cloud is input, the point cloud is divided into the short-range group, the medium-range group and the long-range group. An adaptive parameter adjustment method is then used to obtain the optimal parameters for each of these point cloud groups. However, Lidar’s physical characteristics can cause the point cloud to be insufficient, making an important part of points misjudged as outliers for the medium-range and long-range cases. Thanks to the help of the adaptive parameters, these point clouds, which were originally misjudged as outliers can be preserved in this paper. The second method: In point clouds, the same object in different ranges is represented with different point clouds. Hence, features, such as the roughness and density, can dramatically change with the variation of distance even when it is the same object. To solve this problem, we have designed three different range point cloud feature extraction methods to get more accurate point cloud features, such as the planes or edges. By combining these two steps, the LeGO-LOAM accuracy can be effectively increased by more than 30% while achieving the performance of the real-time SLAM, which is more accurate and faster than the NDT-Mapping used in Autoware.

We introduce SAIL-VOS (Semantic Amodal Instance Level Video Object Segmentation), a new dataset aiming to stimulate semantic amodal segmentation research. Humans can effortlessly recognize partially occluded objects and reliably estimate their spatial extent beyond the visible. However, few modern computer vision techniques are capable of reasoning about occluded parts of an object. This is partly due to the fact that very few image datasets and no video dataset exist which permit development of those methods. To address this issue, we present a synthetic dataset extracted from the photo-realistic game GTA-V. Each frame is accompanied with densely annotated, pixel-accurate visible and amodal segmentation masks with semantic labels. More than 1.8M objects are annotated resulting in 100 times more annotations than existing datasets. We demonstrate the challenges of the dataset by quantifying the performance of several baselines. Data and additional material is available at http://sailvos.web.illinois.edu.

Cloud computing technology has immediate impact on the development of software testing. However, the study about autonomic self-testing of regression and internationalization based on cloud computing is still in its early stage. Our study is based on a regression and i18n testing project about enterprise office collaboration suite. Firstly, automated function test script for baseline version (English) is developed by Java, and it is implemented reusability for i18n testing in other 9 languages. In addition, a comprehensive AST model of regression and i18n is proposed. The detailed design and implementation of this system based on the daily build is presented. Finally, the multi-server architecture and the basis process of testing are elaborated. The above mentioned testing script and platform have been applied to actual testing of web products, and achieved good results.