Search

Real-world multi-objective optimization problems encounter different types of uncertainty that may affect the quality of solutions. One common type is the stochastic noise that contaminates the objective functions. Another type of uncertainty is the different forms of dynamism including changes in the objective functions. Although related work in the literature targets only a single type, in this paper, we study Dynamic Multi-objective Optimization problems (DMOPs) contaminated with stochastic noises by dealing with the two types of uncertainty simultaneously. In such problems, handling uncertainty becomes a critical issue since the evolutionary process should be able to distinguish between changes that come from noise and real environmental changes that resulted from different forms of dynamism. To study both noisy and dynamic environments, we propose a flexible mechanism to incorporate noise into the DMOPs. Two novel techniques called Multi-Sensor Detection Mechanism (MSD) and Welford-Based Detection Mechanism (WBD) are proposed to differentiate between real change points and noise points. The proposed techniques are incorporated into a set of Dynamic Multi-objective Evolutionary Algorithms (DMOEAs) to analyze their impact. Our empirical study reveals the effectiveness of the proposed techniques for isolating noise from real dynamic changes and diminishing the noise effect on performance.

Background: Unplanned hospital readmissions are serious medical adverse events, stressful to patients, and expensive for hospitals. This study aims to develop a probability calculator to predict unplanned readmissions (PURE) within 30-days after discharge from the department of Urology, and evaluate the respective diagnostic performance characteristics of the PURE probability calculator developed with machine learning (ML) algorithms comparing regression versus classification algorithms. Methods: Eight ML models (i.e. logistic regression, LASSO regression, RIDGE regression, decision tree, bagged trees, boosted trees, XGBoost trees, RandomForest) were trained on 5.323 unique patients with 52 different features, and evaluated on diagnostic performance of PURE within 30 days of discharge from the department of Urology. Results: Our main findings were that performances from classification to regression algorithms had good AUC scores (0.62–0.82), and classification algorithms showed a stronger overall performance as compared to models trained with regression algorithms. Tuning the best model, XGBoost, resulted in an accuracy of 0.83, sensitivity of 0.86, specificity of 0.57, AUC of 0.81, PPV of 0.95, and a NPV of 0.31. Conclusions: Classification models showed stronger performance than regression models with reliable prediction for patients with high probability of readmission, and should be considered as first choice. The tuned XGBoost model shows performance that indicates safe clinical appliance for discharge management in order to prevent an unplanned readmission at the department of Urology.

Drug–target interaction is crucial in the discovery of new drugs. Computational methods can be used to identify new drug–target interactions at low costs and with reasonable accuracy. Recent studies pay more attention to machine-learning methods, ranging from matrix factorization to deep learning, in the DTI prediction. Since the interaction matrix is often extremely sparse, DTI prediction performance is significantly decreased with matrix factorization-based methods. Therefore, some matrix factorization methods utilize side information to address both the sparsity issue of the interaction matrix and the cold-start issue. By combining matrix factorization and autoencoders, we propose a hybrid DTI prediction model that simultaneously learn the hidden factors of drugs and targets from their side information and interaction matrix. The proposed method is composed of two steps: the pre-processing of the interaction matrix, and the hybrid model. We leverage the similarity matrices of both drugs and targets to address the sparsity problem of the interaction matrix. The comparison of our approach against other algorithms on the same reference datasets has shown good results regarding area under receiver operating characteristic curve and the area under precision–recall curve. More specifically, experimental results achieve high accuracy on golden standard datasets (e.g., Nuclear Receptors, GPCRs, Ion Channels, and Enzymes) when performed with five repetitions of tenfold cross-validation. Graphical abstract: [Figure not available: see fulltext.]Display graphical of the hybrid model of Matrix Factorization with Denoising Autoencoderswith the help side information of drugs and targets for Prediction of Drug-Target Interactions.

Semi-online models where decisions may be revoked in a limited way have been studied extensively in the last years. This is motivated by the fact that the pure online model is often too restrictive to model real-world applications, where some changes might be allowed. A well-studied measure of the amount of decisions that can be revoked is the migration factor $\beta$: When an object $o$ of size $s(o)$ arrives, the decisions for objects of total size at most $\beta\cdot s(o)$ may be revoked. Usually $\beta$ should be a constant. This means that a small object only leads to small changes. This measure has been successfully investigated for different, classic problems such as bin packing or makespan minimization. The dual of makespan minimization - the Santa Claus or machine covering problem - has also been studied, whereas the dual of bin packing - the bin covering problem - has not been looked at from such a perspective. In this work, we extensively study the bin covering problem with migration in different scenarios. We develop algorithms both for the static case - where only insertions are allowed - and for the dynamic case, where items may also depart. We also develop lower bounds for these scenarios both for amortized migration and for worst-case migration showing that our algorithms have nearly optimal migration factor and asymptotic competitive ratio (up to an arbitrary small $\eps$). We therefore resolve the competitiveness of the bin covering problem with migration.

Because of the billions of transactions that take place every day, the volumes of information is always growing. There are a plethora of categorization techniques accessible for extracting usable information from the massive quantity of information obtained. Swarm optimization methods, as well as hybridizations of these techniques, are now playing a significant role in classifications, and they do so in a very efficient way. An introduction and thorough comparison of many swarm optimization techniques & hybrid swarm optimization techniques that have been published in the academic journals are presented in this article. Bio-inspired computing is a fascinating field of machine learning that investigates how natural occurrences may serve as a rich source of motivation for the development of clever processes that can be turned into strong algorithms in the future. In categorization, prediction, & optimization issues, several of these methods have been utilised effectively. In the field of optimization, swarm intelligence techniques are a kind of microbially algorithm that has been proven to be very effective for quite some time. However, in order for these algorithms to function at their peak levels, the starting variables must be adjusted correctly by a skilled user who knows what they're doing. The development and expansion of machine management are aided greatly by the use of efficient machinery and equipment. Different research investigations are engaged in the execution of the study and reviewers have concluded that machine learning technologies are the most effective means of supporting this expansion of the research. In order to better understand machine learning technologies and machine learning algorithms, the majority of implementations are investigated using swarm intelligence optimization techniques. As a result, they may be used as an essential judgement tool in the manufacturing industry.

Artificial intelligence (AI) focuses on processing and interpreting complex information as well as identifying relationships and patterns among complex data. Artificial intelligence- and machine learning (ML)-driven predictions have shown promising potential in influencing real-time decisions and improving surgical outcomes by facilitating screening, diagnosis, risk assessment, preoperative planning, and shared decision-making. Fundamental understanding of the algorithms, as well as their development and interpretation, is essential for the evolution of AI in surgery. In this article, we provide surgeons with a fundamental understanding of AI-driven predictive models through an overview of common ML and deep learning algorithms, model development, performance metrics and interpretation. This would serve as a basis for understanding ML-based research, while fostering new ideas and innovations for furthering the reach of this emerging discipline.

MetaMorpheus is a free and open-source software program dedicated to the comprehensive analysis of proteomic data. In bottom-up proteomics, protein samples are digested into peptides prior to chromatographic separation and tandem mass spectrometric analysis. The resulting fragmentation spectra are subsequently analyzed with search software programs to obtain peptide identifications and infer the presence of proteins in the samples. MetaMorpheus seeks to maximize the information gleaned from proteomic data through the use of (a) mass calibration, (b) post-translational modification discovery, (c) multiple search algorithms, which aid in the analysis of data from traditional, crosslinking, and glycoproteomic experiments, (d) isotope-based or label-free quantification, (e) multi-protease protein inference, and (f) spectral annotation and data visualization capabilities. This protocol provides detailed descriptions of how use MetaMorpheus and how to customize data analysis workflows using MetaMorpheus tasks to meet the specific needs of the user.

Contours traced by trained phoneticians have been considered to be the most accurate way to identify the midsagittal tongue surface from ultrasound video frames. In this study, inter-measurer reliability was evaluated using measures that quantified both how closely human-placed contours approximated each other as well as how consistent measurers were in defining the start and end points of contours. High reliability across three measurers was found for all measures, consistent with treating contours placed by trained phoneticians as the ‘gold standard.’ However, due to the labour-intensive nature of hand-placing contours, automatic algorithms that detect the tongue surface are increasingly being used to extract tongue-surface data from ultrasound videos. Contours placed by six automatic algorithms (SLURP, EdgeTrak, EPCS, and three different configurations of the algorithm provided in Articulate Assistant Advanced) were compared to human-placed contours, with the same measures used to evaluate the consistency of the trained phoneticians. We found that contours defined by SLURP, EdgeTrak, and two of the AAA configurations closely matched the hand-placed contours along sections of the image where the algorithms and humans agreed that there was a discernible contour. All of the algorithms were much less reliable than humans in determining the anterior (tongue-tip) edge of tongue contours. Overall, the contours produced by SLURP, EdgeTrak, and AAA should be useable in a variety of clinical applications, subject to spot-checking. Additional practical considerations of these algorithms are also discussed.

Funding Information: This work is partially supported by EIT Urban Mobility. The funders had no role in study design, data collection, analysis, decision to publish, or preparation of the manuscript. Publisher Copyright: © 2023, The Author(s). In an attempt to mitigate emissions and road traffic, a significant interest has been recently noted in expanding the use of shared vehicles to replace private modes of transport. However, one outstanding issue has been the hesitancy of passengers to use shared vehicles due to the substandard levels of interior cleanliness, as a result of leftover items from previous users. The current research focuses on developing a novel prediction model using computer vision capable of detecting various types of trash and valuables from a vehicle interior in a timely manner to enhance ambience and passenger comfort. The interior state is captured by a stationary wide-angled camera unit located above the seating area. The acquired images are preprocessed to remove unwanted areas and subjected to a convolutional neural network (CNN) capable of predicting the type and location of leftover items. The algorithm was validated using data collected from two research vehicles under varying conditions of light and shadow levels. The experiments yielded an accuracy of 89% over distinct classes of leftover items and an accuracy of 91% among the general classes of trash and valuables. The average execution time was 65 s from image acquisition in the vehicle to displaying the results in a remote server. A custom dataset of 1379 raw images was also made publicly available for future development work. Additionally, an indoor air quality (IAQ) unit capable of detecting specific air pollutants inside the vehicle was implemented. Based on the pilots conducted for air quality monitoring within the vehicle cabin, an IAQ index was derived which corresponded to a 6-level scale in which each level was associated with the explicit state of interior odour. Future work will focus on integrating the two systems (item detection and air quality monitoring) explicitly to produce a discrete level of cleanliness. The current dataset will also be expanded by collecting data from real shared vehicles in operation.

Sympathovagal balance is important in the pathogenesis of hypertension and independently associated with mortality. We evaluated the value of automated analysis of cross-correlation baroreflex sensitivity (xBRS) and heart rate variability (HRV) and its relationship with clinical covariates in 13,326 participants from the multi-ethnic HELIUS study. Finger blood pressure (BP) was continuously recorded, from which xBRS, standard deviation of normal-to-normal intervals (SDNN), and squared root of mean squared successive difference between normal-to-normal intervals (RMSDD) were determined. A subset of 3356 recordings > 300 s was used to derive the minimally required duration by comparing shortened to complete recordings, defined as intraclass correlation (ICC) > 0.90. For xBRS and SDNN, 120 s and 180 s were required (ICC 0.93); for RMSDD, 60 s (ICC 0.94) was sufficient. We included 10,252 participants (median age 46 years, 54% women) with a recording > 180 s for the regression. xBRS, SDNN, and RMSDD decreased linearly up to 50 years of age. For xBRS, there was a signification interaction with sex, with for every 10 years a decrease of 4.3 ms/mmHg (95%CI 4.0–4.6) for men and 5.9 ms/mmHg (95%CI 5.6–6.1) for women. Using splines, we observed sex-dependent nonlinearities in the relation with BP, waist-to-hip-ratio, and body mass index. Future studies can help unravel the dynamics of these relations and assess their predictive value. Graphical Abstract Panel 1 depicts automatic analysis and filtering of finger BP recordings, panel 2 depicts computation of xBRS from interpolated beat to beat data of systolic BP and interbeat interval, and (IBI) SDNN and RMSDD are computed directly from the filtered IBI dataset. Panel 3 depicts the results of large-scale analysis and relation of xBRS with age, sex, blood pressure and body mass index.

Patients with chronic obstructive pulmonary disease (COPD) may suffer from acute episodes of worsening dyspnea, often associated with increased cough, sputum, and/or sputum purulence. These exacerbations of COPD (ECOPDs) impact health status, accelerate lung function decline, and increase the risk of hospitalization. Importantly, close to 20% of patients are readmitted within 30 days after hospital discharge, with great cost to the person and society. Approximately 25% and 65% of patients hospitalized for an ECOPD die within 1 and 5 years, respectively. Patients with COPD are usually older and frequently have concomitant chronic diseases, including heart failure, coronary artery disease, arrhythmias, interstitial lung diseases, bronchiectasis, asthma, anxiety, and depression, and are also at increased risk of developing pneumonia, pulmonary embolism, and pneumothorax. All of these morbidities not only increase the risk of subsequent ECOPDs but can also mimic or aggravate them. Importantly, close to 70% of readmissions after an ECOPD hospitalization result from decompensation of other morbidities. These observations suggest that in patients with COPD with worsening dyspnea but without the other classic characteristics of ECOPD, a careful search for these morbidities can help detect them and allow appropriate treatment. For most morbidities, a thorough clinical evaluation supplemented by appropriate clinical investigations can guide the healthcare provider to make a precise diagnosis. This perspective integrates the currently dispersed information available and provides a practical approach to patients with COPD complaining of worsening respiratory symptoms, particularly dyspnea. A systematic approach should help improve outcomes and the personal and societal cost of ECOPDs.

In the present paper, a new strategy of multigrid method is introduced to accelerate the convergence speed of numerical simulations via lattice Boltzmann method. Based on the popular V-cycle multigrid algorithm, a simplified multigrid algorithm is presented and validated through the simulations of the classic lid-driven cavity flow for steady states. The novelty of the present algorithm resides in the construction of the information transferring process, in which, for a full cycle, the numerical simulation starts on the coarse mesh, transferring to medium mesh and then streams to the fine mesh through a prolongation operator. Afterwards, instead of using a restriction operator, the fluid information jumps back directly from the fine mesh to the coarse mesh via an assignment operator. The accuracy and efficiency of the simplified new algorithm are validated by comparing the results obtained when employing the classic V-cycle multigrid algorithm and the traditional lattice Boltzmann method with uniform Cartesian grid This research was founded by the project provided by the Northwestern Polytechnical University (G2021KY05103). This work was also sponsored by the foundation of National Key Laboratory of Science and Technology on Aerodynamic Design and Research (No. 614220121030101)

Aims Screening for atrial fibrillation (AF) is recommended in the European Society of Cardiology guidelines. Yields of detection can be low due to the paroxysmal nature of the disease. Prolonged heart rhythm monitoring might be needed to increase yield but can be cumbersome and expensive. The aim of this study was to observe the accuracy of an artificial intelligence (AI)-based network to predict paroxysmal AF from a normal sinus rhythm single-lead ECG. Methods and results A convolutional neural network model was trained and evaluated using data from three AF screening studies. A total of 478 963 single-lead ECGs from 14 831 patients aged ≥65 years were included in the analysis. The training set included ECGs from 80% of participants in SAFER and STROKESTOP II. The remaining ECGs from 20% of participants in SAFER and STROKESTOP II together with all participants in STROKESTOP I were included in the test set. The accuracy was estimated using the area under the receiver operating characteristic curve (AUC). From a single timepoint ECG, the artificial intelligence–based algorithm predicted paroxysmal AF in the SAFER study with an AUC of 0.80 [confidence interval (CI) 0.78–0.83], which had a wide age range of 65–90+ years. Performance was lower in the age-homogenous groups in STROKESTOP I and STROKESTOP II (age range: 75–76 years), with AUCs of 0.62 (CI 0.61–0.64) and 0.62 (CI 0.58–0.65), respectively. Conclusion An artificial intelligence–enabled network has the ability to predict AF from a sinus rhythm single-lead ECG. Performance improves with a wider age distribution.

Medical ultrasound imaging relies heavily on high-quality signal processing to provide reliable and interpretable image reconstructions. Conventionally, reconstruction algorithms have been derived from physical principles. These algorithms rely on assumptions and approximations of the underlying measurement model, limiting image quality in settings where these assumptions break down. Conversely, more sophisticated solutions based on statistical modeling or careful parameter tuning or derived from increased model complexity can be sensitive to different environments. Recently, deep learning-based methods, which are optimized in a data-driven fashion, have gained popularity. These model-agnostic techniques often rely on generic model structures and require vast training data to converge to a robust solution. A relatively new paradigm combines the power of the two: leveraging data-driven deep learning and exploiting domain knowledge. These model-based solutions yield high robustness and require fewer parameters and training data than conventional neural networks. In this work we provide an overview of these techniques from the recent literature and discuss a wide variety of ultrasound applications. We aim to inspire the reader to perform further research in this area and to address the opportunities within the field of ultrasound signal processing. We conclude with a future perspective on model-based deep learning techniques for medical ultrasound.

Fault tree analysis (FTA) is a well-established technique to analyze the safety risks of a system. Two specific prominent FTA methods, largely applied in the aerospace field, are the so-called minimal cut sets (MCS), which uses an approximate evaluation of the problem, and the direct evaluation (DE) of the fault tree, which uses a top-down recursive algorithm. The first approach is only valid for small values of basic event probabilities and has historically yielded faster results than exact solutions for complex fault trees. The second one means exact solutions at a higher computational cost. This article presents several improvements applied to both approaches in order to upgrade the computing performance. First, improvements to the MCS approach have been performed, where the main idea has been to optimize the number of required permutations and to take advantage of the available information from previous solved subsets. Second, improvements to the DE approach have been applied, which deal with a reduction of the number of recursive calls through a deep search for independent events in the fault tree. This could dramatically reduce the computation time for industrial fault trees with a high number of repeated events. Additional implementation improvements have been also applied regarding hash tables, and memory access and usage, but also implementing the so-called “virtual gates”, which enable limitless children on each gate. The results presented hereafter are promising, not only because they show that both approaches have been highly optimized compared to the literature, but also because a DE solution has been achieved, which can compete in time resources (and obviously in precision) with the MCS approach. These improvements are relevant when considering the industrial, and more specifically the aeronautical, implementation and application of both techniques. The author Jordi Pons‐Prats acknowledges the support from Serra Hunter programme, Generalitat de Catalunya, as well as the support through the Severo Ochoa Centre of Excellence (2019‐2023) under the grant CEX2018‐000797‐S funded by MCIN/AEI/10.13039/501100011033.

Deep learning-enabled smartphone-based image processing has significant advantages in the development of point-of-care diagnostics. Conventionally, most deep-learning applications require task specific large scale expertly annotated datasets. Therefore, these algorithms are oftentimes limited only to applications that have large retrospective datasets available for network development. Here, we report the possibility of utilizing adversarial neural networks to overcome this challenge by expanding the utility of non-specific data for the development of deep learning models. As a clinical model, we report the detection of fentanyl, a small molecular weight drug that is a type of opioid, at the point-of-care using a deep-learning empowered smartphone assay. We used the catalytic property of platinum nanoparticles (PtNPs) in a smartphone-enabled microchip bubbling assay to achieve high analytical sensitivity (detecting fentanyl at concentrations as low as 0.23 ng mL

High-resolution functional MRI (fMRI) is largely hindered by random thermal noise. Random matrix theory (RMT)-based principal component analysis (PCA) is promising to reduce such noise in fMRI data. However, there is no consensus about the optimal strategy and practice in implementation. In this work, we propose a comprehensive RMT-based denoising method that consists of 1) rank and noise estimation based on a set of newly derived multiple criteria, and 2) optimal singular value shrinkage, with each module explained and implemented based on the RMT. By incorporating the variance stabilizing approach, the denoising method can deal with low signal-to-noise ratio (SNR) (such as5) magnitude fMRI data with favorable performance compared to other state-of-the-art methods. Results from both simulation and in-vivo high-resolution fMRI data show that the proposed denoising method dramatically improves image restoration quality, promoting functional sensitivity at the same level of functional mapping blurring compared to existing denoising methods. Moreover, the denoising method can serve as a drop-in step in data preprocessing pipelines along with other procedures aimed at removal of structured physiological noises. We expect that the proposed denoising method will play an important role in leveraging high-quality, high-resolution task fMRI, which is desirable in many neuroscience and clinical applications.

Human brain is a complex yet economically organized system, where a small portion of critical hub regions support the majority of brain functions. The identification of common hub nodes in a population of networks is often simplified as a voting procedure on the set of identified hub nodes across individual brain networks, which ignores the intrinsic data geometry and partially lacks the reproducible findings in neuroscience. Hence, we propose a first-ever group-wise hub identification method to identify hub nodes that are common across a population of individual brain networks. Specifically, the backbone of our method is to learn common graph embedding that can represent the majority of local topological profiles. By requiring orthogonality among the graph embedding vectors, each graph embedding as a data element is residing on the Grassmannian manifold. We present a novel Grassmannian manifold optimization scheme that allows us to find the common graph embeddings, which not only identify the most reliable hub nodes in each network but also yield a population-based common hub node map. Results of the accuracy and replicability on both synthetic and real network data show that the proposed manifold learning approach outperforms all hub identification methods employed in this evaluation.

Objectives. To develop an imputation method to produce estimates for suppressed values within a shared government administrative data set to facilitate accurate data sharing and statistical and spatial analyses. Methods. We developed an imputation approach that incorporated known features of suppressed Massachusetts surveillance data from 2011 to 2017 to predict missing values more precisely. Our methods for 35 de-identified opioid prescription data sets combined modified previous or next substitution followed by mean imputation and a count adjustment to estimate suppressed values before sharing. We modeled 4 methods and compared the results to baseline mean imputation. Results. We assessed performance by comparing root mean squared error (RMSE), mean absolute error (MAE), and proportional variance between imputed and suppressed values. Our method outperformed mean imputation; we retained 46% of the suppressed value’s proportional variance with better precision (22% lower RMSE and 26% lower MAE) than simple mean imputation. Conclusions. Our easy-to-implement imputation technique largely overcomes the adverse effects of low count value suppression with superior results to simple mean imputation. This novel method is generalizable to researchers sharing protected public health surveillance data. (Am J Public Health. 2021; 111(10):1830–1838. https://doi.org/10.2105/AJPH.2021.306432 )

To obtain observable physical or molecular properties such as ionization potential and fluorescent wavelength with quantum chemical (QC) computation, multi-step computation manipulated by a human is required. Hence, automating the multi-step computational process and making it a black box that can be handled by anybody are important for effective database construction and fast realistic material design through the framework of black-box optimization where machine learning algorithms are introduced as a predictor. Here, we propose a Python library, QCforever, to automate the computation of some molecular properties and chemical phenomena induced by molecules. This tool just requires a molecule file for providing its observable properties, automating the computation process of molecular properties (for ionization potential, fluorescence, etc.) and output analysis for providing their multi-values for evaluating a molecule. Incorporating the tool in black-box optimization, we can explore molecules that have properties we desired within the limitation of QC computation.

With the increasing use of wearable healthcare devices for remote patient monitoring, reliable signal quality assessment (SQA) is required to ensure the high accuracy of interpretation and diagnosis on the recorded data from patients. Photoplethysmographic (PPG) signals non-invasively measured by wearable devices are extensively used to provide information about the cardiovascular system and its associated diseases. In this study, we propose an approach to optimize the quality assessment of the PPG signals.We used an ensemble-based feature selection scheme to enhance the prediction performance of the classification model to assess the quality of the PPG signals. Our approach for feature and subset size selection yielded the best-suited feature subset, which was optimized to differentiate between the clean and artifact corrupted PPG segments.A high discriminatory power was achieved between two classes on the test data by the proposed feature selection approach, which led to strong performance on all dependent and independent test datasets. We achieved accuracy, sensitivity, and specificity rates of higher than 0.93, 0.89, and 0.97, respectively, for dependent test datasets, independent of heartbeat type, i.e., atrial fibrillation (AF) or non-AF data including normal sinus rhythm (NSR), premature atrial contraction (PAC), and premature ventricular contraction (PVC). For independent test datasets, accuracy, sensitivity, and specificity rates were greater than 0.93, 0.89, and 0.97, respectively, on PPG data recorded from AF and non-AF subjects. These results were found to be more accurate than those of all of the contemporary methods cited in this work.As the results illustrate, the advantage of our proposed scheme is its robustness against dynamic variations in the PPG signal during long-term 14-day recordings accompanied with different types of physical activities and a diverse range of fluctuations and waveforms caused by different individual hemodynamic characteristics, and various types of recording devices. This robustness instills confidence in the application of the algorithm to various kinds of wearable devices as a reliable PPG signal quality assessment approach.

Spectral CT has shown promise for high-sensitivity quantitative imaging and material decomposition. This work presents a new device called a spatial-spectral filter (SSF) which consists of a tiled array of filter materials positioned near the x-ray source that is used to modulate the spectral shape of the x-ray beam. The filter is moved to obtain projection data that is sparse in each spectral channel. To process this sparse data, we employ a one-step direct model-based material decomposition (MBMD) to reconstruct basis material density images directly from the SSF CT data. To evaluate different possible SSF designs, we define a new Fisher-information-based predictive image quality metric called separability index which characterizes the ability of a spectral CT system to distinguish between the signals from two or more materials. This spectral CT performance metric can be used to optimize spectral CT system design. We conducted simulation-based design optimization study to find optimized combinations of filter materials, filter thicknesses, filter widths, and source settings. Finally, we present MBMD results using simulated SSF CT measurements from the optimized designs to demonstrate the ability to reconstruct basis material density images and to show the benefits of the optimized designs. Our results indicate that optimizing SSF CT for separability leads to high-performance at material discrimination tasks.

Hereditary hemochromatosis (HH) causes unbalanced iron deposition in many organs including the joints leading to severe cartilage loss and bone damage in the metacarpophalangeal joints (MCPJ). High-resolution peripheral quantitative computed tomography (HR-pQCT) and its joint space width (JSW) quantification algorithm quantifies in vivo 3D joint morphology. We therefore aimed to (i) determine feasibility and performance of the JSW algorithm in HH, (ii) quantify joint space morphology, and (iii) investigate the relationship between morphological and clinical parameters in HH. Here, we performed an exploratory study on 24 HH patients and sex- and age-matched controls using HR-pQCT imaging of MCPJ. Mineralized bone structure was automatically segmented from the grayscale image data and periosteal surface bone masks and joint space masks were generated. Mean, minimal, and maximal joint space width (JSW; JSW.MIN; JSW.MAX), JSW heterogeneity (JSW.SD), JSW asymmetry (JSW.AS), and joint space volume (JSV) were computed. Demographics and, for HH patients, disease-specific parameters were recorded. Segmentation of JS was very good with 79.7% of MCPJs successfully segmented at first attempt and 20.3% requiring semi-manual correction. HH men showed larger JSV at all MCPs (+ 25.4% p p p = 0.043) at MCP 4 relative to HH women. Compared to controls, both HH men and HH women showed larger JSW.AS and smaller JSW.MIN at all MCP levels, reaching significance for HH men at MCP 2 and 3 (JSW.AS: + 323% p p p = 0.025; JSW.MIN: − 41.8%, p = 0.022). Time since HH diagnosis was correlated positively with MCP 4 JSW.AS and JSW.SD (0.463 ρ p ρ p

In ultra-high-field (UHF) magnetic resonance imaging (MRI) applications, the design and analysis of high-frequency radio frequency (RF) coils requires full-wave electromagnetic (EM) methods that can handle complex field-tissue interactions. Using a Huygens' equivalent surface, the Method of Moments (MoM) and the Finite-Difference Time-Domain (FDTD) algorithm can be combined to accurately model the high-frequency RF coils. In previous research, a uniform FDTD mesh structure was considered, providing a compromised solution for coil-tissue interactions. This paper proposes a hybrid FDTD/MoM algorithm with non-uniform meshes. The fine mesh domain is set at the Huygens' surface, and the other domain uses coarse meshes. The proposed algorithms are strictly validated, and their computational performance is compared against conventional methods. Results show that the new algorithm can improve the calculation efficiency without losing accuracy. Specifically, compared with the uniform FDTD method, the numerical difference between both hybrid methods remains at 3.2%. Still, the calculation time of the non-uniform grid algorithm is reduced by 64.2%, demonstrating the effectiveness of the new algorithm for modeling RF coils for UHF-MRI applications.

IEEEWhile the zeitgeist of today prescribes localization as a key to eliminate the excessive transportation for manufacturing firms, firm-level practical and workable guidelines that go beyond maxims are scant. Laden with analytical methods, this paper offers a phronetic, customizable framework based on Lean Six Sigma (LSS). The framework mimics Define-Measure-Analyze-Improve-Control (DMAIC) process. The early phases of the framework, ex-ante, prescribe the use of a variety tools and techniques from Lean Six sigma (LSS) and employ fuzzy analytic hierarchy process (FAHP) to extract the vital few root-causes that may impede localization. The latter phases suggest the use of design of experiments (DOE) to evaluate the root-causes, ex-post. The paper outlines a clear and established procedure and demonstrates each analytical phase using evidence from a two-year real-world application in steel industry.

The time evolution of (bio)chemical processes, specifically those involving distributed species as in polymer synthesis and recycling, can be obtained using Gillespie-based kinetic Monte Carlo simulations, provided that a sufficiently high (Monte Carlo) control volume is utilized with respect to the simulation targets (e.g., focus on only conversion/yield or a combination of the former and average molar masses, or even the combination with complete distributions with accurate tail prediction). For more detailed kinetics and more demanding simulations targets, currently, simulation results are mostly visually checked to decide which control volume to practically use, taking into account user time constraints. The present work puts forward a convergence strategy that avoids relying on subjective visual analysis to set the minimum control value that guarantees the minimization of the stochastic noise for several simulation target combinations to acceptable values. The strategy is illustrated for two (for simplicity, intrinsic) non-dispersed phase bulk chain-growth polymerization processes, one with high average chain lengths and a broad chain length distribution (CLD), that is, free -radical polymerization (FRP), and one with low average chain lengths and a narrow CLD, that is, nitroxide-mediated polymerization (NMP). It is showcased that the monomer conversion profile converges the fastest, with even the occurrence of noise-free simulation results in the absence of numerical convergence. A sufficiently accurate representation of the tail of the (number) CLD in FRP demands a sufficiently high control volume so that relative errors below 0.5% result for the z-based average chain length or molar mass (Mz). This need to inspect Mz convergence further holds under NMP, in general, reversible deactivation radical polymerization (RDRP) conditions, in which it is uncommon to report such higher order averages. An automated convergence check beyond threshold values is recommended to minimize the impact of possible fluctuations in certain simulation targets, specifically peak representations, for example, the initial spike in the dispersity plot in RDRP. The convergence results are supported by the number of radicals in the control volume with values much higher than 2 to accurately represent termination kinetics for specifically CLD prediction, already under intrinsic conditions.

© 2023, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.Many prediction methods proposed in the literature can be concerned under two main headings: probabilistic and non-probabilistic methods. In particular, as a kind of non-probabilistic model, artificial neural networks (ANNs), having different properties, have been commonly and effectively used in the literature. Some ANNs operate the additive aggregation function in the structure of their neuron models, while others employ the multiplicative aggregation function. Recently proposed dendritic neural networks also have both additional and multiplicative neuron models. The prediction performance of such an artificial neural network will inevitably be negatively affected by the outliers that the time series of interest may contain due to the neuron model in its structure. This study, for the training of a dendritic neural network, presents a robust learning algorithm. The presented robust algorithm is the first for the training of DNM in the literature as far as is known and uses Huber\"s loss function as the fitness function. The iterative process of the robust learning algorithm is carried out by particle swarm optimization. The productivity and efficiency of the suggested learning algorithm were evaluated by analysing different real-life time series. All analyses were performed with original and contaminated data sets under different scenarios. The R-DNM has the best performance for the original data sets with a value of 2.95% in the ABC time series, while the FTSE showed the best performance in approximately 27% and the second best in 33% of all analyses. The proposed R-DNM has been the least affected by outliers in almost all scenarios for contaminated ABC data sets. Moreover, it has been the least affected model by outliers in approximately 71% of the 90 analyses performed for the contaminated FTSE time series. The obtained results show that the dendritic artificial neural network trained by the proposed robust learning algorithm produces the satisfactory predictive results in the analysis of time series with and without outliers.

Epidemic outbreaks, such as the one generated by the coronavirus disease, have raised the need for more efficient healthcare logistics. One of the challenges that many governments have to face in such scenarios is the deployment of temporary medical facilities across a region with the purpose of providing medical services to their citizens. This work tackles this temporary-facility location and queuing problem with the goals of minimizing costs, the expected completion time, population travel and waiting times. The completion time for a facility depends on the numbers assigned to those facilities as well as stochastic arrival times. This work proposes a learnheuristic algorithm to solve the facility location and population assignment problem. Firstly a machine learning algorithm is trained using data from a queuing model (simulation module). The learnheuristic then constructs solutions using the machine learning algorithm to rapidly evaluate decisions in terms of facility completion and population waiting times. The efficiency and quality of the algorithm is demonstrated by comparison with exact and simulation-only (simheuristic) methodologies. A series of experiments are performed which explore the trade offs between solution cost, completion time, population travel and waiting times.

article describes a method used to calibrate 3-D freehand ultrasound systems based on phantoms with parallel wires forming two perpendicular planes, such as the usual general-purpose commercial phantoms. In our algorithm, the phantom pose is co-optimized with the calibration to avoid the need to precisely track the phantom. We provide a geometrical analysis to explain the proposed acquisition protocol. Finally, we give an estimate of the system accuracy and precision based on measurements acquired on an independent test phantom. We obtained error norms of 1.6 mm up to 6 cm of depth and 3.5 mm between 6 and 14 cm of depth, in total average. In conclusion, it is possible to calibrate ultrasound tracked-probe systems with a reasonable accuracy based on a general-purpose phantom. Contrarily to most calibration methods that imply the construction of the phantom, the present algorithm is based on a standard phantom geometry that is commercially available. (E-mail: sandra. marcadent@epfl.ch) (c) 2022 World Federation for Ultrasound in Medicine & Biology. All rights reserved.

With the rapid integration of artificial intelligence into medical practice, there has been an exponential increase in the number of scientific papers and industry players offering models designed for various tasks. Understanding these, however, is difficult for a radiologist in practice, given the core mathematical principles and complicated terminology involved. This review aims to elucidate the core mathematical concepts of both machine learning and deep learning models, explaining the various steps and common terminology in common layman language. Thus, by the end of this article, the reader should be able to understand the basics of how prediction models are built and trained, including challenges faced and how to avoid them. The reader would also be equipped to adequately evaluate various models, and take a decision on whether a model is likely to perform adequately in the real-world setting.

Tomography imaging methods at synchrotron light sources keep evolving, pushing multi-modal characterization capabilities at high spatial and temporal resolutions. To achieve this goal, small probe size and multi-dimensional scanning schemes are utilized more often in the beamlines, leading to rising complexities and challenges in the experimental setup process. To avoid spending a significant amount of human effort and beam time on aligning the X-ray probe, sample and detector for data acquisition, most attention has been drawn to realigning the systems at the data processing stages. However, post-processing cannot correct everything, and is not time efficient. Here we present automatic alignment schemes of the rotational axis and sample pre- and during the data acquisition process using a software approach which combines the advantages of genetic algorithms and human intelligence. Our approach shows excellent sub-pixel alignment efficiency for both tasks in a short time, and therefore holds great potential for application in the data acquisition systems of future scanning tomography experiments.

There is an increased desire for miniature ultrasound probes with small apertures to provide volumetric images at high frame rates for in-body applications. Satisfying these increased requirements makes simultaneous achievement of a good lateral resolution a challenge. As micro-beamforming is often employed to reduce data rate and cable count to acceptable levels, receive processing methods that try to improve spatial resolution will have to compensate the introduced reduction in focusing. Existing beamformers do not realize sufficient improvement and/or have a computational cost that prohibits their use. Here we propose the use of adaptive beamforming by deep learning (ABLE) in combination with training targets generated by a large aperture array, which inherently has better lateral resolution. In addition, we modify ABLE to extend its receptive field across multiple voxels. We illustrate that this method improves lateral resolution both quantitatively and qualitatively, such that image quality is improved compared with that achieved by existing delay-and-sum, coherence factor, filtered-delay-multiplication-and-sum and Eigen-based minimum variance beamformers. We found that only in silica data are required to train the network, making the method easily implementable in practice.

For an arbitrary valued field $(K,v)$ and a given extension $v(K^*)\hookrightarrow\Lambda$ of ordered groups, we analyze the structure of the tree formed by all $\Lambda$-valued extensions of $v$ to the polynomial ring $K[x]$. As an application, we find a model for the tree of all equivalence classes of valuations on $K[x]$ (without fixing their value group), whose restriction to $K$ is equivalent to $v$. In the henselian case, we apply these results to show that there is a complete parallelism between the arithmetic properties of irreducible polynomials $F\in K[x]$, encoded by their Okutsu frames, and the valuation-theoretic properties of their induced valuations $v_F$ on $K[x]$, encoded by their MacLane-Vaqui\'e chains. This parallelism was only known for defectless irreducible polynomials., Comment: V2, April 2022

Sparse PCA methods are used to overcome the difficulty of interpreting the solution obtained from PCA. However, constraining PCA to obtain sparse solutions is an intractable problem, especially in a high-dimensional setting. Penalized methods are used to obtain sparse solutions due to their computational tractability. Nevertheless, recent developments permit efficiently obtaining good solutions of cardinality-constrained PCA problems allowing comparison between these approaches. Here, we conduct a comparison between a penalized PCA method with its cardinality-constrained counterpart for the least-squares formulation of PCA imposing sparseness on the component weights. We compare the penalized and cardinality-constrained methods through a simulation study that estimates the sparse structure’s recovery, mean absolute bias, mean variance, and mean squared error. Additionally, we use a high-dimensional data set to illustrate the methods in practice. Results suggest that using cardinality-constrained methods leads to better recovery of the sparse structure.

Learning graph embeddings for high-dimensional data is an important technology for dimensionality reduction. The learning process is expected to preserve the discriminative and geometric information of high-dimensional data in a new low-dimensional subspace via either manual or automatic graph construction. Although both manual and automatic graph constructions can capture the geometry and discrimination of data to a certain degree, they working alone cannot fully explore the underlying data structure. To learn and preserve more discriminative and geometric information of the high-dimensional data in the low-dimensional subspace as much as possible, we develop a novel Discriminative and Geometry-Preserving Adaptive Graph Embedding (DGPAGE). It systematically integrates manual and adaptive graph constructions in one unified graph embedding framework, which is able to effectively inject the essential information of data involved in predefined graphs into the learning of an adaptive graph, in order to achieve both adaptability and specificity of data. Learning the adaptive graph jointly with the optimized projections, DGPAGE can generate an embedded subspace that has better pattern discrimination for image classification. Results derived from extensive experiments on image data sets have shown that DGPAGE outperforms the state-of-the-art graph-based dimensionality reduction methods. The ablation studies show that it is beneficial to have an integrated framework, like DGPAGE, that brings together the advantages of manual/adaptive graph construction.

In recent years, new Data Mining (DM) algorithms and methodologies are increasingly used as anindustrial solution for manufacturing improvements. In this context, new techniques are widely required bycompanies in the field of maintenance due to the need to reduce breakdowns intervention and take advantage ofthe increasing availability of data. This paper aims to propose a new learning-based algorithm to improveknowledge extraction by combining different DM techniques from a predictive maintenance perspective. First,the J48 algorithm and Random Forest (RF) are used as a predictive model to classify a set of failure modesaccording to their influence on the Overall Equipment Effectiveness (OEE). Then, the Apriori algorithm is usedto identify the relationship among failure events belonging to the lowest OEE range for which, therefore, apredictive maintenance strategy should be defined. In order to describe the learning-based algorithm proposed inthis paper, a real case study is presented and detailed. The experimental results showed a valuable tool forknowledge extraction and the definition of a set of predictive maintenance strategies for those failures mostaffecting the process. In this way, the complexity of decision-making on maintenance strategies can be reducedmainly when dealing with a large amount of information or a challenging dataset.

Protein secondary structure prediction from its amino acids is purposely used to evaluate and improve the accuracy of performance as well as drug design and cell functionality. Various approaches for predicting protein secondary structure have been used, each with varying accuracy, vulnerabilities, and strengths. In view of this, this paper is aimed at training a deep neural network with particle swarm optimization and comparing the results with the state of accuracy. Also, the methodology used is basic particle swarm optimization for training a 20-15-15-15-3 deep neural network. The Java programming language and the Spring Boot framework were employed to implement the various application programming interfaces of the model. The dataset acquired after the training of JPred Server 1.2, which included 1349 training sets and 149 test sets, was used in training the model. Following the training, it was discovered that the model had a highest accuracy of 53.18 percent on epoch 140, indicating that this model is not a best fit or an alternative to the current state of the art for the prediction of protein secondary structure.

BACKGROUND: Despite evidence demonstrating limited benefit, many clinicians continue to perform routine laboratory testing of well-appearing children to medically clear them before psychiatric admission. METHODS: We conducted a quality improvement project to reduce routine laboratory testing among pediatric patients requiring admission to our psychiatric unit. We convened key stakeholders whose input informed the modification of an existing pathway and the development of a medical clearance algorithm. Our outcome was a reduction in routine laboratory testing for children requiring psychiatric admission. Our balancing measure was the number of patients requiring transfer from the inpatient psychiatry unit to a medical service. We used run charts to evaluate nonrandom variation and demonstrate sustained change. RESULTS: Before the introduction of the new medical clearance algorithm, 93% (n = 547/589) of children with psychiatric emergencies received laboratory testing. After implementing the medical clearance algorithm, 19.6% (n = 158/807) of children with psychiatric emergencies received laboratory testing. Despite a decreased rate of routine testing, there were no transfers to the medical service. CONCLUSIONS: Implementing a medical clearance algorithm can decrease routine laboratory testing without increasing transfers to the medical service among children requiring psychiatric admission.

Con la irrupción de las nuevas tecnologías, la Inteligencia Artificial (IA) está convirtiéndose en una especie de nueva religión monoteísta, y los algoritmos, como si fueran oráculos, se tornan cada vez más omniscientes. No importa su grado de opacidad o la falta de trazabilidad que ostenten. Cada día que pasa el culto a los algoritmos deviene cada vez más reverencial. Parecen sagrados. Entre tanto, las cajas negras impiden que se cumplan los principios de transparencia y explicabilidad. A su vez la atribución de responsabilidad tiende a diluirse., With the emergence of new tecnnologies, Artificial Intelligence (IA) is becoming a kind of new monotheistic religion, and algorithms, as if they were oracles, they are more and more omniscient. It does not matter their degree of opacity or the lack of traceability they suffer. With each passing day the cult of algorithms becomes more reverential. They seem sacred. Meanwhile, black boxes prevent the exercise of the principles of transparency and explainability. In turn, the attribution of responsibility tends to be diluted.

At present, sentiment analysis has become a trend; above all, in digital product development companies, as it is essential for rapid and automatic analysis. Sentiment analysis deals with emotions with the help of software, and it is playing an unavoidable role in workplaces. The constant growth of social networks, especially the Twitter social network, has made the ability to understand and comprehend users or clients take a greater scope regarding their needs; and therefore, increase the complexity of analysis of this social network, causing excessive expenses in time, personnel and money. This work presents a solution through the application of Machine Learning (ML) for sentiment analysis and thus improve analysis, execution time and customer satisfaction. The scope of this research is limited to using the Support Vector Machine (SVM) supervised learning technique for the intended analysis. The model derives from the ML technique making use of cross validation. The applied methodology is the CRISP-ML(Q) Methodology. The results show that the use of ML allows efficient sentiment analysis in Twitter communications.

The article investigates various kinds of the strategies of technosensation in artistic project based on bio-parametrisation’s techniques. The category of technosensation is in the article referred to the considerations of Luciana Parisi and Marie-Luise Angerer to define the relationship between body affectivity and computational systems (primarily in relation to automated decision-making systems and machine learning processes). The context for these considerations is the reflection on “technological redlining” as a strategy for racial, gender, and (dis)ability profiling of computational systems, which generate social exclusion, inequalities and oppressiveness. In the article, the author considers (with reference to the considerations of Parisi, Bernard Stiegler, Yuk Hui and Gabbrielle M. Johnson) to what extent the algorithmic biases are the result of automation, and to what extent they result from the absorption of uncertainty, randomness and technodiversity. Technosensation strategies are considered in relation to the artistic practices of Zach Blas, Maja Smrekar, Marija Griniuk and others, pointing to subversive, critical and affirmative variants of technological functionality and agency. The presented projects prove that the functionality of computational technologies is not bipolar, but it is developing as a spectrum of nuanced mechanisms, both in the area of oppressive-exclusionary systems and emancipatory strategies.

ObjectiveWe developed an algorithm that identifies patients at high risk of morbidity/mortality after cytoreductive surgery for advanced ovarian cancer. We have previously shown that the Mayo triage algorithm reduces operative mortality internally, followed by validation using an external low complexity national dataset. However, validation in a higher complexity surgical setting is required before widespread acceptance of this approach, and this was the goal of our study.MethodsWe included patients who underwent debulking surgery (including primary or interval debulking surgery) for stage IIIC/IV ovarian cancer between October 2011 and November 2019 (SCORPION trial patients until May 2016 and non-trial patients thereafter) at Fondazione Policlinico A Gemelli, Italy. Using the algorithm, we classified patients as either high-risk or triage-appropriate and compared 30-day grade 3+ complications and 90-day mortality using a χ2test or Fisher’s exact test.ResultsA total of 625 patients were included. The mean age was 58.7±11.4 years, 73.6% (n=460) were stage IIIC, and 63.0% (n=394) underwent primary debulking surgery. Surgical complexity was intermediate or high in 82.6% (n=516) of patients (95.7% (n=377) for primary surgery and 60.2% (n=139) for interval surgery), and 20.3% (n=127) were classified as high-risk. When compared with triage-appropriate patients, high-risk patients had (1) a threefold higher rate of 90-day mortality (6.3% vs 2.0%, p=0.02); (2) a higher likelihood of 90-day mortality following a grade 3+ complication (25.9% vs 10.0%, p=0.05); and (3) comparable rates of grade 3+ complications (21.3% vs 16.1%, p=0.17).ConclusionThe evidence-based triage algorithm identifies patients at high risk of morbidity/mortality after cytoreductive surgery. Triage high-risk patients are poor candidates for surgery when complex surgery is required. This algorithm has been validated in heterogeneous settings (internal, national, and international) and degree of surgical complexity. Risk-based decision making should be standard of care when planning surgery for patients with advanced ovarian cancer, whether primary or interval surgery.

This paper investigates the theoretical bound to reduce the parameter uncertainty in Bayesian adaptive estimation for psychometric functions and proposes an exploration-exploitation (E-E) approach to improve the computation efficiency for parameter estimations. When the experimental trial goes on, the uncertainty of the parameters decreases dramatically and the space between the maximal mutual information and the theoretical bound gets narrower, so the advantage of classical Bayesian adaptive estimation algorithm diminishes. This approach tries to trade off the exploration (parameter posterior uncertainty) and the exploitation (parameter mean estimation). The experimental results show that the proposed E-E approach estimates parameters for psychometric functions with same convergence and reduces the computation time by more than 34.27%, compared with the classical Bayesian adaptive estimation.

The computer vision community has taken a keen interest in recent developments in activity recognition and classification in sports videos. Advancements in sports have a broadened the technical interest of the computer vision community to perform various types of research. Images and videos are the most frequently used components in computer vision. There are numerous models and methods that can be used to classify videos. At the same time, there no specific framework or model for classifying and identifying sports videos. Hence, we proposed a framework based on deep learning to classify sports videos with their appropriate class label. The framework is to perform sports video classification using two different benchmark datasets, UCF101 and the Sports1-M dataset. The objective of the framework is to help sports players and trainers to identify specific sports from the large data source, then analyze and perform well in the future. This framework takes sports video as an input and produces the class label as an output. In between, the framework has numerous intermediary processes. Preprocessing is the first step in the proposed framework, which includes frame extraction and noise reduction. Keyframe selection is carried out by candidate frame extraction and an enhanced threshold-based frame difference algorithm, which is the second step. The final step of the sports video classification framework is feature extraction and classification using CNN. The proposed framework result is compared with pretrained neural networks such as AlexNet and GoogleNet, and then the results are also compared. Three different evaluation metrics are used to measure the accuracy and performance of the framework.

The development of toxicity classification models using the ToxCast database has been extensively studied. Machine learning approaches are effective in identifying the bioactivity of untested chemicals. However, ToxCast assays differ in the amount of data and degree of class imbalance (CI). Therefore, the resampling algorithm employed should vary depending on the data distribution to achieve optimal classification performance. In this study, the effects of CI and data scarcity (DS) on the performance of binary classification models were investigated using ToxCast bioassay data. An assay matrix based on CI and DS was prepared for 335 assays with biologically intended target information, and 28 CI assays and 3 DS assays were selected. Thirty models established by combining five molecular fingerprints (i.e., Morgan, MACCS, RDKit, Pattern, and Layered) and six algorithms [i.e., gradient boosting tree, random forest (RF), multi-layered perceptron

Recently, generative adversarial networks (GANs) have progressed enormously, which makes them able to learn complex data distributions in particular faces. More and more efficient GAN architectures have been designed and proposed to learn the different variations of faces, such as cross pose, age, expression, and style. These GAN-based approaches need to be reviewed, discussed, and categorized in terms of architectures, applications, and metrics. Several reviews that focus on the use and advances of GAN in general have been proposed. However, to the best of our knowledge, the GAN models applied to the face, which we call facial GANs , have never been addressed. In this article, we review facial GANs and their different applications. We mainly focus on architectures, problems, and performance evaluation with respect to each application and used datasets. More precisely, we review the progress of architectures and discuss the contributions and limits of each. Then, we expose the encountered problems of facial GANs and propose solutions to handle them. Additionally, as GAN evaluation has become a notable current defiance, we investigate the state-of-the-art quantitative and qualitative evaluation metrics and their applications. We conclude this work with a discussion on the face generation challenges and propose open research issues.

Machine learning (ML) models are often modelled to predict cancer prognosis but rarely consider spatial factors in a region. Hence this study explored machine learning algorithms utilising Local Government Areas (LGAs) in Queensland, Australia to spatially predict 3- and 5-year prognosis of oral cancer patients and provide clinical interpretability of the predicted outcome made by the ML model.Data from a total of 3,841 oral cancer patients were retrieved from the Queensland Cancer Registry (QCR). Synthesizing minority oversampling technique together with edited nearest neighbours (SMOTE-ENN) was used to pre-process unbalanced datasets. Five ML models: logistic regression, random forest classifier, XGBoost, Gaussian Naïve Bayes and Voting Classifier were trained. Predictive features were age, sex, LGAs, tumour site and differentiation. Outcomes were 3- and 5-year overall survival of patients. Model performances on test set were evaluated using area under the curve and F1 scores. SHapley Additive exPlanations (SHAP) method was applied to the best performing model for model interpretation of the predicted outcome.The Voting Classifier was the best performing model with F1 score of 0.58 and 0.64 for 3- and 5-year overall survival, respectively. Age was the most important feature in the Voting Classifier in 3- and 5-year prognosis prediction. LGAs at diagnosis was the top 3 predictive feature for both 3- and 5-year models.The Voting Classifier demonstrated the best overall performance in classifying both 3- and 5-year overall survival of oral cancer patients in Queensland. SHAP method provided clinical understanding of the predictive features of the Voting Classifier.

Cerebrovascular segmentation in time-of-flight magnetic resonance angiography (TOF-MRA) volumes is essential for a variety of diagnostic and analytical applications. However, accurate cerebrovascular segmentation in 3D TOF-MRA is faced with multiple issues, including vast variations in cerebrovascular morphology and intensity, noisy background, and severe class imbalance between foreground cerebral vessels and background. In this work, a 3D adversarial network model called A-SegAN is proposed to segment cerebral vessels in TOF-MRA volumes. The proposed model is composed of a segmentation network A-SegS to predict segmentation maps, and a critic network A-SegC to discriminate predictions from ground truth. Based on this model, the aforementioned issues are addressed by the prevailing visual attention mechanism. First, A-SegS is incorporated with feature-attention blocks to filter out discriminative feature maps, though the cerebrovascular has varied appearances. Second, a hard-example-attention loss is exploited to boost the training of A-SegS on hard samples. Further, A-SegC is combined with an input-attention layer to attach importance to foreground cerebrovascular class. The proposed methods were evaluated on a self-constructed voxel-wise annotated cerebrovascular TOF-MRA segmentation dataset, and experimental results indicate that A-SegAN achieves competitive or better cerebrovascular segmentation results compared to other deep learning methods, effectively alleviating the above issues.

This work concentrates on the issue of leader-following bipartite synchronization of multiple memristive neural networks with Markovian jump topology. In contrast to conventional coupled neural network systems, the coupled neural network model under consideration possesses both cooperative and competitive connections among neuron nodes. Specifically, the interaction between neighbors' nodes is described by a signed graph, in which a positive weight represents an alliance relationship between two neuron nodes while a negative weight represents an adversarial relationship between two neuron nodes. By designing a pinning discontinuous controller that makes full use of the mode information, some effective criteria that ensure the stability of bipartite synchronization error states are obtained. All network nodes can synchronize the target node state bipartitely. Finally, two simulation examples are provided to demonstrate the viability of the suggested bipartite synchronization control approach.