Real-time Quality Control Research Articles

Hybrid-DC: A Hybrid Framework Using ResNet-50 and Vision Transformer for Steel Surface Defect Classification in the Rolling Process

This study introduces Hybrid-DC, a hybrid deep-learning model integrating ResNet-50 and Vision Transformer (ViT) for high-accuracy steel surface defect classification. Hybrid-DC leverages ResNet-50 for efficient feature extraction at both low and high levels and utilizes ViT’s global context learning to enhance classification precision. A unique hybrid attention layer and an attention fusion mechanism enable Hybrid-DC to adapt to the complex, variable patterns typical of steel surface defects. Experimental evaluations demonstrate that Hybrid-DC achieves substantial accuracy improvements and significantly reduced loss compared to traditional models like MobileNetV2 and ResNet, with a validation accuracy reaching 0.9944. The results suggest that this model, characterized by rapid convergence and stable learning, can be applied for real-time quality control in steel manufacturing and other high-precision industries, enhancing automated defect detection efficiency.

Free Electron Density Gradients Enhanced Biosensor for Ultrasensitive and Accurate Affinity Assessment of the Immunotherapy Drugs.

Accurate affinity assessments play an important role in drug discovery, screening, and efficacy evaluation. Label-free affinity biosensors are recognized as a dependable and standard technology for addressing this challenge. This study constructs a free electron density gradient-enhanced meta-surface plasmon resonance (FED-MSPR) biosensor through a finite-difference time-domain simulation model, the biosensor demonstrates superior detection performance in accurately determining affinity and kinetic rate constants. By controlling the dielectric properties of the metal on the surface of the nanocup arrays, the plasmon resonance effects are easily tuned without changing the nanostructure design. Compared with the single-layer gold chip, the triple-layer FED-MSPR chip demonstrated a four-fold improvement in resolution at the optimal resonance peak. Additionally, the sensitivity and figure of merit (FOM) of the multi-layer chip increased by 3.5 and 7.99 times, respectively. Following modification with high- and low-staggered carboxylation, the noise-signal ratio and baseline stability of the real-time kinetic curves based on these chips are significantly enhanced. The developed carboxylation FED-MSPR platform is successfully used to perform affinity assays for Adalimumab and TNF-α protein, resulting in favorable dynamic curves. These findings validate the proposed FED-MSPR biosensor platform as cost-effective, rapid, sensitive, and label-free, facilitating real-time quality control in drug development.

In-line porosity and hardness monitoring of tablets by means of optical coherence tomography

In-line monitoring of critical quality attributes (CQAs) during a tableting process is an essential step toward a real-time release strategy. Such CQAs can be the tablet mass, the API content, dissolution, hardness and tensile strength. Since dissolution testing is laborious and time-consuming and cannot be performed in-line, it is desirable to replace dissolution testing with predictive models based on other CQAs that affect the dissolution characteristics, such as the tablet porosity and hardness. Traditionally, porosity is determined offline via gas adsorption methods or other techniques, such as Terahertz spectroscopy or gas in scattering media absorption spectroscopy. Tablet hardness is typically established using a hardness tester. While these destructive tests can readily be performed at-line, they have limited applicability in in-line settings for a high-percentage inspection. Optical coherence tomography (OCT) has recently been proposed as a possible tool for determining quality attributes. This work describes the first application of OCT for the prediction of tablet porosity and hardness. OCT measurements of tablets produced in a ConsiGma 25™ tableting line and a Stylcam 200R compaction simulator in several compaction force settings were performed and correlated with the porosity and hardness. It was demonstrated that OCT can easily be installed in-line and provide real-time information about critical material attributes. These insights confirm the applicability of OCT as a real-time quality control tool and its potential to replace time-consuming and destructive offline measurements.

B-130 Time-of-day dependence of running averages for some components of metabolic panels at our hospital: relation to strategy for patient-based quality control

Abstract Background Our hospital system recently reorganized its lab services so as to restrict in-house testing almost exclusively to inpatient testing. This disruption in patient mix caused us to reassess properties of running averages for our most common analytes, for potential use of such data in real-time patient-based quality control (PBQC). In particular, we examined whether there were analytes for which there was significant dependence of running averages on 24-h clock time (time-of-day, TOD). Methods We performed a retrospective analysis of 3-months' data (FY24, Q2) for patient results of 13 components of the Basic Metabolic Panel (BMP) and Comprehensive Metabolic Panel: Albumin*, Alkaline phosphatase, Alanine amino transferase, Aspartate aminotransferase*, Calcium*, Chloride*, CO2*, Creatinine, Potassium*, Total Protein*, Sodium, Total Bilirubin, and Urea. Running averages for 20 consecutive results (20-mers) were computed for data restricted to results within reference intervals. This produced an overall mean (X) and standard-deviation (SD) of 20-mers for each analyte. We then computed the average 20-mer result (Y) reported within 1-h bins across 24-hour clock time (t). Y(t) was regarded as having TOD-dependence if: (a) there was an 8-h interval of contiguous results for Y that were strictly either below or above X; and (b) either nadir or apex values for |Y-X| exceeded 0.5 SD. Results Seven analytes (*, above) demonstrated TOD-dependence of running means for 20-mers according to criteria given above. An example of TOD-dependent Y(t) is shown in Figure (Albumin). Conclusions For our hospital, TOD-dependence of running means was identified for 7 of 13 metabolic panel analytes. TOD-dependence of running means is a circumstance in which use of only a fixed target for running means in PBQC would lead to inherent TOD-dependent variation in capability for systematic error detection. Use of TOD-dependent targets for PBQC would be appropriate in these cases.

Real world feasibility of patient-based real time quality control (PBRTQC) using five analytes in a South African laboratory

BackgroundPatient-based real-time quality control (PBRTQC) identifies possible bias in methods by utilising shifts in trend of statistical measures in laboratory results. In this study we aimed to compare and optimize various PBRTQC procedures for serum alanine aminotransferase, albumin, calcium, ferritin and sodium. MethodsIn a bias simulation study, we added artificial bias to intervals of patient data and then evaluated the efficiency with which various PBRTQC procedures were able to detect this bias. PBRTQC procedures used included block size, moving statistic calculation, control limits as well as truncation limits. The number of patients till error detection, the false alarm rate as well as validation charts were utilised to select the optimal PBRTQC procedure for each analyte. ResultsThe optimal PBRTQC procedures identified for each analyte were: ALT − MA T0 50 MaxMin; Albumin − MA T5 100 Perc; Calcium − MM T5 50 Perc; Ferritin − MA T0 100 MaxMin; and Sodium − MA T5 75 MaxMin. (T, Truncation limits; Control limits- MA, Moving Average, MM, Moving median; Perc, percentile) ConclusionsThe use of large, real patient datasets allows for the reliable determination of laboratory-specific PBRTQC procedures. This study demonstrates that the moving average calculation excels in both normal and transformed analyte distributions. Whilst, the benefits of PBRTQC procedures are proven, the complex and time-consuming optimisation process may be a barrier to the rapid implementation in under-resourced countries.

Soft sensor for melting flow rate prediction based on data-enhanced classification method

Abstract The melt flow rate (MFR) is a crucial quality indicator for polyolefin products. However, the offline laboratory measurement method suffers from significant time delays between sampling and obtaining test values, making it inconvenient for real-time product quality control. Additionally, installing an online analyzer can be expensive. In this study, we propose a soft sensor method that combines mechanism analysis and data-driven approaches to predict the MFR in the extruder process. Firstly, we analyze the physical mechanism of the extruder process, and the time series data are processed into independent enhanced data sets by reasonably selecting data enhancement parameters. Secondly, we statistically model the distribution of particle MFR and classify the data into four sections. We train Deep Neural Network (DNN) and Gradient Boosting Decision Tree (GBDT) regression models separately for low and non-low MFR sections while also developing a global classification model to invoke corresponding regression models when needed. The results demonstrate that our classification model achieves 100% accuracy in distinguishing between low and non-low MFR categories; furthermore, our regression models achieve an RMSE of 0.600, and an R2 value of 0.998, outperforming existing models on various indicators. Finally, considering viscosity variations in polyolefin products' influence on prediction performance, we recommend setting data enhancement parameters contrary to particle MFR trends.

Efficient fabric anomaly detection: A transfer learning framework with expedited training times

In industrial quality control, anomaly detection plays a critical role in identifying defective products. However, because of the rarity and time-consuming nature of defect collection, training models often rely solely on defect-free samples. This necessitates the use of unsupervised anomaly-detection techniques trained exclusively on defect-free data. Alternatively, defect data can be synthesized to augment the dataset with defective samples. In the textile industry, expeditious model training is crucial to ensure a smooth production flow. Unfortunately, most unsupervised methods require extensive training time. This paper proposes a novel transfer learning approach designed to achieve training times in seconds while effectively adapting the model to the target domain of fabric anomaly detection. The key contributions of our method include significantly reduced training times, up to 10 times faster than current state-of-the-art methods, and comparable performance in anomaly detection, achieving results on par with state-of-the-art approaches on benchmark datasets (MVTEC Anomaly Detection, TILDA, AITEX and DAGM). Additionally, our approach improves inference times, ensuring expedited and efficient anomaly detection during production. The proposed method offers a practical and efficient solution for real-time industrial quality control.

Transformative Role of Artificial Intelligence in the Pharmaceutical Sector

Background: The pharmaceutical sector is a critical component of healthcare, driving innovation in drug discovery, development, and delivery. With the increasing integration of artificial intelligence (AI), digital health technologies, and biotechnology, the industry is transforming rapidly. This review examines the key areas of the pharmaceutical industry and highlights the growing impact of AI in enhancing various processes, from drug discovery to clinical trials. To explore the applications of AI in drug discovery, development, manufacturing, clinical trials, personalized medicine, and regulatory compliance. This review also addresses the challenges, such as data privacy and interoperability, that accompany the adoption of AI in the pharmaceutical sector. Methods: A comprehensive review of existing literature and case studies on the application of AI in pharmaceutical research and operations was conducted. Key areas of focus include AI's role in predictive analytics, target identification, manufacturing, supply chain management, clinical trial optimization, and pharmacovigilance. Results: AI significantly enhances drug discovery by improving target identification, predictive modeling, and high-throughput screening. It optimizes manufacturing through real-time quality control and process automation. In clinical trials, AI facilitates patient recruitment and adaptive trial designs, while in personalized medicine, it enables biomarker discovery and treatment optimization. AI also supports regulatory compliance through automated monitoring and risk assessment. Conclusion: AI is transforming the pharmaceutical sector, making processes more efficient, precise, and tailored to individual patients. However, challenges such as data privacy, ethical considerations, and interoperability must be addressed to fully harness AI's potential. Standardization and collaboration will be essential in driving the next phase of innovation in pharmaceutical development and healthcare delivery.

Synchronisation of thermal imaging and multi-channel EIS to interpret planar array PCB fuel cell performance

The planar design of printed circuit board (PCB) fuel cells offers the advantages of simple structure, rapid prototyping and easy manufacturing. However, uneven reactants and heat distribution in each PCB cell module often result in inconsistent overall performance. Current characterisation methods mainly focus on the study of the overall performance and lack the analysis of individual cells, which provide a limited understanding of localised electrochemical and thermal effects that might play a critical part in further optimising and diagnosing the cell performance of this category. We have developed a multi-channel impedance diagnostic technique for PCB fuel cells. Combining with thermal imaging, we could simultaneously diagnose individual open cathode cell modules by measuring the heat distribution and electrochemical impedance spectra (EIS) of 11 PCB cell modules. Hence, the correlation between heat distribution, internal resistance and performance of different sections of PCB fuel cells can be established. By comparing the fuel cell’s pristine and degraded, we demonstrated how this non-destructive technique could identify and locate cell modules with deteriorated performance, which would be beneficial to real-time diagnostics, quality control and optimisation. The multi-channel impedance measurement takes several seconds only through optimised multi-frequency perturbation and can be readily extended to stacks containing more cells by simply increasing the number of voltage sensors, which can be further adapted to other electrochemical devices.

Machine Learning for Patient-Based Real-Time Quality Control (PBRTQC), Analytical and Preanalytical Error Detection in Clinical Laboratory.

The rapidly evolving field of machine learning (ML), along with artificial intelligence in a broad sense, is revolutionising many areas of healthcare, including laboratory medicine. The amalgamation of the fields of ML and patient-based real-time quality control (PBRTQC) processes could improve the traditional PBRTQC and error detection algorithms in the laboratory. This narrative review discusses published studies on using ML for the detection of systematic errors, non-systematic errors, and combinations of different types of errors in clinical laboratories. The studies discussed used ML for detecting bias, the requirement for re-calibration, samples contaminated with intravenous fluid or EDTA, delayed sample analysis, wrong-blood-in-tube errors, interference or a combination of different types of errors, by comparing the performance of ML models with human validators or traditional PBRTQC algorithms. Advantages, limitations, the creation of standardised ML models, ethical and regulatory aspects and potential future developments have also been discussed in brief.

Construction and application of an ICU nursing electronic medical record quality control system in a Chinese tertiary hospital: a prospective controlled trial

BackgroundICU nurses provide critical care and meticulously document electronic medical records (EMRs), tracking vital signs, interventions, and medication hourly. Despite China’s ICUs effectively integrating real-time monitor and ventilator data into EMRs, challenges persist. Patient movements can introduce inaccuracies, and the demands of critical care may lead nurses to miss assessments like pain and nutrition. Traditional manual EMR verification is inefficient and error-prone, highlighting the urgent need for standardized, technology-aided EMR practices in ICU nursing.ObjectiveThis study aimed to describe the development and evaluation of an electronic medical records quality control system implemented in a Chinese tertiary care ICU setting, where current practices impact the accuracy of electronic medical records.MethodsA prospective controlled trial was conducted with 600 ICU patients in Zhejiang Province from January to December 2023. An automated EMR quality control system was implemented in July 2023, facilitating real-time data collection and quality control for vital signs, medication management, and nursing evaluations.ResultsAfter implementing the ICU nursing electronic medical record quality control system, the prevalence of false data on vital signs decreased from 9 to 1.33%. Additionally, the incidence of incomplete medication administration dropped from 3.33 to 1.67%, and the rate of missing evaluations of assessment items in EMRs was reduced from 8 to 1.33%. Besides, the average time spent on quality control of the electronic medical records was 62 (48,76) seconds per record, which was significantly lower than the 264 (195.5,337.5) seconds using the traditional method. The nurses’ satisfaction with the nursing electronic medical record quality control was (105.73 ± 9.31).ConclusionsThe ICU nursing electronic medical record quality control system has led to substantial improvements in the quality and reliability of EMRs. The reduction in false data on vital signs, instances of incomplete medication administration, and missing evaluations of assessment items demonstrates the system’s positive impact on nursing documentation practices. These improvements not only enhance the accuracy of patient records but also contribute to better patient care and safety within the ICU setting.

Noninvasive, label-free image approaches to predict multimodal molecular markers in pluripotency assessment

Manufacturing regenerative medicine requires continuous monitoring of pluripotent cell culture and quality assessment while eliminating cell destruction and contaminants. In this study, we employed a novel method to monitor the pluripotency of stem cells through image analysis, avoiding the traditionally used invasive procedures. This approach employs machine learning algorithms to analyze stem cell images to predict the expression of pluripotency markers, such as OCT4 and NANOG, without physically interacting with or harming cells. We cultured induced pluripotent stem cells under various conditions to induce different pluripotent states and imaged the cells using bright-field microscopy. Pluripotency states of induced pluripotent stem cells were assessed using invasive methods, including qPCR, immunostaining, flow cytometry, and RNA sequencing. Unsupervised and semi-supervised learning models were applied to evaluate the results and accurately predict the pluripotency of the cells using only image analysis. Our approach directly links images to invasive assessment results, making the analysis of cell labeling and annotation of cells in images by experts dispensable. This core achievement not only contributes for safer and more reliable stem cell research but also opens new avenues for real-time monitoring and quality control in regenerative medicine manufacturing. Our research fills an important gap in the field by providing a viable, noninvasive alternative to traditional invasive methods for assessing pluripotency. This innovation is expected to make a significant contribution to improving regenerative medicine manufacturing because it will enable a more detailed and feasible understanding of cellular status during the manufacturing process.

Application of Patient-based Real-time Quality Control

Application of Patient-based Real-time Quality Control

Establishment and evaluation of Voting algorithm-based internal quality control (ViQC), a patient-based real-time quality control

BackgroundPatient-Based Real-Time Quality Control (PBRTQC) has emerged as a supplementary programme to traditional internal quality control (iQC) mechanisms. Despite its growing popularity, practical applications in clinical settings reveal several challenges. The primary objective of this research is to introduce and develop an Artificial Intelligence (AI)-based method, named Voting algorithm based iQC (ViQC), designed to enhance the precision and reliability of existing PBRTQC systems. MethodsIn this study, we conducted a retrospective analysis of 111,925 inpatient serum glucose test results from Nanjing Drum Tower Hospital, Nanjing, China, to provide an unbiased data set. The Voting iQC (ViQC) algorithm, established by the principles of the Voting algorithm, was then developed. Its analytical performance was evaluated through the calculation of random errors (RE). Subsequently, its clinical efficacy was assessed by comparison with five statistical algorithms: Moving Average (MA), Exponentially Weighted Moving Average (EWMA), Moving Median (movMed, MM), Moving Quartile (MQ), and Moving Standard Deviation (MovSD). ResultsThe ViQC model incorporates a variety of machine learning models, including logistic regression, Bayesian methods, K-Nearest Neighbor, decision trees, random forests, and gradient boosting decision trees, to establish a robust predictive framework. This model consistently maintains a false positive rate below 0.002 across all six evaluated error factors, showcasing exceptional precision. Notably, its performance further excels with an error factor of 3.0, where the false positive rate drops below 0.001, and achieves an accuracy rate as high as 0.965 at an error factor of 2.0. The classification effectiveness of ViQC model is evaluated by an area under the curve (AUC) exceeding 0.97 for all error factors. In comparison to five conventional PBRTQC statistical methods, ViQC significantly enhances error detection efficiency, maximum reducing the trimmed average number of patient samples required for detecting errors from 724 to 168, thereby affirming its superior error detection capability. ConclusionThe new established PBRTQC using artificial intelligence yielded satisfactory performance compared to the traditional PBBTQC in real world setting.

Comparative study on the quality control effectiveness of AI-PBRTQC and traditional PBRTQC model in identifying quality risks.

We compared the quality control efficiency of artificial intelligence-patient-based real-time quality control (AI-PBRTQC) and traditional PBRTQC in laboratories to create favorable conditions for the broader application of PBRTQC in clinical laboratories. In the present study, the data of patients with total thyroxine (TT4), anti-Müllerian hormone (AMH), alanine aminotransferase (ALT), total cholesterol (TC), urea, and albumin (ALB) over five months were categorized into two groups: AI-PBRTQC group and traditional PBRTQC group. The Box-Cox transformation method estimated truncation ranges in the conventional PBRTQC group. In contrast, in the AI-PBRTQC group, the PBRTQC software platform intelligently selected the truncation ranges. We developed various validation models by incorporating different weighting factors, denoted as λ. Error detection, false positive rate, false negative rate, average number of the patient sample until error detection, and area under the curve were employed to evaluate the optimal PBRTQC model in this study. This study provides evidence of the effectiveness of AI-PBRTQC in identifying quality risks by analyzing quality risk cases. The optimal parameter setting scheme for PBRTQC is TT4 (78-186), λ = 0.03; AMH (0.02-2.96), λ = 0.02; ALT (10-25), λ = 0.02; TC (2.84-5.87), λ = 0.02; urea (3.5-6.6), λ = 0.02; ALB (43-52), λ = 0.05. The AI-PBRTQC group was more efficient in identifying quality risks than the conventional PBRTQC. AI-PBRTQC can also effectively identify quality risks in a small number of samples. AI-PBRTQC can be used to determine quality risks in both biochemistry and immunology analytes. AI-PBRTQC identifies quality risks such as reagent calibration, onboard time, and brand changes.

Active Learning in Feature Extraction for Glass-in-Glass Detection

In the food industry, ensuring product quality is crucial due to potential hazards to consumers. Though metallic contaminants are easily detected, identifying non-metallic ones like wood, plastic, or glass remains challenging and poses health risks. X-ray-based quality control systems offer deeper product inspection than RGB cameras, making them suitable for detecting various contaminants. However, acquiring sufficient defective samples for classification is costly and time-consuming. To address this, we propose an anomaly detection system requiring only non-defective samples, automatically classifying anything not recognized as good as defective. Our system, employing active learning on X-ray images, efficiently detects defects like glass fragments in food products. By fine tuning a feature extractor and autoencoder based on non-defective samples, our method improves classification accuracy while minimizing the need for manual intervention over time. The system achieves a 97.4% detection rate for foreign glass bodies in glass jars, offering a fast and effective solution for real-time quality control on production lines.

Enhanced patient-based real-time quality control using the graph-based anomaly detection.

Patient-based real-time quality control (PBRTQC) is an alternative tool for laboratories that has gained increasing attention. Despite the progress made by using various algorithms, the problems of data volume imbalance between in-control and out-of-control results, as well as the issue of variation remain challenges. We propose a novel integrated framework using anomaly detection and graph neural network, combining clinical variables and statistical algorithms, to improve the error detection performance of patient-based quality control. The testing results of three representative analytes (sodium, potassium, and calcium) and eight independent variables of patients (test date, time, gender, age, department, patient type, and reference interval limits) were collected. Graph-based anomaly detection network was modeled and used to generate control limits. Proportional and random errors were simulated for performance evaluation. Five mainstream PBRTQC statistical algorithms were chosen for comparison. The framework of a patient-based graph anomaly detection network for real-time quality control (PGADQC) was established and proven feasible for error detection. Compared with classic PBRTQC, the PGADQC showed a more balanced performance for both positive and negative biases. For different analytes, the average number of patient samples until error detection (ANPed) of PGADQC decreased variably, and reductions could reach up to approximately 95 % at a small bias of 0.02 taking calcium as an example. The PGADQC is an effective framework for patient-based quality control, integrating statistical and artificial intelligence algorithms. It improves error detection in a data-driven fashion and provides a new approach for PBRTQC from the data science perspective.

High-resolution imaging for in-situ non-destructive testing by quantitative lensless digital holography

Quantitative imaging technologies for in-situ non-destructive testing (NDT) demand high-resolution, wide-field, and stable metrology capabilities. Moreover, live processing and automation are vital for real-time quality control and inspection. Conventional methods use complex optical setups, resulting in large, immobile systems which can solely operate within controlled environmental conditions due to temporal instabilities, rendering them unsuitable for in-situ measurements of micro-to nano-scale physical phenomena. This article delves into the multiphysics application of lensless digital holography, emphasizing its metrological capacity for various in-situ scenarios, while acknowledging and characterizing the differing constraints imposed by various physical phenomena, both transient and steady-state. The digital reconstruction of holograms is computed in real-time, and numerical focusing capabilities allow for instantaneous retrieval of the optical phase at various working distances without the need of complex optical setups, making lensless digital holography well-suited for in-situ quantitative imaging under various types of environments. Current NDT capabilities are demonstrated, including high-resolution and real-time reconstructions, simultaneous measurements for comparative metrology, and practical applications ranging from vibrations and acoustics to thermo-mechanics. Furthermore, methodologies to enhance overall metrology capabilities are exploited, addressing the study of existing physical phenomena, thereby expanding the applicability of holographic techniques across diverse industrial sectors.

Real-time quality assurance and quality control for a high frequency radar network

This paper recommends end to end quality assurance methods and quality control tests for High Frequency Radar Networks. We focus on the network that is operated by the Mid Atlantic Regional Association Coastal Ocean Observing System (MARACOOS). The network currently consists of 38 radars making real-time measurements of the surface currents over the continental shelf for a variety of applications including search and rescue planning, oil spill trajectory modelling and providing a transport context for marine biodiversity observing networks. MARACOOS has been delivering surface current measurements to the United States Coast Guard (USCG) since May 2009. Data quality is important for all applications; however, since the USCG uses this surface current information to plan life-saving missions, delivery of the best quality data is crucial. We have mapped the components of the HF radar data processing chain onto the data levels presented in the NASA Earth Science Reference Handbook and have applied quality assurance and quality control techniques at each data level to achieve the highest quality data. There are approximately 400 High Frequency radars (HFRs) deployed globally and the presented techniques can provide a foundation for data quality checks and standardization of the data collected by the large number of systems operating today.

Risk-based quality control plan incorporating patient based real time quality control using moving average and internal quality control

Risk-based quality control plan incorporating patient based real time quality control using moving average and internal quality control