Real-time Predictor Research Articles

Benchmarking of early ontogeny parameters as predictors of the first-feeding larvae vitality in spotted rose snapper Lutjanus guttatus

A common problem in marine fish hatcheries is the lack of functional embryonic parameters reliably predicting favorable larval outcomes. Embryos characterized as high-quality by typical assessment methods frequently result in weak larvae with high mortality during the endo-exotrophic transition. Thus, because there is no universal consensus on what parameters are most associated with successful larval survivability, this study aims to determine the most reliable, practical, and real-time predictor of vitality at the earliest developmental stage for first-feeding larvae of L. guttatus. From 18 mass volitional spawnings, the reproductive performance and quality of embryos and larvae were assessed by following quality parameters: total fecundity per spawning, percentage of floating eggs, fertilization rate, blastomere morphology score (at 8–32 cell stages), egg and oil globule volumes, coefficient of variation of eggs and oil globule diameters, percentage of poly-globular eggs, oil globule-to-egg volume percentage, morphological appearance and survival rates at 24 (hatching), 48 and 72 h post-spawning, larvae transition rate, and survival activity index (SAI). SAI was used as a larval vitality benchmark to determine the predictive capacity of the abovementioned parameters. Our results show that blastomere morphology score was the only parameter positively correlated with SAI. However, when examining each component of blastomere morphology to SAI prediction, only the presence of cell inclusions, but not size, symmetry, margins, or adhesions of blastomeres, predict the SAI. This reveals that embryos with defective blastomeres can develop into normal-appearing larvae but are weak to starvation. Therefore, cell inclusions are the best real-time, early, reliable, and functional predictor of first-feeding larvae vitality in L. guttatus.

Less-invasive subdural electrocorticography for investigation of spreading depolarizations in patients with subarachnoid hemorrhage.

Wyler-strip electrodes for subdural electrocorticography (ECoG) are the gold standard for continuous bed-side monitoring of pathological cortical network events, such as spreading depolarizations (SD) and electrographic seizures. Recently, SD associated parameters were shown to be (1) a marker of early brain damage after aneurysmal subarachnoid hemorrhage (aSAH), (2) the strongest real-time predictor of delayed cerebral ischemia currently known, and (3) the second strongest predictor of patient outcome at 7 months. The strongest predictor of patient outcome at 7 months was focal brain damage segmented on neuroimaging 2 weeks after the initial hemorrhage, whereas the initial focal brain damage was inferior to the SD variables as a predictor for patient outcome. However, the implantation of Wyler-strip electrodes typically requires either a craniotomy or an enlarged burr hole. Neuromonitoring via an enlarged burr hole has been performed in only about 10% of the total patients monitored. In the present pilot study, we investigated the feasibility of ECoG monitoring via a less invasive burrhole approach using a Spencer-type electrode array, which was implanted subdurally rather than in the depth of the parenchyma. Seven aSAH patients requiring extraventricular drainage (EVD) were included. For electrode placement, the burr hole over which the EVD was simultaneously placed, was used in all cases. After electrode implantation, continuous, direct current (DC)/alternating current (AC)-ECoG monitoring was performed at bedside in our Neurointensive Care unit. ECoGs were analyzed following the recommendations of the Co-Operative Studies on Brain Injury Depolarizations (COSBID). Subdural Spencer-type electrode arrays permitted high-quality ECoG recording. During a cumulative monitoring period of 1,194.5 hours and a median monitoring period of 201.3 (interquartile range: 126.1-209.4) hours per patient, 84 SDs were identified. Numbers of SDs, isoelectric SDs and clustered SDs per recording day, and peak total SD-induced depression duration of a recording day were not significantly different from the previously reported results of the prospective, observational, multicenter, cohort, diagnostic phase III trial, DISCHARGE-1. No adverse events related to electrode implantation were noted. In conclusion, our findings support the safety and feasibility of less-invasive subdural electrode implantation for reliable SD-monitoring.

Real-time physical field reconstruction for nanofluids convection using deep learning with auxiliary tasks

Nanofluids has become one of the most promising cooling fluids due to its great thermophysical properties, especially for microchannels. In recent years, to avoid prohibitively expensive simulations in design process or iterative optimization, massive data-driven and machine learning methods have been employed to construct surrogate models for thermophysical properties or concerned performance characteristics. Though the favorable accuracy is validated, this kind of surrogate models ignore the importance of physical fields, which are also an evaluating criterion in heat transfer design scenarios. In this article, we develop a deep learning framework to achieve the main task of real-time reconstruction mapping from design variables to physical fields. In addition, the auxiliary tasks of prediction mapping from physical fields to performance characteristics are tailored to improve the accuracy and generalization of reconstruction mapping. To validate the feasibility of this framework, we assess the ability of the generator alone to reconstruct physical fields and predict performance characteristics for the flow and heat transfer of nanofluid first. The results show that the introduction of auxiliary task can greatly improve the performance of main task. The computational performance analysis indicates that a well-trained deep learning model with GPU-accelerated has three orders of magnitude faster than CFD solver. This proposed approach was expected to provide a real-time predictor with a reasonable level of accuracy for nanofluid convective heat transfer.

Rothman Index As a Real-Time Predictor of Intensive Care Unit Admissions and Mortality in Acute Myeloid Leukemia Patients Undergoing Induction Chemotherapy

Introduction: In-hospital mortality is higher for patients (Pts) with Acute Myeloid Leukemia(AML) requiring intensive care unit(ICU) admissions (43.1% Vs 9.3%) as described by Halpren,A.B., et al (JAMA Oncol. 2017. 3(3): p. 374-381). Various demographic factors identify patients at risk but may not fully provide primary prevention strategies. The Rothman Index (RI) is a dynamic acuity metric that incorporates 26 variables of physiological measures, clinical assessments, and laboratory results into a composite index that is trended over time. In our study, we hypothesize that RI at admission for induction chemotherapy (IRx) and the number of unstable days defined as a drop in RI by 15 points in a 24-hour period; can predict ICU admissions. This is a score that is trended in real time providing potential opportunities for Pt stabilization and preventing ICU admissions.Methods: This is a single center retrospective observational study of AML pts admitted between 8/24/12 to 4/22/17. Data collection included age at diagnosis, RI at admission, unstable days, days of stability before first unstable day, total length of hospital stay (LOS) for IRx, days to first readmission, total number of unstable days for readmission and mortality. The primary end point was ICU admission. Continuous variables were reported as means and standard deviations or means with confidence intervals or median with inter quartile range(IQR). Dichotomous variables were reported as proportions. A logistic regression model was used to predict probability of ICU admission.Results: A total of 83 pts (54.21% males) were studied with a median age at diagnosis of 66 (IQR 19) years(y) and mean LOS of 21.79±16.19 days. Twenty (24.09%) pts had ICU stay in their index admission and the age at admission was very similar to the non-ICU group [Median 66y (IQR 21) Vs 67y (IQR 19)]. Nineteen pts died with a mortality rate of 22.89%. The RI at admission for the ICU group [Mean 73.34 (CI 63.14-83.54) Vs 82.74 (CI 79.72-85.76)] and days of stability before the first unstable day [Mean 2.89 (CI 1.05-4.73) Vs 5.63 (CI 4.23-7.03)] were lower. Number of unstable days [Mean 8.57 (CI 6.20-10.95) Vs 3.36 (CI 2.32-4.40)] and LOS for the IRx admission [Mean 26.10 (CI 20.23-31.97) Vs 20.38 (CI 16.05-24.70)] was higher in the ICU group. Number of unstable days was the best predictor of ICU admissions [odds ratio 1.25 (CI 1.10-1.41 p 0.00)] as well as days of stability before first unstable day [odds ratio 0.83 (CI 0.70-0.98 p 0.03)] and RI at admission [odds ratio 0.96 (CI 0.93-0.99 p 0.04)]. However, as mortality predictors, number of unstable days [odds ratio 1.08 (CI 0.96-1.21 p 0.18)], days of stability before first unstable day [odds ratio 1.06 (CI 0.95- 1.18 p 0.28)] and RI at admission [odds ratio (CI 0.96-1.04 p 0.74)] did not meet statistical significance and this could be due to the smaller sample size.Conclusions:Real time trends in RI can potentially predict AML pts undergoing IRx that may need critical care. Higher number of unstable RI days along with fewer stable days before first unstable day predict higher likelihood for ICU stay. Finally, RI also can predict the longer LOS in patients with ICU admissions. The Clinical implications of using RI can be far reaching as it provides a real-time opportunity for utilizing clinical resources to stabilize pts on the floor, prevent ICU admissions and reduce LOS. DisclosuresIyer:Genentech: Research Funding; Takeda: Research Funding.

Real-time neural network based predictor for cov19 virus spread.

Since the epidemic outbreak in early months of 2020 the spread of COVID-19 has grown rapidly in most countries and regions across the World. Because of that, SARS-CoV-2 was declared as a Public Health Emergency of International Concern (PHEIC) on January 30, 2020, by The World Health Organization (WHO). That’s why many scientists are working on new methods to reduce further growth of new cases and, by intelligent patients allocation, reduce number of patients per doctor, what can lead to more successful treatments. However to properly manage the COVID-19 spread there is a need for real-time prediction models which can reliably support various decisions both at national and international level. The problem in developing such system is the lack of general knowledge how the virus spreads and what would be the number of cases each day. Therefore prediction model must be able to conclude the situation from past data in the way that results will show a future trend and will possibly closely relate to the real numbers. In our opinion Artificial Intelligence gives a possibility to do it. In this article we present a model which can work as a part of an online system as a real-time predictor to help in estimation of COVID-19 spread. This prediction model is developed using Artificial Neural Networks (ANN) to estimate the future situation by the use of geo-location and numerical data from past 2 weeks. The results of our model are confirmed by comparing them with real data and, during our research the model was correctly predicting the trend and very closely matching the numbers of new cases in each day.

Expansion and additional validation of PKA17: A fast real-time and web-based pKa predictor

We have further improved and validated PKA17, our fast software for predicting p[Formula: see text] values of protein residues. The methodology employs coarse-grained lattice-based model of proteins. It was previously demonstrated to perform ca. an order of magnitude faster than such successful and widely used frameworks as PROPKA without losing accuracy of the calculations. In this paper, we report the following improvements: (i) We have expanded our training and testing sets of protein residues by 128%, from 442 to 1009 cases; (ii) we have added and parameterized PKA17’s capability to predict acidity constants of cysteine residues that are important in many biomedical applications, including but not limited to binding of such transition metal ions as copper(I) and platinum(II); (iii) we have carried out the comparison of accuracy of predicted Asp and Glu p[Formula: see text] values not only between PKA17 and PROPKA, but also with DelPhiPKa and H[Formula: see text]. The computational speed of PKA17 remains the highest of all the methods used in our studies, and the accuracy of PKA17 is somewhat inferior only to those of such more sophisticated methods as Multi-Conformation Continuum Electrostatic (MCCE) ones. For instance, the average unsigned deviations of predicted p[Formula: see text] values from the experiment for 416 Glu residues were found to be 0.706, 0.766, 0.867, and 0.520[Formula: see text]pH units when obtained with PROPKA, DelPhiPKa, H[Formula: see text], and PKA17, respectively (0.487[Formula: see text]pH units with PKA17 after refitting). The average unsigned errors for cysteine p[Formula: see text] values calculated with PROPKA, DelPhiPKa, H[Formula: see text], and PKA17 were 3.50, 2.06, 3.17, and 1.26[Formula: see text]pH units. PKA17 has also performed well in assessing the cysteine acidity constants of the CXXC motif of CopZ protein involved in binding of copper(I) metal ions. Our results demonstrate that the PKA17 methodology and current parameters are accurate and robust, and its computational speed makes it possible to be employed in large-scale p[Formula: see text] screening calculations and in constant-pH protein dynamics simulations. The resulting PKA17 software has been deployed online at http://kaminski.wpi.edu/PKA17/pka_calc.html .

Abstract 4295: Early circulating tumor DNA dynamics as a real-time predictor of FOLFOX efficacy in advanced colorectal cancer patients

Abstract Background: In recent years, various efforts have been made to identify biomarkers in metastatic colorectal cancer patients (mCRC), all with the goal of improving patients' outcomes, including maximizing therapeutic response and minimizing exposure to ineffective treatments. Nevertheless, there is still no valid biomarker for the early assessment of therapeutic efficacy in the patient management strategy. To address this issue, we aimed to describe the predictive value of circulating tumor DNA (ctDNA) on the efficacy of FOLFOX treatment in mCRC patients. Patients and methods: A total of 11 patients with mCRC receiving FOLFOX therapy were included in this study. During the 48-hour FOLFOX cycle, we performed serial liquid biopsy-based ctDNA analysis before treatment start (T1) and at eight further time points (T2-T9). Among the 11 patients tested, all patients had detectable variants identified by either tumor tissue genotyping or baseline sample ctDNA sequencing for longitudinal analysis. We assessed genome-wide somatic copy number alterations (SCNAs) and highly sensitive sequencing approaches were utilized to monitor changes in the ctDNA mutant allele frequencies (mAFs) between baseline and on-treatment samples. Results: We were able to clearly assess different mAF patterns across patients with either stable disease, partial response or progressive disease. However, we invariably observed an early and deep ctDNA mAF decrease in all patients, as comparing the baseline (T1) mAF levels with the mAFs at all time points revealed that the most significant ctDNA drop was 23 hours after treatment start, i.e. at time point T5 (p <0.005). In addition, we observed in patients with stable disease or partial response (n=8) not only a significant decline of mAFs between T5 (p=0.0391) but also at T9 (52 hours after treatment start; p=0.0156) compared to the baseline sample. Conversely, mAFs of patients with progressive disease increased again within the next 29 hours (T9) after the initial ctDNA drop at time point T5, leading to higher ctDNA levels at T9 than the baseline values prior to treatment start. Conclusion: Our observations clearly demonstrated for the first time that ctDNA dynamics allow real-time, individualized evaluation of treatment response using two early time points, i.e. at T5 and T9. Therefore, our data suggest that sequential ctDNA analysis can contribute to an improved patient treatment strategy. Nevertheless, further studies are clearly required to validate this promising early on-treatment indicator of therapeutic efficacy for CRC patients. Citation Format: Tina Moser, Julie Waldispuehl-Geigl, Jelena Belic, Samantha Perakis, Sabrina Weber, Qing Zhou, Heinz Sill, Sigurd Lax, Karl Kashofer, Gerald Hoefler, Helmut Schoellnast, Thomas Bauernhofer, Ellen Heitzer, Jochen B. Geigl, Michael R. Speicher. Early circulating tumor DNA dynamics as a real-time predictor of FOLFOX efficacy in advanced colorectal cancer patients [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4295.

Characterising recovery from renal transplantation and live-related donation using cardiopulmonary exercise testing

Background An association between end-stage renal failure and exercise intolerance exists. Whether live kidney donation impacts on exercise tolerance is unknown. Here recovery post renal transplant and donation using cardiopulmonary exercise testing is investigated. Methods Renal donors (n = 28) and recipients (n = 24) undertook a cardiopulmonary exercise test, Duke activity score index and patient reported health score questionnaires pre-operatively and in the 7th and 14th week post-operatively. Anaerobic threshold, peak oxygen uptake and ventilatory equivalents were measured in relation to activity and reported health scores. Haemoglobin and renal function was recorded. Results Recipients showed impaired cardiopulmonary function compared to donors with lower anaerobic threshold (10.5 vs. 14.4 ml/kg/min) and peak oxygen uptake (18.5 vs 23.0 ml/kg/min). Post-operatively the anaerobic threshold of recipients improved and normalised by the 14th week, whereas that in donors fell by ∼20% by the 7th (mean 11.4 ml/kg/min), recovering by the 14th (mean 15.6 ml/kg/min). Reported health but not activity scores showed similar changes. Conclusions Recovery following renal transplantation and donation differ. Transplantation improves renal function resulting in an increase in anaerobic threshold and peak oxygen uptake which essentially normalise by the 14th week post-operatively. Donors suffer a 20% reduction in cardiopulmonary reserve post-operatively, which recovers by the 14th week, suggesting no associated chronic exercise intolerance. IMPLICATIONS FOR REHABILITATION Cardiopulmonary exercise testing is a real-time predictor of functional capacity and thus is used as a pre-operative tool to measure physiological fitness and predict outcomes. Renal failure is associated with exercise intolerance and transplantation is transformational in terms of quality of life, longevity and healthcare cost. Live – related renal donation is increasingly available but whether donation itself carries a long-term health burden has not been previously well established. This study suggests that renal donation is not associated with long-term cardiopulmonary compromise and patients who donate their kidneys recover their previous fitness within 14 weeks.

Forecasting arrivals and occupancy levels in an emergency department

This is a sequel to Whitt and Zhang (2017), in which we developed an aggregate stochastic model of an emergency department (ED) based on the publicly available data from the large 1000-bed Rambam Hospital in Haifa, Israel, from 2004–7, associated with the patient flow analysis by Armony et al. (2015). Here we focus on forecasting future daily arrival totals and predicting hourly occupancy levels in real time, given recent history (previous arrival and departure times of all patients) and useful exogenous variables. For the arrival forecasting, we divide the dataset into an initial training set for fitting the models and a final test set to evaluate the performance. By using 200 weeks of data instead of the previous 25, we identify (i) long-term trends in both the arrival process and the length-of-stay distributions and (ii) dependence among successive daily arrival totals, which were undetectable before. From several forecasting methods, including artificial neural network models, we find that a seasonal autoregressive integrated moving average with exogenous (holiday and temperature) regressors (SARIMAX) time-series model is most effective. We then combine our previous ED model with the arrival prediction to create a real-time predictor for the future ED occupancy levels.

Time-domain simulator for short-term ship manoeuvring prediction: development and applications

ABSTRACTThis paper presents a short-term time-domain algorithm for ship motion simulation. Plane motion was considered, and thus, a three degrees of freedom model was used. The main inputs of the model are: propellers speed, rudder angles and initial conditions (initial position/orientation and linear/rotational velocity). The outputs are final ship position and orientation after a given simulation time. The data available in the Voyage Data Recorder and the results of sea trials performed before the ship delivery were used to estimate the needed parameters, implementing a simple but reliable linear regression scheme. The proposed model was applied to Costa Concordia manoeuvring simulation in order to assess its reliability in terms of trajectory prediction. Finally, as an example of application, the simulator was implemented in a real-time manoeuvring predictor, which is capable of foreseeing the ship position up to 40 s ahead in time, supporting the Commander in emergency operations.

Estimating probability of insemination success using milk progesterone measurements

The aim of this study was to quantify the effects of progesterone profile features and other cow-level factors on insemination success to provide a real-time predictor equation of probability of insemination success. Progesterone profiles from 26 dairy herds were analyzed and the effects of profile features (progesterone slope, cycle length, and cycle height) and cow traits (milk yield, parity, insemination during the previous estrus) on likelihood of artificial insemination success were estimated. The equation was fitted on a training data set containing data from 16 herds (6,246 estrous cycles from 3,404 lactations). The equation was tested on a testing data set containing data from 10 herds (8,105 estrous cycles from 3,038 lactations). Predictors were selected to be implemented in the final equation if adding them to a base model correcting for timing of insemination and parity decreased the overall likelihood distance of the model. Selected variables (cycle length, milk yield, cycle height, and insemination during the previous estrus) were used to build the final model using a stepwise approach. Predictors were added 1 by 1 in different order, and the model that had the smallest likelihood distance was selected. The final equation included the variables timing of insemination, parity, milk yield, cycle length, cycle height, and insemination during the previous estrus, respectively. The final model was applied to the testing data set and area under the curve (AUC) was calculated. On the testing data set, the final model had an AUC of 58%. When the farm effect was taken into account, the AUC increased to 63%. This equation can be implemented on farms that monitor progesterone and can support the farmer in deciding when to inseminate a cow. This can be the first step in moving the focus away from the current paradigm associated with poorer estrus detection, where each detected estrus is automatically inseminated, to near perfect estrus detection, where the question is which estrous cycle is worth inseminating?

Semi-Automated Needle Steering in Biological Tissue Using an Ultrasound-Based Deflection Predictor.

The performance of needle-based interventions depends on the accuracy of needle tip positioning. Here, a novel needle steering strategy is proposed that enhances accuracy of needle steering. In our approach the surgeon is in charge of needle insertion to ensure the safety of operation, while the needle tip bevel location is robotically controlled to minimize the targeting error. The system has two main components: (1) a real-time predictor for estimating future needle deflection as it is steered inside soft tissue, and (2) an online motion planner that calculates control decisions and steers the needle toward the target by iterative optimization of the needle deflection predictions. The predictor uses the ultrasound-based curvature information to estimate the needle deflection. Given the specification of anatomical obstacles and a target from preoperative images, the motion planner uses the deflection predictions to estimate control actions, i.e., the depth(s) at which the needle should be rotated to reach the target. Ex-vivo needle insertions are performed with and without obstacle to validate our approach. The results demonstrate the needle steering strategy guides the needle to the targets with a maximum error of 1.22mm.

Geographically distributed real-time digital simulations using linear prediction

Real-time (RT) simulator is a powerful tool for analyzing operational and control algorithms in electric power systems engineering. For understanding the dynamic and transient behavior of a power systems, significant RT computation capabilities are essential. A single unit of RT simulator has limited simulation capabilities. The most common way of augmenting simulation capability is using a bank of locally connected RT simulators. However, creating a large-sized bank of RT simulators involves significant financial investments and hence may not be feasible at all research facilities. Power and energy systems research facilities that use RT simulators are at diverse physical locations. In addition to RT simulators, research facilities around the world house an array of facilities with unique power, energy, and control systems for innovative research. To leverage these unique research facilities, geographically distributed RT simulation based on Wide Area Network (WAN) is required. Typical RT simulators perform simulations with time-steps in the order of milliseconds to microseconds, whereas data latency for communication on WAN may be as high as a few hundred milliseconds. Such communication latency between RT simulators may lead to inaccuracies and instabilities in geographically distributed RT simulations. In this paper, the effect of communication latency on geographically distributed RT simulation is discussed and analyzed. In order to reduce the effect of the communication latency, a Real-Time Predictor (RTP), based on linear curve fitting is developed and integrated into the distributed RT simulation environment. Two geographically distributed digital RT simulators are used to perform dynamic simulations of an electric power system with a fixed communication latency and the predictor. Empirical results demonstrate the effects of communication latency on the simulation and the performance of the RTP to improve the accuracy of simulations.

Adherent Use of Digital Health Trackers Is Associated with Weight Loss.

We study the association between weight fluctuation and activity tracking in an on-line population of thousands of individuals using digital health trackers (1,749 ≤ N ≤ 14,411, depending on the activity tracker considered) with millions of recorded activities (119,292 ≤ N ≤ 2,221,382) over the years 2013–2015. In a first between-subject analysis, we found a positive association between activity tracking frequency and weight loss. Users who log food with moderate frequency lost an additional 0.63% (CI [0.55, 0.72]; p < .001) of their body weight per month relative to low frequency loggers. Frequent workout loggers lost an additional 0.38% (CI [0.20, 0.56]; p < .001) and frequent weight loggers lost an additional 0.40% (CI [0.33, 0.47]; p < .001) as compared to infrequent loggers. In a subsequent within-subject analysis on a subset of the population (799 ≤ N ≤ 6,052) with sufficient longitudinal data, we used fixed effect models to explore the temporal relationship between a change in tracking adherence and weight change. We found that for the same individual, weight loss is significantly higher during periods of high adherence to tracking vs. periods of low adherence: +2.74% of body weight lost per month (CI [2.68, 2.81]; p < .001) during adherent weight tracking, +1.35% per month (CI [1.26, 1.43]; p < .001) during adherent food tracking, and +0.60% per month (CI [0.44, 0.76]; p < .001) during adherent workout tracking. The findings suggest that adherence to activity tracking can be utilized as a convenient real-time predictor of weight fluctuations, enabling large-scale, personalized intervention strategies.

Trending autoregulatory indices during treatment for traumatic brain injury.

Our goal is to use automatic data monitoring for reliable prediction of episodes of intracranial hypertension in patients with traumatic brain injury. Here we test the validity of our method on retrospective patient data. We developed the Continuous Hemodynamic Autoregulatory Monitor (CHARM), that siphons and stores signals from existing monitors in the surgical intensive care unit (SICU), efficiently compresses them, and standardizes the search for statistical relationships between any proposed index and adverse events. CHARM uses an automated event detector to reliably locate episodes of elevated intracranial pressure (ICP), while eliminating artifacts within retrospective patient data. A graphical user interface allowed data scanning, selection of criteria for events, and calculating indices. The pressure reactivity index (PRx), defined as the least square linear regression slope of intracranial pressure versus arterial BP, was calculated for a single case that spanned 259h. CHARM collected continuous records of ABP, ICP, ECG, SpO2, and ventilation from 29 patients with TBI over an 18-month period. Analysis of a single patient showed that PRx data distribution in the single hours immediately prior to all 16 intracranial hypertensive events, significantly differed from that in the 243h that did not precede such events (p<0.0001). The PRx index, however, lacked sufficient resolution as a real-time predictor of IH in this patient. CHARM streamlines the search for reliable predictors of intracranial hypertension. We report statistical evidence supporting the predictive potential of the pressure reactivity index.

Credit Indicators as Predictors of Economic Activity: A Real‐Time VAR Analysis

Using readily available indicators of the profitability, price, and availability of credit—the term spread, junk‐bond spread, and banks’ “willingness to lend” as reported by the Federal Reserve—we show that it is possible to significantly improve on the real‐time output and employment predictions of forecasting professionals at the medium‐run horizons that are most relevant to policymakers and private decision makers. Key to this improvement is a flexible state–space model of data revisions. The willingness‐to‐lend variable is the best real‐time predictor of GDP growth. For forecasting job growth, all three credit indicators prove helpful.

Integration and Validation of a Disruption Predictor Simulator in JET

Disruptions in tokamak devices are inevitable and can severely damage a tokamak device’s wall. For this reason, different protection mechanisms have to be implemented. In the Joint European Torus (JET), these protection systems are structured in different levels. At the lowest level are those systems that are responsible for protecting the machine’s integrity, which must be highly reliable. More complex systems are located at higher levels; these higher-level systems have been designed to take action before low-level systems. Since the installation of the new metallic wall in JET, new protection systems have been being developed to improve the overall protection of the device. This work focuses on a software application – a disruption predictor – that detects an incoming disruption. This software application simulates the behavior of a real-time implementation.In recent years, efforts have been devoted to developing and optimizing a reliable system that is capable of predicting disruptions. This has been accomplished by the novel combination of machine-learning techniques based on supervised learning methods. Disruptions must be predicted early enough so that the protection systems can mitigate the effects of disruptions. This paper summarizes the software development of the JET disruption predictor. This software simulates the real-time data acquisition and data processing. It has been an essential software tool to both optimize the disruption prediction model and implement a simulator of the real-time predictor.

Real-Time, Model-Based Spray-Cooling Control System for Steel Continuous Casting

This article presents a new system to control secondary cooling water sprays in continuous casting of thin steel slabs (CONONLINE). It uses real-time numerical simulation of heat transfer and solidification within the strand as a software sensor in place of unreliable temperature measurements. The one-dimensional finite-difference model, CON1D, is adapted to create the real-time predictor of the slab temperature and solidification state. During operation, the model is updated with data collected by the caster automation systems. A decentralized controller configuration based on a bank of proportional-integral controllers with antiwindup is developed to maintain the shell surface-temperature profile at a desired set point. A new method of set-point generation is proposed to account for measured mold heat flux variations. A user-friendly monitor visualizes the results and accepts set-point changes from the caster operator. Example simulations demonstrate how a significantly better shell surface-temperature control is achieved.

An advanced disruption predictor for JET tested in a simulated real-time environment

Disruptions are sudden and unavoidable losses of confinement that may put at risk the integrity of a tokamak. However, the physical phenomena leading to disruptions are very complex and non-linear and therefore no satisfactory model has been devised so far either for their avoidance or their prediction. For this reason, machine learning techniques have been extensively pursued in the last years. In this paper a real-time predictor specifically developed for JET and based on support vector machines is presented. The main aim of the present investigation is to obtain high recognition rates in a real-time simulated environment. To this end the predictor has been tested on the time slices of entire discharges exactly as in real world operation.Since the year 2000, the experiments at JET have been organized in campaigns named sequentially beginning with campaign C1. In this paper results from campaign C1 (year 2000) and up to C19 (year 2007) are reported. The predictor has been trained with data from JET's campaigns up to C7 with particular attention to reducing the number of missed alarms, which are less than 1%, for a test set of discharges from the same campaigns used for the training. The false alarms plus premature alarms are of the order of 6.4%, for a total success rate of more than 92%. The robustness of the predictor has been proven by testing it with a wide subset of shots of more recent campaigns (from C8 to C19) without any retraining. The success rate over the period between C8 and C14 is on average 88% and never falls below 82%, confirming the good generalization capabilities of the developed technique. After C14, significant modifications were implemented on JET and its diagnostics and consequently the success rates of the predictor between C15 and C19 decays to an average of 79%. Finally, the performance of the developed detection system has been compared with the predictions of the JET protection system (JPS). The new predictor clearly outperforms JPS up to about 180 ms before the disruptions.

An adaptive predictor for dynamic system forecasting

A reliable and real-time predictor is very useful to a wide array of industries to forecast the behaviour of dynamic systems. In this paper, an adaptive predictor is developed based on the neuro-fuzzy approach to dynamic system forecasting. An adaptive training technique is proposed to improve forecasting performance, accommodate different operation conditions, and prevent possible trapping due to local minima. The viability of the developed predictor is evaluated by using both gear system condition monitoring and material fatigue testing. The investigation results show that the developed adaptive predictor is a reliable and robust forecasting tool. It can capture the system's dynamic behaviour quickly and track the system's characteristics accurately. Its performance is superior to other classical forecasting schemes.