Classification Threshold Research Articles

Incident atrial fibrillation prediction using ECG-based deep learning at a specialized tertiary cardiac care center

Abstract Background Deep learning (DL) applied to electrocardiograms (ECG) is an emerging modality in the prediction of incident atrial fibrillation or atrial flutter (AF). The generalizability of such methods to a tertiary heart center (THC) population has not yet been formally explored. Purpose Evaluate the performance of DL in predicting 5-year incident AF using a resting 12-lead ECG in sinus rhythm acquired at a THC. Methods In a retrospective study, we examined 1.4 million ECGs acquired from over 250,000 adults at a THC between 2004 and 2022. ECGs were excluded if they were performed within 30 days of cardiac surgery, captured a rhythm other than sinus rhythm, or were acquired in patients with pre-existing AF. Incident AF at 5 years was modelled as a binary outcome and was determined on the basis of available outpatient and inpatient clinical databases and ECG diagnoses. Included ECGs were randomly split by distributing patients into training (70%), validation (10%), and test (20%) sets. A ResNet-50 model was trained using the training set. Hyperparameters were optimized using the validation set. The tuned model's performance is reported on the test set. The results at the patient level were derived by averaging the model's probability outputs for ECGs grouped according to both their AF outcome and the patient's identity. Bootstrapping was used to report confidence intervals (CI). Sensitivity and specificity are reported at a classification threshold based on the Matthews correlation coefficient. Saliency maps were used to enhance the model’s explainability. Results A total of 669,782 ECGs (47% of the screened ECGs) were included among 145,323 patients. Mean age was 63±15 years and 62% were male. The 5-year incident AF outcome was observed in 12% of ECGs and 16% of patients. The performance of the tuned model was first evaluated at the ECG level on the test set demonstrating an area under the receiver operating curve (AUC) of 0.75 (95% CI: 0.745-0.753) and an integrated calibration index of 0.009 (95% CI: 0.008-0.011). When testing the model at the patient level to simulate a deployment scenario, the AUC improved to 0.78 (95% CI: 0.768-0.783) with a sensitivity of 50% (95% CI: 49-52) and specificity of 87% (95% CI: 86.5-87.3). Stratified results by sex showed a better AUC in females, 0.81 (95% CI: 0.80-0.82), compared to males, 0.75 (95% CI: 0.74-0.76). Subgroup analyses revealed a trend of improved performance at the patient level with an increasing number of ECGs. Saliency maps highlighted the P-wave area as having the highest influence on the model’s prediction, thereby enhancing the model’s plausibility and explainability. Conclusion A unimodal ECG-based deep learning model showed promising 5-year incident AF prediction performance in a cohort of all-comer patients at a tertiary heart center. Further studies could explore the use of multimodal prediction models that integrate ECGs with other clinical and imaging data.

Integrating D4Z4 methylation analysis into clinical practice: improvement of FSHD molecular diagnosis through distinct thresholds for 4qA/4qA and 4qA/4qB patients

BackgroundFacioscapulohumeral dystrophy (FSHD) is a myopathy characterized by the loss of repressive epigenetic features affecting the D4Z4 locus (4q35). The assessment of DNA methylation at two regions (DUX4-PAS and DR1) of D4Z4 locus proved to be an effective method to detect epigenetic signatures compatible with FSHD. The present study aims at validating the employment of this method into clinical practice and improving the protocol by refining the classification thresholds of 4qA/4qA patients. To this purpose, 218 subjects with clinical suspicion of FSHD collected in 2022–2023 were analyzed. Each participant underwent in parallel the traditional FSHD molecular testing (D4Z4 sizing) and the proposed methylation assay. The results provided by both analyses were compared to evaluate the concordance and calculate the performance metrics of the methylation test.ResultsAmong the 218 subjects, the 4q variant type distribution was 54% 4qA/4qA, 43% 4qA/4qB and 3% 4qB/4qB. The methylation analysis was performed only on carriers of at least one 4qA allele. After refining the classification threshold, the test reached the following performance metrics: sensitivity = 0.90, specificity = 1.00 and accuracy = 0.93. These results confirmed the effectiveness of the methylation assay in identifying patients with genetic signature compatible with FSHD1 and FSHD2 based on their DUX4-PAS and DR1 profile, respectively. The methylation data were also evaluated with respect to the clinical information.ConclusionsThe study confirmed the ability of the method to accurately identify methylation profiles compatible with FSHD genetic signatures considering the 4q genotype. Moreover, the test allows the detection of hypomethylated profiles in asymptomatic patients, suggesting its potential application in identifying preclinical conditions in patients with positive family history and FSHD genetic signatures. Furthermore, the present work emphasizes the importance of interpreting methylation profiles considering the patients’ clinical data.

Multiple-Criteria Decision Analysis for Assessments of Chemical Alternatives (MCDA-ACA).

A comprehensive assessment of chemical alternatives (ACA) is necessary to avoid regrettable substitution. In a preceding study, an analysis of six hazard assessment methods found that none of them are fully aligned with the hazard assessment criteria of Article 57 of the European REACH regulation, indicating a need for a method better reflecting hazard assessment schemes in European chemical regulations. This paper presents a multiple-criteria decision analysis (MCDA) method for the ACA that takes the criteria of Article 57 of REACH into account. Investigated and presented are objective hierarchies, the aggregation of objectives, the curvature of the value functions, weights, and the introduction of a classification threshold. The MCDA-ACA method allows for the aggregation of hazards in such a way that poor performance in one hazard cannot be compensated for by good performance in another hazard. The method parameters were developed and tested using two data sets with the aim to classify chemical alternatives into acceptable (nonregrettable) and unacceptable (regrettable) alternatives according to the regulations set in Europe. The flexibility of the general method was explored by adapting the method to align with two hazard assessment schemes, Article 57 of REACH and GreenScreen. The results show that MCDA-ACA is so flexible and transparent that it can easily be adapted to various hazard assessment schemes.

Predictive diagnostic value of mean platelet volume to platelet count and neutrophil to lymphocyte ratios in the gray zone of prostate cancer with tPSA between 4 to 10 ng/mL.

Exploration of the Predictive Diagnostic Value of Mean Platelet Volume to Platelet Count Ratio (MPV/PLT,PVI) and Neutrophil-to-Lymphocyte Ratio (NLR) in the tPSA Gray Zone of Prostate Cancer. A retrospective study was conducted on 65 prostate cancer (Pca) patients and 52 benign prostatic hyperplasia (BPH) patients who underwent transperineal prostate biopsy at Xiangtan Central Hospital from December 2021 to December 2023. Descriptive statistics and logistic regression models were used to investigate the predictive diagnostic value of PVI and NLR in the tPSA gray zone of prostate cancer. Receiver operating characteristic (ROC) curves were constructed based on PVI and NLR values to determine the classification thresholds. A total of 117 patients were enrolled, including 65 cases of prostate cancer (PCa) and 52 cases of benign prostatic hyperplasia (BPH). There were no statistically significant differences in age, BMI, history of hypertension, history of diabetes, history of coronary heart disease, pre-biopsy white blood cell count, history of drinking, history of smoking, and tPSA between the PCa and BPH patients. The results of logistic regression analysis showed that PVI (OR=2.03, 95%CI: 1.34~3.07, P<0.00) and NLR (OR=0.32, 95%CI: 0.18~0.58, P<0.00) were independent predictors for diagnosing prostate cancer in the tPSA gray zone (VIF=1.04).The maximum area under the curve (AUC) for PVI was 0.70, with an optimal cut-off value of 0.05 (P ≤ 0.01). The maximum AUC for NLR was 0.76, with an optimal cut-off value of 2.86 (P ≤ 0.01).The calibration curve showed good consistency between the predicted and actual outcomes in both the PCa and BPH groups, indicating that the nomogram model had good predictive performance.When using PVI and NLR to plot the receiver operating characteristic (ROC) curves to predict the assessment of PCa in the tpsa gray zone, the area under the curve (AUC) for PVI was the largest at 0.70, with an optimal cutoff value of 0.05 (P ≤ 0.01). The AUC for NLR was the largest at 0.76, with an optimal cutoff value of 2.86 (P ≤ 0.01). PVI and NLR have certain predictive diagnostic value for Pca in the tPSA gray zone, and appropriate use of PVI and NLR can improve the positive rate of early screening for Pca in the gray zone.

Left Atrial Structural and Functional Changes in Adults with Congenital Septal Defects and Paroxysmal Atrial Fibrillation.

Aims. To identify the difference between adult patients with septal defects and paroxysmal atrial fibrillation (AF) and patients without a history of arrhythmia using the left atrial (LA) volume and function parameters, to reveal the parameters associated with AF development. Methods and results. In this prospective study, 81 patients with septal defects and left-to-right shunts were enrolled between 2021 and 2023 and divided into two groups: with paroxysmal AF and without AF. Left atrial function was analyzed based on the indexed left atrial volumes (LAVI and preA-LAVI), ejection fraction (LAEF), expansion index (LAEI), reservoir (LAS-r), conduit (LAS-cd) and contractile (LAS-ct) strain, and stiffness index (LASI) using a Philips CVx3D ultrasound system (Philips, Amsterdam, The Netherlands) and corresponding software. In total, 26 patients with paroxysmal atrial fibrillation (mean age: 59.6 ± 11.7 years, female: 80.8%) and 55 patients with septal defects without any history of arrhythmias (mean age: 44.8 ± 11.6 years, female: 81.8%) were included. All patients were in the NYHA class I or II at baseline. Our findings demonstrated a significant difference between all LA function parameters in the two groups. Upon univariable analysis, the LAVI, preA-LAVI, LASI, LAEF, LAEI, LAS-r, LAS-c, LAS-ct, age, cardiac index, E/A, and RV pressure were found to be associated with AF. The multivariate analysis identified LAVI (OR 1.236, 95% CI 1.022-1.494, p = 0.03), LAS-r (OR 0.723, 95% CI 0.556-0.940, p = 0.02), and LAS-ct (OR 1.518, 95% CI 1.225-1.880, p < 0.001) as independent predictors of AF development. The proposed model demonstrated high sensitivity and specificity with an adjusted classification threshold of 0.38 (AUC: 0.97, 95% CI 0.93-1.00, sensitivity 92% and specificity 92%, p < 0.001). Conclusions. The assessment of LA function using speckle-tracking echocardiography demonstrated significantly different values in the AF group among patients with congenital septal defects. This technique can therefore be implemented in routine clinical management. The key message. Atrial fibrillation development in adult patients with congenital septal defects and a left-to-right shunt is associated with the changes in left atrial function under conditions of an increased preload.

Imbalanced image classification algorithm based on fine-grained analysis

Fine-grained attribute analysis and data imbalance have always been research hotspots in the field of computer vision. Due to the complexity and diversity of fine-grained attribute images, traditional image classification methods have shortcomings in paying attention to fine-grained attributes of images and perform poorly when dealing with imbalanced data sets. To overcome these problems, this study proposes a fine-grained image threshold classification algorithm based on deep metric learning. By introducing a metric learning method, the focus on fine-grained attributes of images is enhanced. At the same time, by applying pairwise loss and proxy loss, the classification accuracy of the model is improved and the model convergence speed is accelerated. In order to deal with the problem of data imbalance, a classifier based on threshold analysis is designed. The classifier uses threshold analysis technology to achieve multi-level classification of fine-grained images, thereby improving the problem of low classification accuracy of a few categories in imbalanced data sets. Experimental results show that the proposed fine-grained image threshold classification algorithm based on deep metric learning is significantly superior to other methods in terms of classification accuracy.

Megabenthic Diversity Patterns on a Seamount in the Philippine Sea: Implications for Conservation Planning on the Kyushu-Palau Ridge.

The oligotrophic tropical western Pacific region is characterized by a high density of seamounts, with the Kyushu-Palau Ridge (KPR) being the longest seamount chain here. Effective spatial management plans for seamount ecosystems necessitate an understanding of distribution patterns and key environmental factors influencing benthic communities. However, knowledge regarding deep-sea biodiversity patterns over intricate topography remains limited. In this study, we investigated a seamount with a water depth of 522 m at the summit located in the southern section of KPR. Survey transects were conducted from 522 m to 4059 m. By analyzing video-recorded data obtained by a human-occupied vehicle (HOV) during dives and environmental variables derived from bathymetry, distinct assemblages were identified through noise clustering. α- and β-diversity patterns within the seamount megabenthic community were analyzed across the depth gradient, along with investigation of their environmental drivers. A total of 10,596 megafauna individuals were documented, categorized into 88 morphospecies and statistically separated into six distinct community clusters using noise clustering analysis. Species abundance and richness were highest within the 700-800 m water depth range, declining notably beyond 2100 m, indicating a critical threshold for habitat classification in this region. The β-diversity of megabenthic communities was high (0.836). Although β-diversity patterns along the depth gradient were mostly dominated by differences in species richness, the contribution of species replacement increased with depth, becoming dominant at depths greater than 3000 m. Depth emerged as the primary driver of spatial variation in community structure, while near-bottom current velocity, topographic parameters (bathymetric position index, slope), and substrate type also influenced the formation of microhabitats. The study highlights the depth gradients, thresholds, and other intricate environmental factors shaping the spatial heterogeneity of these communities. It provides valuable insights for the future development of effective survey and conservation strategies for benthic biodiversity on the KPR.

Precise and Accurate Short-term Forecasting of Solar Energetic Particle Events with Multivariate Time-series Classifiers

Solar energetic particle (SEP) events are one of the most crucial aspects of space weather that require continuous monitoring and forecasting using robust methods. We demonstrate a proof of concept of using a data-driven supervised classification framework on a multivariate time-series data set covering solar cycles 22, 23, and 24. We implement ensemble modeling that merges the results from three proton channels (E ≥ 10 MeV, 50 MeV, and 100 MeV) and the long-band X-ray flux (1–8 Å) channel from the Geostationary Operational Environmental Satellite missions. Our task is binary classification, such that the aim of the model is to distinguish strong SEP events from nonevents. Here, strong SEP events are those crossing the Space Weather Prediction Center’s “S1” threshold of solar radiation storm and proton fluxes below that threshold are weak SEP events. In addition, we consider periods of nonoccurrence of SEPs following a flare with magnitudes ≥C6.0 to maintain a natural imbalance of sample distribution. In our data set, there are 244 strong SEP events comprising the positive class. There are 189 weak events and 2460 “SEP-quiet” periods for the negative class. We experiment with summary statistic, one-nearest neighbor, and supervised time-series forest (STSF) classifiers and compare their performance to validate our methods for prediction windows from 5 minutes up to 60 minutes. We find the STSF model to perform better under all circumstances. For an optimal classification threshold of ≈0.3 and a 60 minutes prediction window, we obtain a true skill statistic TSS = 0.850, Heidke skill score HSS = 0.878, and Gilbert skill score GSS = 0.783.

A stethoscope-guided interpretable deep learning framework for powder flow diagnosis in cold spray additive manufacturing

Cold spray (CS) particle deposition, also known as cold spray additive manufacturing, has garnered great attention as an advanced additive manufacturing (AM) and surface deposition technology, facilitating rapid and scalable production of functional parts and surfaces in a solid-state manner. In CS, consistent and precise feeding of functional feedstock powders is crucial for achieving effective particle deposition. However, vibratory-based powder feeders often face challenges associated with powder delivery and powder segregation. This underscores the critical need for a precise diagnostic framework to effectively control powder flow during cold spraying. To this end, the present study proposes a powder flow monitoring framework for the CS process using a stethoscope sound-guided interpretable deep learning (IDL) model. Internal sound data from the vibrated powder feeder is collected through a stethoscope sensor to train a two-stage model. In the first stage, a convolutional autoencoder (CAE) is trained to build an unsupervised learning-based anomaly detector, which identifies classification thresholds based on the receiver operation characteristic curve. In the second stage, a convolutional neural network (CNN) model is trained as the powder flow diagnostic tool by considering process anomalies, namely (i) no powder flow; (ii) feeder clogging; and (iii) no gas flow. The results reveal that the stethoscope sound-guided model achieves a classification accuracy of 95% on the test set, significantly outperforming benchmark utilizing typical external-sound recording microphones in diagnosing CS powder flow. Furthermore, the model is visualized and interpreted by employing t-distribution stochastic neighbor embedding and integrated gradients techniques to enhance reliability of CS powder flow diagnosis. This research highlights the effectiveness of the stethoscope sound-guided IDL model for in situ powder flow monitoring and process diagnosis in the domain of cold spray additive manufacturing, contributing to effective particle deposition.

Biotic homogenization in multisystem cascade reservoirs: insights from sedimentary photopigment analysis.

The existing literature provides limited insights into the dynamics of phytoplankton communities and the spatial heterogeneity of physicochemical parameters in multisystem cascade reservoirs (interconnected reservoirs derived from different rivers). The existing studies are concentrated on cascade reservoirs (interconnected reservoirs derived from the same river). To address this knowledge gap, the aims of the present study were as follows: (1) investigate the spatial heterogeneity, within and between reservoirs, of geochemical parameters associated with the eutrophication process, considering total phosphorus, chlorophyll-a, pheophytin, and metals (chromium, copper, nickel, lead, zinc, iron, and manganese); (2) evaluate sediment quality at the designated locations; (3) assess differences in the richness and concentration of sedimentary photopigments between the reservoirs. Application of principal component analysis revealed discernible gradients for the abiotic variables, although the differences were not statistically significant (one-way PERMANOVA test, p > 0.05). The observations suggested a tendency towards spatial homogeneity within and between the reservoirs. The metal concentrations were consistent with regional reference values, while phosphorus levels in the sediment approached the threshold for classification as pollution (~ 2000mg/kg). Analysis of pigments indicated low dissimilarity among the reservoirs, which could be mainly attributed to the eutrophication process and high connectivity of the sampled areas. To counteract ongoing biotic homogenization, it is essential to reduce nutrient inputs and invest in ecological protection and restoration programs. The analysis of sedimentary photopigments provides an efficient and cost-effective alternative way to assess phytoplankton communities.

Classification Criteria for Axial Disease in Youth with Juvenile Spondyloarthritis.

To develop and validate classification criteria for axial disease in youth with juvenile spondyloarthritis (SpA; AxJSpA). This international initiative consisted of four phases: 1) Item generation; 2) Item reduction; 3) Criteria development; and 4) Validation of the AxJSpA criteria by an independent team of experts in an internationally representative Validation cohort. These criteria are intended to be used on youth with a physician diagnosis of juvenile SpA and for whom axial disease is suspected. Item generation consisted of a systematic literature review and a free-listing exercise using input from international physicians and collectively resulted in 108 items. After the item reduction exercise and expert panel input, 37 items remained for further consideration. The final AxJSpA criteria domains included: imaging: active inflammation, imaging: structural lesions, pain chronicity, pain pattern, pain location, stiffness, and genetics. The most heavily weighted domains were active inflammation and structural lesions on imaging. Imaging typical of sacroiliitis was deemed necessary, but not sufficient, to classify a youth with AxJSpA. The threshold for classification of AxJSpA was a score of ≥55 (out of 100). When tested in the validation data set, the final criteria had a specificity of 97.5% (95% CI: 91.4-99.7), sensitivity of 64.3% (95% CI: 54.9-73.1) and Area Under the Receiver Operating Characteristic (AUROC) curve of 0.81 (95% CI: 0.76-0.86). The new AxJSpA classification criteria require an entry criterion, physician diagnosis of juvenile SpA, and include seven weighted domains. The AxJSpA classification criteria are validated and designed to identify participants for research studies.

Perceived Effect of Shipborne Sewage Discharge on Water Quality of Apapa Sea Port

The Apapa Sea Port in Lagos, Nigeria, faces significant environmental concerns due to shipborne sewage discharge. Over 3 billion tons of sewage are discharged annually, posing health risks and causing disease outbreaks. Understanding this relationship is crucial for developing policies to protect marine environments and port operations. The study investigates the perceived effects of sewage wastewater pollution on the marine environment, focusing on Apapa Port, Nigeria’s largest and busiest port. The study used a sample of 475 individuals, with a stratified frame of 323 respondents, resulting in a sample size of 245 using Yaro Yamane’s formula. Descriptive statistical methods were used to summarize and interpret data, with a mean score of 2.5 as the threshold for classification. Inferential statistics, including ANOVA and regression analysis, were used to test the research hypotheses. The results shown that the perceived impact of sewage pollution on water quality, with findings showing a strong awareness and concern among port personnel regarding its deleterious effects, including eutrophication (mean rating: 3.825), harmful algal blooms (mean rating: 3.921), habitat degradation (mean rating: 3.868), and waterborne diseases (mean rating: 4.018). The Hypotheses testing confirmed that sewage discharge significantly impacts the study area (p-value = 0.048) and that the implementation of MARPOL Annex IV has a significant influence on pollution levels (p-value &lt; 0.05). The study concludes with recommendations to improve treatment processes, enforcement mechanisms, infrastructure, and stakeholder collaboration to ensure compliance and protect the marine ecosystem at Lagos Port, among many others. Keywords: Apapa seaports, sewage, shipborne, pollution, waste management.

Development of the 2023 ACR/EULAR Antiphospholipid Syndrome Classification Criteria, Phase III-D Report: Multicriteria Decision Analysis.

The 2023 American College of Rheumatology/EULAR antiphospholipid syndrome (APS) classification criteria development, which aimed to identify patients with high likelihood of APS for research, employed a four-phase methodology. Phase I and II resulted in 27 proposed candidate criteria, which are organized into laboratory and clinical domains. Here, we summarize the last stage of phase III efforts, employing a consensus-based multicriteria decision analysis (MCDA) to weigh candidate criteria and identify an APS classification threshold score. We evaluated 192 unique, international real-world patients referred for "suspected APS" with a wide range of APS manifestations. Using proposed candidate criteria, subcommittee members rank ordered 20 representative patients from highly unlikely to highly likely to have APS. During an in-person meeting, the subcommittee refined definitions and participated in an MCDA exercise to identify relative weights of candidate criteria. Using consensus decisions and pairwise criteria comparisons, 1000Minds software assigned criteria weights, and we rank ordered 192 patients by their additive scores. A consensus-based threshold score for APS classification was set. Premeeting evaluation of 20 representative patients demonstrated variability in APS assessment. MCDA resolved 81 pairwise decisions; relative weights identified domain item hierarchy. After assessing 192 patients by weights and additive scores, the Steering Committee reached consensus that APS classification should require separate clinical and laboratory scores, rather than a single-aggregate score, to ensure high specificity. Using MCDA, candidate criteria preliminary weights were determined. Unlike other disease classification systems using a single-aggregate threshold score, separate clinical and laboratory domain thresholds were incorporated into the new APS classification criteria.

Predicting whether patients in an acute medical unit are physiologically fit-for-discharge using machine learning: A proof-of-concept

IntroductionDelays in discharging patients from Acute Medical Units hamper patient flows throughout the hospital. The decision to discharge a patient is mainly based on the patients’ physiological condition, but may vary between physicians. An objective decision-support system based on patients’ physiological data may help minimizing unnecessary delays in discharge. The aim of this proof-of-concept study is to assess the feasibility of predicting whether patients in an Acute Medical Unit are physiologically fit-for-discharge using machine learning with commonly available hospital data. Furthermore, this study investigated how long before actual time of discharge from the Acute Medical Unit we could predict discharge fitness. Also, the predictive importance of features extracted from these data was assessed. MethodsElectronic Medical Records of patients who participated in a Randomized Controlled Trial conducted in an Acute Medical Unit were used retrospectively (N = 199). Only commonly available hospital data were used. Logistic Regression and Random Forest models were applied to predict every hour whether patients were physiologically fit-for-discharge. Nested 5-fold cross-validation with 5 repeats was used to optimize the model hyperparameters and to estimate the predictive performances. ResultsPhysiological discharge fitness was predictable with reasonable performance for Logistic Regression (mean AUROC: 0.67) and Random Forest (mean AUROC: 0.69). For an intuitively chosen classification threshold of 0.8, mean specificity was 93.3 % and sensitivity 14.1 %. Models could predict physiological discharge fitness more than 24 h earlier than actual time of discharge for most patients who were correctly predicted to be fit-for-discharge. Patient characteristics, vital signs and laboratory results were shown to be important predictors. ConclusionThis proof-of-concept study showed that it is feasible to predict with machine learning whether patients in an Acute Medical Unit are physiologically fit-for-discharge using commonly available hospital data.

Turning the attention to time-resolved EPID-images: treatment error classification with transformer multiple instance learning

Objective. The aim of this work was to develop a novel artificial intelligence-assisted in vivo dosimetry method using time-resolved (TR) dose verification data to improve quality of external beam radiotherapy. Approach. Although threshold classification methods are commonly used in error classification, they may lead to missing errors due to the loss of information resulting from the compression of multi-dimensional electronic portal imaging device (EPID) data into one or a few numbers. Recent research has investigated the classification of errors on time-integrated (TI) in vivo EPID images, with convolutional neural networks showing promise. However, it has been observed previously that TI approaches may cancel out the error presence on γ-maps during dynamic treatments. To address this limitation, simulated TR γ-maps for each volumetric modulated arc radiotherapy angle were used to detect treatment errors caused by complex patient geometries and beam arrangements. Typically, such images can be interpreted as a set of segments where only set class labels are provided. Inspired by recent weakly supervised approaches on histopathology images, we implemented a transformer based multiple instance learning approach and utilized transfer learning from TI to TR γ-maps. Main results. The proposed algorithm performed well on classification of error type and error magnitude. The accuracy in the test set was up to 0.94 and 0.81 for 11 (error type) and 22 (error magnitude) classes of treatment errors, respectively. Significance. TR dose distributions can enhance treatment delivery decision-making, however manual data analysis is nearly impossible due to the complexity and quantity of this data. Our proposed model efficiently handles data complexity, substantially improving treatment error classification compared to models that leverage TI data.

A Gate-Level Power Estimation Approach with a Comprehensive Definition of Thresholds for Classification and Filtering of Inertial Glitch Pulses

Estimation of power consumption in digital circuits is performed at gate-level simulation. Its accuracy depends on the models of gate delays that capture the effects of spurious signal transitions, called “glitches”. Electronic Design Automation (EDA) software considers inertial gate delays and represses a glitch in the cell’s output if its width is below a threshold. Selecting threshold values for the inertial glitch classification and filtering is crucial for precise power estimations. In this direction, we explore the effectiveness of automatically adjusting such thresholds on a cell-specific basis according to the local cell’s information. We used a commercial industry-standard gate-level power estimation tool and a 32 nm CMOS standard cell library. Via power measurements in circuit simulations, we created customized lookup tables for each library cell employed in the benchmark circuits. We compared the proposed approach’s performance with other methods for glitch threshold definition. Our method demonstrated good power estimation accuracy while presenting the lowest mean absolute error among all the cells of the circuits under test and the smallest standard deviation. The latter suggests that the proposed method achieves better cell-specific accuracy, which is expected to allow for more precise circuit-level power estimations in complex circuits with a large number of combinational cells.

Comparative investigations of silanes in UN Test O.2 (liquid oxidizer)

Decachlorotetrasilane, octachlortrisilane and hexachlorodisilane show false-positive results in the UN Test O.2 (liquid oxidizer). The cause of this unexpected result had been explained in former investigations. To clarify whether it is possible to obtain false-positive test results in this statutory and legally required test also with other silanes, altogether 12 substances were compared with each other. The focus was on investigating chlorinated silanes. All silanes produced the expected result “non-oxidizing liquid” as required by the theory showing that the international UN O.2 test for other (chloro)silanes gives reliable results. The averaged pressure increase time of all investigated silanes were slower than those of the reference oxidant HNO3. Therefore, organo-silanes were outside the classification threshold for packing group III. Weaknesses and possibilities for improvement of this legally required test also had been provided.

Adaptive iterative transfer learning for effective snapping shrimp sound detection.

This study aims to detect the bioacoustics signal in the underwater soundscape, specifically those produced by snapping shrimp, using adaptive iterative transfer learning. The proposed network is initially trained with pre-classified snapping shrimp sounds and Gaussian noise, then applied to classify and remove snapping-free noise from field data. This separated ambient noise is subsequently used for transfer learning. This process was iterated to distinguish more effectively between ambient noise and snapping shrimp sounds characteristics, resulting in improved classification. Through iterative transfer learning, significant improvements in precision and recall were observed. The application to field data confirmed that the trained network could detect signals that were difficult to identify using existing threshold classification methods. Furthermore, it was found that the rate of false detection decreased, and detection probability improved with each stage. This research demonstrates that incorporating the noise characteristics of field data into the trained network via iterative transfer learning can generate more realistic training data. The proposed network can successfully detect signals that are challenging to identify using existing threshold classification methods.

Frame-Based Change Detection Using Histogram and Threshold to Separate Moving Objects from Dynamic Background

Detecting and subtracting the Motion objects from backgrounds is one of the most important areas. The development of cameras and their widespread use in most areas of security, surveillance, and others made face this problem. The difficulty of this area is unstable in the classification of the pixels (foreground or background). This paper proposed a suggested background subtraction algorithm based on the histogram. The classification threshold is adaptively calculated according to many tests. The performance of the proposed algorithms was compared with state-of-the-art methods in complex dynamic scenes.

Mathematical Model of Logistic Regression for Binary Classification. Part 2. Data Preparation, Learning and Testing Processes

This article reviews the theoretical aspects of logistic regression for binary data classification, including data preparation processes, training, testing, and model evaluation metrics. Requirements for input data sets are formulated, methods of coding categorical data are described, methods of scaling input features are defined and substantiated. A scheme for learning logistic regression using the gradient descent method has been developed to minimize the loss function by the appropriate adjustment of the weights of the features of the sample of objects intended for classification. Features of the construction of recurrent methods of classical and stochastic gradient descent are determined. The requirements for the organization of the data sample for the multi-stage learning model in order to avoid overtraining or undertraining of logistic regression are described. The scheme of testing the trained logistic regression is given and the main quality metrics of binary classification are described. The influence of the height of the classification threshold on the efficiency of logistic regression was noted. According to the results of the work, the directions of perspective research of logistic regression are outlined.