To evaluate the performance of a body mass index (BMI)-based sub-milliSievert low-dose CT (LDCT) protocol with multiple reconstruction algorithms for image quality and lung nodule assessment. This prospective study included 214 participants who underwent standard-dose CT (SDCT, 3.68 ± 1.53 mSv) reconstructed with 50% adaptive statistical iterative reconstruction (ASIR-V-50%) and LDCT. LDCT was randomly divided into a higher-dose group (LD-A, 0.57-1.15 mSv, n = 108) and a lower-dose group (LD-B, 0.33-0.63 mSv, n = 106). Each group was stratified into four BMI-based subgroups with individualized protocols reconstructed with deep learning image reconstruction (DLIR-H and DLIR-M), ASIR-V-50%, and filtered back projection (FBP). Image quality, nodule detection across BMI subgroups, and the performance of four algorithms in detection, size measurement accuracy, and Lung-RADS v2022 consistency were analyzed. In LDCT, DLIR-H provided superior image quality (p < 0.001) and the highest overall nodule detection rate (99.04%), surpassing ASIR-V-50% (98.55%) and FBP (97.87%) (both p < 0.05). The advantage was most evident for nodules < 6 mm, while all nodules ≥ 6 mm were consistently detected across algorithms. Detection rates showed no significant variation among BMI subgroups (all p > 0.05). For measurement accuracy, FBP and ASIR-V-50% performed better in LD-A (all p < 0.05), whereas DLIR-M was superior in LD-B (p < 0.001). All algorithms demonstrated excellent Lung-RADS agreement (κ > 0.9, p < 0.001). A BMI-based sub-milliSievert LDCT protocol significantly reduced radiation exposure while maintaining nodule detection across BMI subgroups, with DLIR offering superior image quality and diagnostic performance. Question Evidence remains scarce on BMI-based sub-milliSievert low-dose CT using different reconstruction algorithms, regarding image quality and nodules detection (particularly < 6 mm). Findings BMI-based sub-milliSievert low-dose CT ensured balanced detectability across populations, while deep learning reconstruction improved image quality and achieved excellent sensitivity for lung nodule detection. Clinical relevance Deep learning reconstruction enhanced BMI-based sub-milliSievert low-dose CT, supporting its application in personalized sub-milliSievert low-dose lung cancer screening.

Traditional police combat training relies heavily on subjective evaluation by human instructors, which lacks consistency and comprehensive coverage of complex movement patterns in real-world scenarios. This paper presents an enhanced deep spatio-temporal graph convolutional network (ST-GCN) framework specifically designed for automated police combat action recognition and quality assessment. The proposed method incorporates adaptive graph topology learning mechanisms that dynamically adjust spatial connectivity patterns based on action-specific joint relationships, multi-modal fusion strategies combining skeletal and RGB video data for robust recognition under diverse environmental conditions, and comprehensive quality assessment algorithms providing objective evaluation of technique execution. The enhanced ST-GCN architecture features attention-guided feature extraction, curriculum learning-based training strategies, and real-time processing capabilities suitable for practical deployment in training facilities. Experimental validation on a comprehensive police combat dataset demonstrates superior performance with 96.7% recognition accuracy across twelve action categories and real-time processing at 42.8 frames per second. The multi-dimensional evaluation framework successfully assesses action completion, standardization compliance, and movement fluency, providing immediate feedback for skill development. The proposed system offers significant improvements over conventional approaches, enabling standardized evaluation criteria, data-driven curriculum development, and enhanced training effectiveness for law enforcement personnel.

In intelligent sports education, current action quality assessment (AQA) methods face significant limitations: regression-based methods are heavily dependent on high-quality annotated data, while unsupervised methods lack sufficient accuracy and degrade performance when handling long-duration sequences. To address these challenges, this paper introduces a novel indirect scoring method integrating action anomaly detection with a Quick Action Quality Assessment (QAQA) algorithm. In this method, the proposed anomaly detection module dynamically adjusts action quality scores by identifying and analyzing acceleration outliers between frames, effectively improving the robustness and accuracy of sports AQA. Moreover, the QAQA algorithm utilizes a multi-resolution approach, including coarsening, projection, and refinement, to significantly reduce computational complexity to O(n), alleviating the computational burden typically associated with long sequence analyses. Experimental results demonstrate that our method outperforms traditional methods in execution efficiency and scoring accuracy. The proposed system improves algorithmic performance and effectively contributes to intelligent sports training and education.

As the use of wearable electrocardiogram (ECG) data for modeling purposes continues to rise, there is a pressing need for signal quality assessment (SQA) algorithms capable of identifying segments of signal from which reliable data can be obtained. Manually annotated ECG data, obtained through expert visual inspection, is often used as reference in the development of ECG SQA algorithms. In this approach, the quality of a signal segment is assessed based on the level of noise present. Yet, the data extracted from noise-corrupted ECG signal segments might still be of sufficient accuracy depending on the target application. The current work proposes a paradigm shift by presenting a SQA algorithm that performs template matching and physiological feasibility checks to determine the quality of ECG signals acquired by textile-based wearable systems. Signal segments were classified into four different quality classes based on the estimated accuracy of RR intervals extracted from the signal segments of each class. Our findings show that the proposed SQA algorithm is effective in identifying ECG signal segments from which accurate RR intervals can be derived, and that the proportion of the data across the different classes is sensitive to different factors known to have an effect on signal quality.

Motion blur degrades face image quality and impairs recognition accuracy. This paper evaluates five face image quality assessment (FIQA) algorithms for motion blur detection, focusing on accuracy and demographic fairness. Experiments on the EDAMB and MST-E datasets employed Kullback–Leibler (KL) divergence to compare algorithm scores against expert consensus, partial area under the curve (pAUC) from error-versus-discard curves to report prediction of recognition performance, and the Gini coefficient to assess fairness. Densenet169 had the lowest KL divergence, while CNN-R showed the best predicting performance, achieving the lowest pAUC. Fusing CNN-R with Densenet161 further reduced the pAUC by 1.3%. The fairness analysis found that the Fourier Transform and CNN-R methods were the most fair, whereas Laplace was the least fair.

With the deepening of economic and cultural globalization and the popularity of cross-cultural communication, Mandarin, as a key carrier of Chinese culture, has become increasingly important for both domestic language education and foreign Chinese learning. However, traditional Mandarin teaching faces limitations such as difficulty in real-time detection of individual reading errors (e.g., missing reading and back reading) and heavy reliance on teacher experience, which restricts the efficiency of error correction and teaching quality. Meanwhile, with the rapid development of information technology, deep learning has shown strong advantages in speech signal processing, providing a new technical path for intelligent Mandarin reading error detection. Against this background, this study focuses on the detection of missing reading and back reading errors in Mandarin reading aloud, and conducts research based on the deep learning framework. To improve the accuracy and efficiency of error detection, this study takes the traditional Deep Neural Network (DNN) as the basic model, and optimizes the core Reading Quality Assessment (GOP) algorithm: first, it extends the GOP algorithm to the DNN-based error detection system, and modifies the GOP calculation formula by introducing the average posterior probability of non-target senones and weight coefficients, which solves the problem of unreliable phoneme segmentation caused by non-standard pronunciation; second, it addresses the issue that missing-reading errors of the current phoneme affect the GOP calculation of adjacent phonemes in the traditional framework, further optimizing the algorithm’s robustness. Additionally, this study introduces DNN adaptive technology based on KL divergence regularization to align the standard and non-standard reading models, enhancing the algorithm’s adaptability to different speakers. Experiments are conducted on two databases (MPE database for domestic Mandarin speakers and ICALL database for foreign Chinese learners). The results show that the improved GOP algorithm combined with DNN adaptive technology significantly outperforms traditional methods: compared with the GMM-CM algorithm, the accuracy and recall of error detection are increased by 13.4%; compared with the original DNN-GOP algorithm, the improved DNN-GOP2 algorithm reduces the Top1 error rate by 1.7% and the Top5 error rate by 2.0%. This study not only provides a more accurate and efficient technical solution for Mandarin reading error detection but also lays a foundation for the development of intelligent Mandarin teaching systems, which is of great significance for promoting the modernization of Mandarin teaching and the popularization of Chinese language education globally.

Solution of the problem of limiting the spread of the dynamic and acoustic emissions from rail and road traffic requires application of new elastomeric materials are characterized by both high acoustic insulation as well as being part of vibroisolation system between the source of vibration associated with the movement of vehicles and construction engineering structure. The article presents the results of experimental vibration isolation effectiveness of three types of rail sleepers used in vibration isolation systems of railway and tramway tracks . In the article the methodology of research and quality assessment algorithm vibration isolation system based on the recorded measurement data is also presented.

In the application of wrist-based Photoplethysmography (PPG) devices for health monitoring, assessing the quality of PPG signals is essential for accurately monitoring cardiovascular parameters. However, the wrist-based PPG signal is susceptible to motion and light interference in practical applications. A machine learning-based signal quality assessment algorithm for wrist PPG signals was proposed to improve the accuracy and reliability of the monitoring data. The algorithm's performance was evaluated on two datasets: the publicly available Wearable and Clinical Signals (WCS) dataset, containing 3,038 wrist-based PPG segments collected from 18 volunteers using an Empatica E4 device; our LAB dataset, comprising 2,426 wrist-based PPG segments acquired from 12 volunteers under varied interference conditions via a custom-developed wearable watch system. Data pre-processing encompassed denoising and normalization, followed by the extraction of 11 mathematical statistical features in time and frequency domains based on pulse wave morphology and 2 features based on template matching (Euclidean Distance and Correlation Coefficient). The classifier, constructed using the LightGBM algorithm, achieved high performance under rigorous leave-one-subject-out cross-validation (LOSO-CV) on the WCS dataset (accuracy = 92.6%, precision = 96.6%, recall = 89.8%, F1-score = 91.4%, AUC = 0.925) and the LAB dataset (accuracy = 96.1%, precision = 98.1%, recall = 95.2%, F1-score = 96.6%, AUC = 0.941). The results show that the machine learning algorithm for wrist-based PPG signal quality assessment, combining the mathematical statistical features in time and frequency domains and the template matching features, can effectively enhance the performance of signal quality assessment, and provides a powerful tool for improving the accuracy of wearable devices in cardiovascular health monitoring.

This study aims to develop tools for constructing an automated rating system to objectively assess vocational education quality in the Kyrgyz Republic. It led to the creation of a conceptual model for an automated rating system, identification of key evaluation criteria, and development of a framework for component interaction. Methods for automated data collection were developed, including database integration via a software interface, electronic questionnaires, and web scraping. Processes for data cleaning, normalisation, and weighting were established to prepare information for analysis. Algorithms were implemented to integrate quantitative and qualitative indicators in assessing educational institutions' performance. The conceptual model reflects the specific features of vocational education and the regional environment. Key methods include data cleaning, normalisation, assigning weights to criteria, and analysing institutional effectiveness. Algorithms for automated data collection via software interfaces and web scraping ensure access to up-to-date information from educational portals. Prohierarchycessing algorithms, such as data cleaning and normalisation, ensure high-quality data preparation. Quality assessment algorithms, based on hierar-chy analysis and efficiency evaluation methods, objectively incorporate qualitative and quantitative indicators like teaching quality, student performance, scientific publications, and material resources.

Recently, there has been increasing research on image quality assessment. Among the existing mainstream approaches, image feature extraction tends to be simplistic, leading to insufficient quality information extraction and underutilization of the extracted data. Additionally, the correlation between different regions of the image is often neglected. This study proposes an image quality assessment algorithm based on global-local feature fusion (IQA-GL). First, the global and local features of the image are extracted separately, and irrelevant information in the local features is filtered out. Then, a global-local feature fusion model is constructed to enhance the interaction of feature information and gather image quality data across all feature channels. Finally, the relationship between individual image patches and the global image is modeled, adjusting the weights of each image patch to aggregate a quality score for the global image. Experimental results show the IQA-GL performs excellently on public datasets. This study innovatively combines global and local features, offering a new perspective for image quality assessment.

VIF–TOPSIS coupling algorithm for image quality assessment in smart construction site management

With the swift progression of machine learning technology, its potential applications in education are vast. This study introduces a convolutional neural network (CNN)-based algorithm designed to evaluate the quality of reforms in physical education classroom teaching. The goal is to improve the effectiveness and impact of physical education instruction through technological advancements. The research initially scrutinizes the limitations associated with traditional methods of evaluating the quality of physical education classrooms and delves into the possibilities afforded by employing CNN for assessing teaching quality. By formulating and implementing a CNN model tailored to the context of physical education classes, this study adeptly analyzes real-time student movement patterns, participation levels, and interaction effects. The experimental outcomes demonstrate the algorithm’s effectiveness in discerning critical teaching elements within the classroom, including students’ enthusiasm for engagement in activities and teachers’ instructional methodologies. The algorithm enables a quantitative assessment of these components. Additionally, this research explores the practical implementation of the algorithm in teaching contexts, including providing real-time feedback to educators and adjusting teaching strategies based on assessment outcomes. The results offer a fresh perspective for enhancing the quality of physical education instruction and lay the foundation for the wider adoption of machine learning technologies in the education sector.

A majority of individuals with advanced Parkinson's disease develop severe gait and balance deficits that significantly restrict mobility, independence, and quality of life. Despite impressive technological advances, gait impairments remain poorly understood and notoriously hard to treat. Most attempts to characterize gait deficits or to identify neural biomarkers of gait dysfunction have been confined to well-controlled, sophisticated laboratory settings. However, such environments fail to capture many critical contextual factors that exacerbate gait deficits in real-world conditions, such as stress, fatigue, or unexpected distractions and environmental constraints. Replicating these unpredictable conditions in laboratory settings remains a significant challenge. To address this, we developed a wearable neuromonitoring platform integrating (i) smart, sensorized shoes with embedded microchips running A.I. algorithms for online gait quality assessment, along with (ii) wireless recordings of local field potentials from chronically implanted electrodes for deep brain stimulation. This platform enables continuous monitoring of motor and neural biomarkers of gait function and dysfunction across activities of daily life. 13 participants with advanced Parkinson's disease were monitored during a comprehensive set of out-of-laboratory locomotor tasks, both in the ON and OFF medication conditions. Our platform allowed to assess gait quality, quantify modulations in response to medication, fatigue, and context-specific demands, and to map the underlying neural correlates. This neuromonitoring platform offers valuable possibilities for refining neuromodulation protocols and supporting closed-loop therapies for gait impairments in real-life conditions.

Pancreatic cancer has the highest mortality rate among all major cancers, which highlights the urgent needs of non-invasive early detection. Circulating extracellular vesicles (EVs) have gained significant attention for discovering tumor biomarkers. However, isolating EVs with well-defined homogeneous populations from complex biological samples is challenging. Different methods have been found to derive different EV populations carrying different biomolecules, which significantly confound biomarker discovery for developing clinical diagnostics. Building a rigorous EV isolation and standardizing assessment platform associated with -omics is essential to overcome this challenge. We introduced a novel isolation approach using a pH-responsive peptide conjugated with NanoPom magnetic beads (ExCy) for homogeneous EV isolation. Additionally, we introduced the first statistical algorithm for EV quality assessment (ExoQuality Index, EQI), which enables multi-assay quantification to provide a consistent and accurate definition of EV purity and quality; ExoQuality’s algorithm intakes multi-assay information to deconvolute complex EV heterogeneity. We performed the next generation sequencing on EV RNAs from pancreatic cancer patient plasma using four isolation methods; results highlighting ExCy’s isolation and EQI assessment improved biomarker identification. We identified a novel EV biomarker for pancreatic tumor, ATP6V0b, validated with quantitative PCR (qPCR) by screening a pilot cohort of 22 plasma samples. 16 were from pancreatic cancer patients, 6 with matched tumor tissue, and 6 healthy plasma samples. Through modelling the ATP6V0B cycling threshold, we reported 3 models with AUCs between 0.86 and 0.88, showcasing an enabling and clinically translatable liquid biopsy approach for early detection of pancreatic cancer using circulating EVs.Graphical

Assessing the quality of complex technical products is a time-consuming task, which becomes more complicated when an assessment needs to be obtained at the design or modernization stage. The integral mobility index can play the role of a quality measure for automotive equipment. The term “mobility” in the works of Professor V.V. Belyakov is interpreted as characterizing the ability to perform the task assigned to the transport and technological machine with optimal adaptability to the operating conditions and technical condition of the machine itself. In previous works authors demonstrated the development and application of an algorithm for calculating the integral mobility index for transport of various operational and functional purposes. The aim of this work is to develop and apply a methodology for evaluating and improving the mobility of transport and technological machines based on the obtained algorithm for multi-criteria quality assessment, which involves the decomposition of an integral property into the simplest components, and then aggregation of empirical indicators by means of additive convolution, taking into account the weighting coefficients of the criteria. The algorithm was implemented in the Matlab Simulink simulation environment. The article shows the result of applying the obtained methodology, analyzes the possibility of increasing mobility by making changes to the design of transport and technological facilities and complexes. The wheeled snowmobile “Вея ЗВМ-39083” was taken as an object for research. As a result, threshold values of individual criteria were obtained, which make it possible to increase the value of the integral mobility index to a predetermined level. The developed methodology can form the basis of a product quality management system at the design stage of automotive equipment.

Fluorescence microscopy enabled by smartphone-coupled 3D instruments has shown utility in different biomedical applications ranging from diagnostics to biomanufacturing. Recently, we have designed and developed these devices and have demonstrated their utility in micro-nano particle sensing and leukocyte imaging. Here, we present a novel application for enhancing the imaging performance of smartphone fluorescence microscopes (SFM) and reducing their operational complexity. Computational noise correction is employed using 3D Averaging and 3D Gaussian filters of different kernel sizes (3 × 3 × 3, 7 × 7 × 7, 11 × 11 × 11, 15 × 15 × 15, and 21 × 21 × 21) and various standard deviations σ (for Gaussian only). Fluorescent beads of different sizes (8.3, 2, 1, 0.8 µm) were imaged using a custom-designed SFM. The application of the computational filters significantly enhanced the signal quality of particle detection in the captured fluorescent images. Amongst the Averaging filters, a kernel size of 21 × 21 × 21 produced the best results for all bead sizes, and similarly, amongst Gaussian filters, σ equal to 5 and a kernel size equal to 21 × 21 × 21 produced the best results. This visual improvement was then quantified by calculating the signal-difference-to-noise ratio (SDNR) and contrast-to-noise ratio (CNR) of filtered and unfiltered original images using a custom-developed quality assessment algorithm (AQAFI). Lastly, noise correction using Averaging and Gaussian filters with the previously identified optimal parameters was applied to images of fluorescently tagged human peripheral blood leukocytes captured using an SFM under various conditions. The ubiquitous nature and simplistic application of these filters enable their utility with a range of existing fluorescence microscope designs, thus allowing us to enhance their imaging capabilities.

Physical rehabilitation exercises suggested by healthcare professionals can help recovery from various musculoskeletal disorders and prevent re-injury. However, patients' engagement tends to decrease over time without direct supervision, which is why there is a need for an automated monitoring system. In recent years, there has been great progress in quality assessment of physical rehabilitation exercises. Most of them only provide a binary classification if the performance is correct or incorrect, and a few provide a continuous score. This information is not sufficient for patients to improve their performance. In this work, we propose an algorithm for error classification of rehabilitation exercises, thus making the first step toward more detailed feedback to patients. We focus on skeleton-based exercise assessment, which utilizes human pose estimation to evaluate motion. Inspired by recent algorithms for quality assessment during rehabilitation exercises, we propose a Transformer-based model for the described classification. Our model is inspired by the HyperFormer method for human action recognition, and adapted to our problem and dataset. The evaluation is done on the KERAAL dataset, as it is the only medical dataset with clear error labels for the exercises, and our model significantly surpasses state-of-the-art methods. Furthermore, we bridge the gap towards better feedback to the patients by presenting a way to calculate the importance of joints for each exercise.

BackgroundAssociation and cooperation among structural domains play an important role in protein function and drug design. Despite remarkable advancements in highly accurate single-domain protein structure prediction through the collaborative efforts of the community using deep learning, challenges still exist in predicting multi-domain protein structures when the evolutionary signal for a given domain pair is weak or the protein structure is large.ResultsTo alleviate the above challenges, we proposed M-DeepAssembly, a protocol based on multi-objective protein conformation sampling algorithm for multi-domain protein structure prediction. Firstly, the inter-domain interactions and full-length sequence distance features are extracted through DeepAssembly and AlphaFold2, respectively. Secondly, subject to these features, we constructed a multi-objective energy model and designed a sampling algorithm for exploring and exploiting conformational space to generate ensembles. Finally, the output protein structure was selected from the ensembles using our in-house developed model quality assessment algorithm. On the test set of 164 multi-domain proteins, the results show that the average TM-score of M-DeepAssembly is 15.4% and 2.0% higher than AlphaFold2 and DeepAssembly, respectively. It is worth noting that there are models with higher accuracy in ensembles, achieving an improvement of 20.3% and 6.4% relative to the two baseline methods, although these models were not selected. Furthermore, when compared to the prediction results of AlphaFold2 for CASP15 multi-domain targets, M-DeepAssembly demonstrates certain performance advantages.ConclusionsM-DeepAssembly provides a distinctive multi-domain protein assembly algorithm, which can alleviate the current challenges of weak evolutionary signals and large structures to some extent by forming diverse ensembles using multi-objective protein conformation sampling algorithm. The proposed method contributes to exploring the functions of multi-domain proteins, especially providing new insights into targets with multiple conformational states.

Exploration of simulated human olfactory system and its integration with machine learning algorithms for food quality assessment: A review

Assessing the quality of artificial intelligence-generated images (AIGIs) plays a crucial role in their application in real-world scenarios. However, traditional image quality assessment (IQA) algorithms primarily focus on low-level visual perception, while existing IQA works on AIGIs overemphasize the generated content itself, neglecting its effectiveness in real-world applications. To bridge this gap, we propose AIGI-VC, a quality assessment database for AI-Generated Images in Visual Communication, which studies the communicability of AIGIs in the advertising field from the perspectives of information clarity and emotional interaction. The dataset consists of 2,500 images spanning 14 advertisement topics and 8 emotion types. It provides coarse-grained human preference annotations and fine-grained preference descriptions, benchmarking the abilities of IQA methods in preference prediction, interpretation, and reasoning. We conduct an empirical study of existing representative IQA methods and large multi-modal models on the AIGI-VC dataset, uncovering their strengths and weaknesses.