Real-time Threshold Research Articles

Adaptive metrics for an evolving pandemic: A dynamic approach to area-level COVID-19 risk designations

Throughout the COVID-19 pandemic, policymakers have proposed risk metrics, such as the CDC Community Levels, to guide local and state decision-making. However, risk metrics have not reliably predicted key outcomes and have often lacked transparency in terms of prioritization of false-positive versus false-negative signals. They have also struggled to maintain relevance over time due to slow and infrequent updates addressing new variants and shifts in vaccine- and infection-induced immunity. We make two contributions to address these weaknesses. We first present a framework to evaluate predictive accuracy based on policy targets related to severe disease and mortality, allowing for explicit preferences toward false-negative versus false-positive signals. This approach allows policymakers to optimize metrics for specific preferences and interventions. Second, we propose a method to update risk thresholds in real time. We show that this adaptive approach to designating areas as "high risk" improves performance over static metrics in predicting 3-wk-ahead mortality and intensive care usage at both state and county levels. We also demonstrate that with our approach, using only new hospital admissions to predict 3-wk-ahead mortality and intensive care usage has performed consistently as well as metrics that also include cases and inpatient bed usage. Our results highlight that a key challenge for COVID-19 risk prediction is the changing relationship between indicators and outcomes of policy interest. Adaptive metrics therefore have a unique advantage in a rapidly evolving pandemic context.

Sea-Surface Small Target Detection Based on Four Features Extracted by FAST Algorithm

On account of current algorithm and parameter design difficulties and low detection accuracy in feature extractions of small target detections in sea clutter environment, this paper proposes a correspondingly improved four feature extraction method by FAST. After the short-time Fourier transform is applied, a time–frequency distribution spectrogram of original data is generated. Candidate feature points (CFP) are first extracted by FAST algorithm, and then a four-feature extraction is implemented with FAST and DBSCAN combined. The feature distinction is enhanced through a feature optimization. Upon the construction of the four-dimensional feature vectors, XGBoost classifier algorithm classifies and detects these feature vectors. The genetic algorithm optimizes the hyperparameters in XGBoost and updates the decision threshold in real time to control the detection method’s false alarm rate. The IPIX dataset is employed for experimental verification. Verification results confirm that this proposed detection method has better performance than several other currently used detection methods. The detection performance is improved by 7% and 13.8% when observation time is set at 0.512 s and 1.024 s, respectively.

Traumatic brain injury risk assessment with smart technology

Mild traumatic brain injuries (mTBIs) continue to burden our warfighters. The high-tech industry has delivered wearable Micro-Electro-Mechanical System (MEMS) head-impact sensors to monitor impact forces. So far, these MEMS sensors have categorically failed to detect mTBIs and are therefore of no clinical utility for diagnosis. Our recent studies have shown that human head kinematics is anisotropic with respect to pitch–roll–yaw degrees of freedom of the head and neck. In the present project, we generated head acceleration datasets on non-injurious impacts and mTBI events based on mean values from the literature. We then augmented the simulated data with pitch–roll–yaw information followed by machine learning with a Classification and Regression Tree analysis. Our results revealed that head angular acceleration in pitch is the best predictor. More than 81.3 % of concussive injuries had head angular accelerations in pitch exceeding 3527 rad/s2. Out of 18.6% of concussive injuries with head angular accelerations in pitch under 3527 rad/s2, 75% of these cases had head angular accelerations in roll exceeding 1679 rad/s2. This study shows that artificial intelligence and machine learning should be able to provide accurate identification of subject-specific concussive thresholds in real time and in the field, thereby moving concussion diagnosis toward precision medicine.

Intelligent flow control algorithm for microservice system

In microservice systems, availability can be ensured through a variety of measures, such as fault tolerance and flow limiting, which are collectively called the flow control. In the current mainstream system design, the flow control rules are usually fixed and set manually, which cannot be dynamically adjusted according to the flow shape. The performance of the system is thus not fully explored. To mitigate this problem, an adaptive dynamic flow control algorithm is proposed. Based on the system's monitoring data and current flow, the algorithm calculates the flow-limiting threshold in real time, and then it implements fine-grained service adaptive flow control to improve the resource utilization. Experimental results show that the performance of the adaptive automatic flow control is better than that of the traditional static method on resource utilization.

SURG-03. IMMERSIVE VIRTUAL REALITY APPLICATIONS IN NEUROSURGICAL ONCOLOGY

Tridimensional (3D) rendering of volumetric neuroimaging is increasingly been used to assist surgical management of brain tumors. New technologies allowing immersive virtual reality (VR) visualization of obtained models offer the opportunity to appreciate neuroanatomical details and spatial relationship between the tumor and normal neuroanatomical structures to a level never seen before. We present our preliminary experience with the Surgical Theatre, a commercially available 3D VR system, in 60 consecutive neurosurgical oncology cases. 3D models were developed from volumetric CT scans and MR standard and advanced sequences. The system allows the loading of 6 different layers at the same time, with the possibility to modulate opacity and threshold in real time. Use of the 3D VR was used during preoperative planning allowing a better definition of surgical strategy. A tailored craniotomy and brain dissection can be simulated in advanced and precisely performed in the OR, connecting the system to intraoperative neuronavigation. Smaller blood vessels are generally not included in the 3D rendering, however, real-time intraoperative threshold modulation of the 3D model assisted in their identification improving surgical confidence and safety during the procedure. VR was also used offline, both before and after surgery, in the setting of case discussion within the neurosurgical team and during MDT discussion. Finally, 3D VR was used during informed consent, improving communication with families and young patients. 3D VR allows to tailor surgical strategies to the single patient, contributing to procedural safety and efficacy and to the global improvement of neurosurgical oncology care.

Dynamic evolutionary model based on a multi-sampling inherited HAPFNN for an aluminium electrolysis manufacturing system

It is technically difficult to accurately establish a dynamic adaptive model of an aluminium electrolysis manufacturing system (AEMS) because the system has many complex characteristics, such as multiple parameters, dynamic time variance, and a non-Gaussian distribution of process data. Inspired by the overall superiority of particle filtering theory in dealing with non-linear non-Gaussian problems, this paper presents a novel method based on a multi-sampling inherited hybrid annealed particle filter neural network (MSI-HAPFNN). Firstly, the neural network’s (NN’s) weights and thresholds are used as the state variables of hybrid annealed particle filter; Secondly, the hybrid proposal distribution obtained by sampling the above state variables is employed to replace the posterior proposal distribution in the standard particle filter (PF) algorithm as the importance density function, thereby adjusting the NN’s weights and thresholds in real time. Thirdly, the model achieves the features of multi-sampling and inheritance by introducing NN and PF weights, and using adaptive inheritance method. Therefore, this paper systematically proposes the theoretical construction framework and experimental procedure of MSI-HAPFNN. Furthermore, this article also introduces a genetic algorithm to thoroughly evaluate the prediction potential. The proposed model has been tested on the real-world system for aluminium electrolysis manufacturing and compared with several closely related frameworks. The experimental results show that the MSI-HAPFNN model can significantly improve the self-adaptive ability of the object system to working conditions and the prediction accuracy of power consumption, which is helpful in finding optimal design parameters in an AEMS.

Preventing rangeland degradation: a shared problem for Australia and China

Rangeland degradation continues in Australia, China and elsewhere. The stocking rate/animal production relationship has been a successful concept for pastoralists wanting to avoid degradation and/or raise incomes. However, there are no means available of alerting pastoralists to the approach of critical thresholds that would ‘flip’ rangelands into alternative states when grazing-stressed. Critical threshold forecasting for avoiding degradation (and seizing restoration opportunities) could be made available online. Research has yet to find, assemble and test the set of indicators needed to forecast the approach of critical thresholds envisaged in State-and-Transition thinking. Forecasting at paddock, property and regional scales would have to involve high-performance computing because the thresholds will be space and time dependent. The case for Australia and China to contribute cooperatively to this research effort rests on the large number of contrasting rangeland ecosystems across the two countries that represent rangelands globally. A proven history of past collaboration is extant with existing research programs on plant population dynamics, landscape patchiness/leakiness and soil biota status, and their responses to the separate and combined effects of climate and grazing animals. The road to adoption would involve partnerships with pastoralists throughout the process, remote sensing to identify approaching thresholds in real time, application of high-performance computing and possibly artificial intelligence, and packaging of forecasts for different socio-economic rangeland systems.

Enhanced soil profile visualization using portable X-ray fluorescence (PXRF) spectrometry

Soil horizonation has traditionally relied upon morphological description, field sampling, and laboratory analysis as a means of establishing key diagnostic horizons and features within a soil profile. While this technique is useful for many soil properties, other quantifiable properties associated with elemental abundance may be visually imperceptible. Portable X-ray fluorescence (PXRF) spectrometry allows for rapid elemental data acquisition in-situ, providing more than 20 elements commonly detectable within a soil profile. In this study, a grid containing 130 cells was physically imposed upon two soil profiles in West Texas, USA. PXRF was used to sequentially scan each grid cell and acquire elemental data. The profiles were morphologically described by a panel of soil scientists, sampled, and subjected to traditional laboratory characterization. Data from the PXRF and field sampling were statistically compared, then visualized as depth functions. A data visualization tool was created to overly elemental data on top of a digital photograph of the soil profile in semitransparency. Using elemental data as a proxy, PXRF data was able to detect both argillic and calcic horizons within one soil profile and visually illustrate their extent and assist in delineating horizons in a second profile with nondescript horizons. The soil visualization tool has other features as well such as correlation analysis between elements, adjustable correlation threshold in real time, rasterized vs. smoothed data overlay, and box-plots of elemental concentrations. Summarily, this research developed a new tool for PXRF data visualization in support of soil pedon description, allowing for elemental data which may be visually imperceptible to be considered during soil profile characterization. Future advancements will seek to integrate this tool directly into the PXRF software for on-site visualization of any element specified by the analyst. Applications of such could include real-time mapping of contaminant metals in soil profiles.

SEAL: SDN based secure and agile framework for protecting smart city applications from DDoS attacks

Evolution of smart cities induces critical challenges related to cyber and network security. The increased reliance of a smart city on Information and Communication Technologies (ICT) infrastructure improves automation, efficiency, and sustainability of city services. However, it also poses enormous challenges for ensuring continued operations and services at all times and especially under cyber-attacks. Any lapse in cyber security can lead to critical disaster across the city. Distributed Denial of Service (DDoS) attacks are considered to be the most predominant and prevalent cyber-attacks. We believe that smart city could consist of numerous applications with varying level of network and security requirements. Therefore, providing an adaptive mechanism against DDoS attacks for all applications in a smart city is a key challenge. Further, considering the wide-scale requirements of a smart city, developing an adaptive and flexible solution is a key requirement. Considering these requirements, this paper presents SEAL (SEcure and AgiLe) – a novel Software Defined Networking (SDN) based adaptive framework for protecting smart city applications against DDoS attacks. The SEAL framework leverages key characteristics of SDN such as the global visibility, centralized control, and programmability to enhance the security and resilience. SEAL is capable of effectively detecting and mitigating DDoS attacks not only on application servers but also on network resources. SEAL is also unique in this regard that it provides application specific DDoS attack security solution instead of static threshold mechanism. Moreover, inherently distributed architecture of the SEAL framework ensures fault tolerance, scalability and reliability of the smart city. The SEAL framework comprises three modules, namely D-Defense, A-Defense and C-Defense. These modules collectively provide a mechanism to detect and mitigate DDoS attack on smart city applications and the network infrastructure. Adaptability in SEAL is achieved through implementing customized version of estimated-weighted moving average (EWMA) filters. Three types of filters, Proactive Filter, Active Filter, and Passive Filter are proposed and implemented to compute the dynamic threshold in real time for various types of applications. Experimental evaluation of the SEAL framework has been conducted to establish the efficacy of the framework and its components in detecting and mitigating DDoS attacks. The results prove that SEAL is able to detect and mitigate DDoS attacks effectively. The focus of the SEAL framework is to protect smart city applications, however, the SEAL framework can potentially be utilized in a wide range of systems.

Programmable flow rate scanner for evaluating detector sensitivity regime

In order to expedite characterization of on-line detection systems, we have developed a flow rate scanner comprising a pump, an inexpensive single-board computer, and an associated Python program. The program controls the pump, triggers the detector, acquires raw data, filters/smooths the data, calculates the first derivative of the temporal signal trend, and determines the sensitivity regime threshold in real time. The slope of the curve relating the detector response with the sample flow rate is used to determine the sensitivity regime for a given analyte and a given detector. The key part of the data analysis is on-the-fly calculation of the first derivatives using a fourth-order polynomial fit to the raw data by the Savitzky-Golay algorithm. The device has been tested by determining the flow rate profiles for an ultraviolet absorption detector and a mass spectrometer operated with electrospray ionization, atmospheric pressure chemical ionization, and a dual ion source. The sensitivity regime thresholds for various test analytes have been determined automatically and robustly in time as short as a few minutes.

Design and performance evaluation of collision protection-based safety operation for a haptic robot-assisted catheter operating system.

The robot-assisted catheter system can increase operating distance thus preventing the exposure radiation of the surgeon to X-ray for endovascular catheterization. However, few designs have considered the collision protection between the catheter tip and the vessel wall. This paper presents a novel catheter operating system based on tissue protection to prevent vessel puncture caused by collision. The integrated haptic interface not only allows the operator to feel the real force feedback, but also combines with the newly proposed collision protection mechanism (CPM) to mitigate the collision trauma. The CPM can release the catheter quickly when the measured force exceeds a certain threshold, so as to avoid the vessel puncture. A significant advantage is that the proposed mechanism can adjust the protection threshold in real time by the current according to the actual characteristics of the blood vessel. To verify the effectiveness of the tissue protection by the system, the evaluation experiments in vitro were carried out. The results show that the further collision damage can be effectively prevented by the CPM, which implies the realization of relative safe catheterization. This research provides some insights into the functional improvements of safe and reliable robot-assisted catheter systems.

A Novel Zero Velocity Interval Detection Algorithm for Self-Contained Pedestrian Navigation System with Inertial Sensors.

Zero velocity update (ZUPT) plays an important role in pedestrian navigation algorithms with the premise that the zero velocity interval (ZVI) should be detected accurately and effectively. A novel adaptive ZVI detection algorithm based on a smoothed pseudo Wigner–Ville distribution to remove multiple frequencies intelligently (SPWVD-RMFI) is proposed in this paper. The novel algorithm adopts the SPWVD-RMFI method to extract the pedestrian gait frequency and to calculate the optimal ZVI detection threshold in real time by establishing the function relationships between the thresholds and the gait frequency; then, the adaptive adjustment of thresholds with gait frequency is realized and improves the ZVI detection precision. To put it into practice, a ZVI detection experiment is carried out; the result shows that compared with the traditional fixed threshold ZVI detection method, the adaptive ZVI detection algorithm can effectively reduce the false and missed detection rate of ZVI; this indicates that the novel algorithm has high detection precision and good robustness. Furthermore, pedestrian trajectory positioning experiments at different walking speeds are carried out to evaluate the influence of the novel algorithm on positioning precision. The results show that the ZVI detected by the adaptive ZVI detection algorithm for pedestrian trajectory calculation can achieve better performance.

Thresholded Smoothed <inline-formula> <tex-math notation="LaTeX">$\ell_0$</tex-math></inline-formula> Norm for Accelerated Sparse Recovery

Smoothed $\boldsymbol{\ell_0} $ norm (SL0) is a fast and complex domain extendible sparse recovery algorithm which is suitable for many practical real-time applications. In this letter, we propose an improved algorithm termed “Thresholded Smoothed $\boldsymbol{\ell_0} $ Norm (T-SL0) ” for accelerating the iterative process of SL0. T-SL0 introduces an iterative efficiency indicator and compares it with a preset threshold in real time to determine whether or not the current iteration should be executed. Through identifying and bypassing low efficient iterations, our approach converges much faster than the original SL0 algorithm. Experimental results are presented to demonstrate that our approach can accelerate SL0 significantly without loss of accuracy.

In vivo dose measurements in interventional radiology

Introduction Interventional radiology is constantly evolving. During these last years, acts performed under radiological guidance have significantly increased. However, doses delivered to patients may be significant and result in deterministic effects. Therefore, it is necessary to set up tools for monitoring the dose to the patient's skin. Materials and methods Interventional Reference Levels (IRL) peculiar to the department are generated by applying 75th percentile method. They allow creating warning thresholds in real time. At the same time as the use of the indicators, Nanodot® pastilles marketed by the company Landauer are used. These dosimeters with deferred reading are based on OSL technology (Optically Stimulated Luminescence) and are analyzed on-site by reader Microstar®. Dose delivered to patients is determined a posteriori and then compared to the indicators estimated by the local IRL. Results A retrospective collection of dosimetric quantities (DAP in cGy cm2 and DLP in mGy cm) has been performed for different types of practiced examinations. This review allowed determining local reference levels such they already exist in conventional radiology. They are intended to alert and warn operators when exceeding a threshold during an examination. OSL calibration for 0,010 to 10 Gy dose range is currently in process and will allow the putting in clinical routine of the in vivo dosimetry for acts of interventional radiology the most irradiant. Conclusion The complementarity of these dosimetric assessments will allow us to ensure dosimetric patients monitoring. We hope to obtain a better control on doses delivered by taking greater account of the radiation risk. Patients who will receive a significant dose (dose in the skin superior to 3 Gy) will be informed by the radiologist. They will receive information relative to the surveillance of the exposed zone and the involved risks as well as specific medical care if necessary.

Reanalyzing auditory sensitivity: The functional audiogram as modeled by the bat detecting moth ear.

Auditory sensitivity has often been measured by identifying neural threshold in real time (online) which can introduce bias in the audiograms that are produced. This was tested by recording auditory nerve activity of the Notodontid moth Nadata gibbosa elicited by bat-like ultrasound and analyzing the response offline. The offline audiogram was compared to a published online audiogram showing that the bias introduced can result in a difference in both the best frequency of this moth ear and the audiogram shape. The offline neural threshold audiogram was then compared to audiograms produced using behavioral threshold definitions based on (1) spike period and (2) the latency to first spike, showing that audiograms produced using these theoretical behavioral definitions to have a similar shape. Finally, predictions on the distance at which Notodontid moths may evade bats using negative phonotaxis or the acoustic startle response are made, using the number of spikes elicited as a proxy for distance.

Developed Incident Detection Algorithm Compared with Neural Network Algorithms

The CUSUM (cumulative sum of log-likelihood ratio) algorithm is an optimization-based algorithm that is attractive for many applications because it can minimize detection delay and can explicitly incorporate the characteristics of processes before and after changes. One such application is freeway incident detection, where field-measurable traffic-flow parameters are used to flag incidents in real time in an expedient and reliable manner. In the presented study, the special characteristics of traffic processes associated with incidents are incorporated into the CUSUM algorithm for freeway incident detection. In the algorithm evaluation, the most recently developed neural networks are compared with an enhanced CUSUM algorithm. The neural network algorithms are systematically evaluated first among themselves, and then the best of them is compared with the CUSUM algorithm. The results demonstrate that the CUSUM incident detection algorithm can perform better than the neural network algorithms. The neural network algorithm may show inferior performance because it cannot adjust its decision threshold in real time.

Real-Time Isocontouring and Texture Mapping Meet New Challenges in Interactive Molecular Graphics Applications

A technique to extract solid isosurfaces from three-dimensional electron density data at high speed is presented. The ability to change the contouring threshold in real time renders the method a powerful tool for interactive analysis of proteins and their supramolecular assemblies, in order to compare and combine structural information gathered by different data acquisition methods. Chemical properties can effectively be mapped onto these isosurfaces by the use of texture mapping. The implications of these methods in combination with other visualization techniques are discussed, and a framework for their integration into a general-purpose molecular graphics toolkit is proposed.