Performance Of Model Research Articles

A Preoperative Prediction Model for Lymph Node Metastasis in Patients with Gastric Cancer Using a Machine Learning-based Ultrasomics Approach.

This study aimed to develop an ultrasomics model for predicting lymph node metastasis preoperative in patients with gastric cancer (GC). This study enrolled GC patients who underwent preoperative ultrasound examination. Manual segmentation of the region of interest (ROI) was performed by an experienced radiologist to extract radiomics features using the Pyradiomics software. The Z-score algorithm was used for feature normalization, followed by the Wilcoxon test to identify the most informative features. Linear prediction models were constructed using the least absolute shrinkage and selection operator (LASSO). The performance of the ultrasomics model was evaluated using the area under curve (AUC), sensitivity, specificity, and the corresponding 95% confidence intervals (CIs). A total of 464 GC patients (mean age: 60.4 years ±11.3 [SD]; 328 men [70.7%]) were analyzed, of whom 291 had lymph node metastasis. The patients were randomly assigned to either the training (n=324) or test (n=140) sets, using a 7:3 ratio. An ultrasomics model that consisted of 19 radiomics features was developed using Wilcoxon and LASSO algorithms in the training set. Our ultrasomics model showed moderate performance for lymph node metastasis prediction in both the training (AUC: 0.802, 95%CI: 0.752-0.851, P<0.001) and test sets (AUC: 0.802, 95%CI: 0.724-0.879, P<0.001). The calibration curve analysis indicated good agreement between the predicted probabilities of ultrasomics and actual lymph node metastasis status. Our study highlights the potential of a machine learning-based ultrasomics model in predicting lymph node metastasis in GC patients, offering implications for personalized therapy approaches.

Improved Association Rule Mining-based Data Sanitisation with Blockchain for Secured Supply Chain Management

A supply chain management (SCM) method must include information sharing as a vital component in order to improve supply chain performance and boost an organisation’s strategic advantage. Since, due to a lack of trust concern over information leakage, and security breaches by nefarious individuals or groups, several organisations are hesitant to share information with their supply chain partners. This work presents a new supply chain management-based secure data transmission method. By using blockchain-based data storage, it is assumed that the manufacturers, suppliers, and customers would transfer data that must be kept private during transmission. As a consequence, this paper aims to provide an improved association rule mining with a data sanitisation scheme with an improved Apriori algorithm used in the proposed data sanitisation process. In particular, the Long Short-Term Memory (LSTM) will generate keys by considering the objective relying on the value of the preservation ratio, false rule generation, hiding failure, and degree of modification. The weights are adjusted via a novel Minkowski distance-based Namib beetle optimisation (MDNBO) technique, which also improves the performance of the LSTM model. The reverse process of encryption occurs when encrypted data are restored at the receiving end. By contrasting it with the old methods with regard to security as well, the proposed protected data in SCM with blockchain technology will be proved to be efficient.

Global prototype distillation for heterogeneous federated learning.

Federated learning is a distributed machine learning paradigm where the goal is to collaboratively train a high quality global model while private training data remains local over distributed clients. However, heterogenous data distribution over clients is severely challenging for federated learning system, which severely damage the quality of model. In order to address this challenge, we propose global prototype distillation (FedGPD) for heterogenous federated learning to improve performance of global model. The intuition is to use global class prototypes as knowledge to instruct local training on client side. Eventually, local objectives will be consistent with the global optima so that FedGPD learns an improved global model. Experiments show that FedGPD outperforms previous state-of-art methods by 0.22% ~1.28% in terms of average accuracy on representative benchmark datasets.

Regional evaluation of global geopotential models and three types of digital elevation models with ground-based gravity and GNSS/levelling data using several techniques over Sudan

Abstract To evaluate the performance of the global geopotential models (GGMs) in a more unbiased way, ground-based gravity and GNSS/levelling datasets are highly required. In this study, the eight latest releases of the satellite-only and combined GGMs are evaluated on the regional scale using the available terrestrial gravity and GNSS/Levelling data over Sudan, considering the spectral consistency issue by applying the spectral enhancement method (SEM). The evaluation process consists of three stages: firstly, the eight GGMs are evaluated globally with each other by using different degree variances in terms of geoid heights, gravity anomalies, and signal-to-noise ratio (SNR); secondly, the GGMs are compared against the Earth Gravitational Model 2008 (EGM2008) on a regional scale over Sudan; thirdly, apply the SEM strategy by incorporating high (SEM_WITHOUT_RTM technique) and very-high (SEM technique) frequencies of the gravity field spectrum from the EGM2008 and high-resolution residual terrain model (RTM), respectively. For reliable robustness of the latter evaluation process, three different DEMs are used, namely, SRTM30, ASTER30, and GTOPO30. Our findings on the evaluation process using SEM_WITHOUT_RTM technique show improved gravity anomalies solutions regarding differences of standard deviations (STD) from 19–20.7 mGal to about 14 mGal. When applying the SEM technique, more improvements are achieved, providing STD differences in gravity anomalies and geoid heights of about 12 mGal and 45 cm, respectively. Among the three applied DEMs, it has been found that despite the slight refinements, the ASTER30 and GTOPO30 models show better performance than the SRTM30 model.

Population pharmacokinetics and dosage optimization of linezolid in Chinese older patients.

To assess the pharmacokinetics and pharmacodynamics of linezolid in a retrospective cohort of hospitalized Chinese older patients. Patients > 60years of age, who received intravenous linezolid (600mg), were included. A population pharmacokinetics (PPK) model was established using nonlinear mixed-effects modeling. The predictive performance of the final model was assessed using goodness-of-fit plots, bootstrap analyses, and visual predictive checks. Monte Carlo simulations were used to evaluate the achievement of a pharmacodynamics target for the area under the serum concentration-time curve/minimum inhibitory concentration (AUC0-24/MIC). A total of 210 samples were collected from 120 patients. A one-compartment PPK model with linear elimination best predicted the linezolid plasma concentrations. Linezolid clearance (CL) was 4.22 L h-1 and volume of distribution (Vd) was 45.80 L; serum uric acid (SUA) was a significant covariate of CL. The results of this study indicated that the standard dose was associated with a risk of overexposure in older patients, particularly those with high SUA values; these patients would benefit from a lower dose (300mg every 12h).

Priority Setting and Resource Allocation in Coastal Local Government Marine Regulatory Reform: Application of Machine Learning in Resource Optimization

This study investigates the prioritization and resource allocation strategies adopted by the coastal local governments of Qingdao, Dalian, and Xiamen in the context of marine regulatory reform aimed at enhancing regulatory efficiency. Data on relevant opinions, departmental requirements, and existing resource allocations were collected through a questionnaire survey. A backpropagation (BP) neural network was then applied to analyze the survey data, prioritize regulatory tasks, and propose resource allocation schemes. The findings demonstrate that integrating machine learning into marine regulation can significantly improve resource utilization efficiency, optimize task execution sequences, and enhance the scientific and refined nature of regulatory work. The BP neural network model exhibited strong predictive capabilities on the training set and demonstrated good generalization abilities on the test set. The performance of the BP neural network model varied slightly across different management levels. For the management level, the accuracy, precision, and recall rates were 85%, 88%, and 82%, respectively. For the supervisory level, these metrics were 81%, 83%, and 78%, respectively. At the employee level, the accuracy, precision, and recall rates were 79%, 81%, and 76%, respectively. These results indicate that the BP neural network model can provide differentiated resource allocation recommendations based on the needs of different management levels. Additionally, the model’s performance was assessed based on the employees’ years of experience. For employees with 0–5 years of experience, the accuracy, precision, and recall rates were 82%, 84%, and 79%, respectively. For those with 5–10 years of experience, the metrics were 83%, 86%, and 80%, respectively. For employees with over 10 years of experience, the accuracy, precision, and recall rates were 85%, 88%, and 82%, respectively. These data further confirm the applicability and effectiveness of the BP neural network model across different experience groups. Thus, the adoption of machine learning technologies for optimizing marine regulatory resources holds significant practical value, aiding in the enhancement of regulatory capacity and effectiveness within coastal local governments.

A non-invasive preoperative prediction model for predicting axillary lymph node metastasis in breast cancer based on a machine learning approach: combining ultrasonographic parameters and breast gamma specific imaging features.

The most common route of breast cancer metastasis is through the mammary lymphatic network. An accurate assessment of the axillary lymph node (ALN) burden before surgery can avoid unnecessary axillary surgery, consequently preventing surgical complications. In this study, we aimed to develop a non-invasive prediction model incorporating breast specific gamma image (BSGI) features and ultrasonographic parameters to assess axillary lymph node status. Cohorts of breast cancer patients who underwent surgery between 2012 and 2021 were created (The training set included 1104 ultrasound images and 940 BSGI images from 235 patients, the test set included 568 ultrasound images and 296 BSGI images from 99 patients) for the development of the prediction model. six machine learning (ML) methods and recursive feature elimination were trained in the training set to create a strong prediction model. Based on the best-performing model, we created an online calculator that can make a linear predictor in patients easily accessible to clinicians. The receiver operating characteristic (ROC) and calibration curve are used to verify the model performance respectively and evaluate the clinical effectiveness of the model. Six ultrasonographic parameters (transverse diameter of tumour, longitudinal diameter of tumour, lymphatic echogenicity, transverse diameter of lymph nodes, longitudinal diameter of lymph nodes, lymphatic color Doppler flow imaging grade) and one BSGI features (axillary mass status) were selected based on the best-performing model. In the test set, the support vector machines' model showed the best predictive ability (AUC = 0.794, sensitivity = 0.641, specificity = 0.8, PPV = 0.676, NPV = 0.774 and accuracy = 0.737). An online calculator was established for clinicians to predict patients' risk of ALN metastasis ( https://wuqian.shinyapps.io/shinybsgi/ ). The result in ROC showed the model could benefit from incorporating BSGI feature. This study developed a non-invasive prediction model that incorporates variables using ML method and serves to clinically predict ALN metastasis and help in selection of the appropriate treatment option.

Predicting the responses of European grassland communities to climate and land cover change.

European grasslands are among the most species-rich ecosystems on small spatial scales. However, human-induced activities like land use and climate change pose significant threats to this diversity. To explore how climate and land cover change will affect biodiversity and community composition in grassland ecosystems, we conducted joint species distribution models (SDMs) on the extensive vegetation-plot database sPlotOpen to project distributions of 1178 grassland species across Europe under current conditions and three future scenarios. We further compared model accuracy and computational efficiency between joint SDMs (JSDMs) and stacked SDMs, especially for rare species. Our results show that: (i) grassland communities in the mountain ranges are expected to suffer high rates of species loss, while those in western, northern and eastern Europe will experience substantial turnover; (ii) scaling anomalies were observed in the predicted species richness, reflecting regional differences in the dominant drivers of assembly processes; (iii) JSDMs did not outperform stacked SDMs in predictive power but demonstrated superior efficiency in model fitting and predicting; and (iv) incorporating co-occurrence datasets improved the model performance in predicting the distribution of rare species. This article is part of the theme issue 'Ecological novelty and planetary stewardship: biodiversity dynamics in a transforming biosphere'.

Multi-Instance Learning for Vocal Fold Leukoplakia Diagnosis Using White Light and Narrow-Band Imaging: A Multicenter Study.

Vocal fold leukoplakia (VFL) is a precancerous lesion of laryngeal cancer, and its endoscopic diagnosis poses challenges. We aim to develop an artificial intelligence (AI) model using white light imaging (WLI) and narrow-band imaging (NBI) to distinguish benign from malignant VFL. A total of 7057 images from 426 patients were used for model development and internal validation. Additionally, 1617 images from two other hospitals were used for model external validation. Modeling learning based on WLI and NBI modalities was conducted using deep learning combined with a multi-instance learning approach (MIL). Furthermore, 50 prospectively collected videos were used to evaluate real-time model performance. A human-machine comparison involving 100 patients and 12 laryngologists assessed the real-world effectiveness of the model. The model achieved the highest area under the receiver operating characteristic curve (AUC) values of 0.868 and 0.884 in the internal and external validation sets, respectively. AUC in the video validation set was 0.825 (95% CI: 0.704-0.946). In the human-machine comparison, AI significantly improved AUC and accuracy for all laryngologists (p < 0.05). With the assistance of AI, the diagnostic abilities and consistency of all laryngologists improved. Our multicenter study developed an effective AI model using MIL and fusion of WLI and NBI images for VFL diagnosis, particularly aiding junior laryngologists. However, further optimization and validation are necessary to fully assess its potential impact in clinical settings. 3 Laryngoscope, 2024.

Adaptive sampling by citizen scientists improves species distribution model performance: A simulation study

Adaptive sampling by citizen scientists improves species distribution model performance: A simulation study

Endangered Bird Species Classification using Machine Learning

The conservation of endangered bird species is a critical aspect of biodiversity preservation. Traditional methods of identifying and classifying these species are often labor-intensive and time-consuming. In recent years, advancements in machine learning have offered promising alternatives for enhancing the accuracy and efficiency of such tasks. This paper explores the application of various machine learning algorithms to the classification of endangered bird species. By leveraging a dataset comprising images and audio recordings of bird calls, we train and evaluate models such as Convolutional Neural Networks (CNNs) for image recognition and Recurrent Neural Networks (RNNs) for audio classification. Our results demonstrate that these models can achieve high accuracy rates, significantly surpassing traditional methods. Furthermore, we discuss the importance of feature selection, data augmentation, and the integration of multimodal data in improving model performance. The findings underscore the potential of machine learning to revolutionize wildlife conservation efforts, providing a scalable and robust tool for the timely identification and protection of endangered bird species.

Screening and Identification of Neutrophil Extracellular Trap-related Diagnostic Biomarkers for Pediatric Sepsis by Machine Learning.

Neutrophil extracellular trap (NET) is released by neutrophils to trap invading pathogens and can lead to dysregulation of immune responses and disease pathogenesis. However, systematic evaluation of NET-related genes (NETRGs) for the diagnosis of pediatric sepsis is still lacking. Three datasets were taken from the Gene Expression Omnibus (GEO) database: GSE13904, GSE26378, and GSE26440. After NETRGs and differentially expressed genes (DEGs) were identified in the GSE26378 dataset, crucial genes were identified by using LASSO regression analysis and random forest analysis on the genes that overlapped in both DEGs and NETRGs. These crucial genes were then employed to build a diagnostic model. The diagnostic model's effectiveness in identifying pediatric sepsis across the three datasets was confirmed through receiver operating characteristic curve (ROC) analysis. In addition, clinical pediatric sepsis samples were collected to measure the expression levels of important genes and evaluate the diagnostic model's performance using qRT-PCR in identifying pediatric sepsis in actual clinical samples. Next, using the CIBERSORT database, the relationship between invading immune cells and diagnostic markers was investigated in more detail. Lastly, to evaluate NET formation, we measured myeloperoxidase (MPO)-DNA complex levels using ELISA. A group of five important genes (MME, BST1, S100A12, FCAR, and ALPL) were found among the 13 DEGs associated with NET formation and used to create a diagnostic model for pediatric sepsis. Across all three cohorts, the sepsis group had consistently elevated expression levels of these five critical genes as compared to the normal group. Area under the curve (AUC) values of 1, 0.932, and 0.966 indicate that the diagnostic model performed exceptionally well in terms of diagnosis. Notably, when applied to the clinical samples, the diagnostic model also showed good diagnostic capacity with an AUC of 0.898, outperforming the effectiveness of traditional inflammatory markers such as PCT, CRP, WBC, and NEU%. Lastly, we discovered that children with high ratings for sepsis also had higher MPO-DNA complex levels. In conclusion, the creation and verification of a five-NETRGs diagnostic model for pediatric sepsis performs better than established markers of inflammation.

Handwritten Based Alzheimer Disease Prediction from One Dimensional Datasets Using Deep Learning

Because to their high cost, sensitivity, and difficulty in completing surgeries, brain-related disorders are among the most challenging conditions. On the other hand, since the procedure's outcomes could be negative, the operation itself does not have to succeed. Alzheimer's disease, which affects adults and causes varied degrees of memory loss and knowledge forgetfulness, is one of the most prevalent brain diseases. based on the state of each patient. For these reasons, using user-handwritten datasets to categorise memory loss and determine the patient's evaluation of Alzheimer's disease at every given level is crucial. This work offers a novel method for predicting Alzheimer's disease by using advanced deep learning techniques on handwritten data. Alzheimer's is a degenerative. Alzheimer's disease, a neurological condition that progresses and necessitates prompt diagnosis and appropriate treatment. Traditional diagnostic techniques are mostly based on clinical evaluations and imaging, which are frequently inaccessible and expensive. This study investigates the unrealized potential of handwritten data as a special kind of Alzheimer's disease prediction. The dataset provides a broad picture of cognitive impairments by containing handwritten samples collected from individuals in varying states of cognition. Using deep learning architectures, such as the multi-layer perceptron method, the suggested model takes use of the temporal dependencies found in sequential handwritten patterns. These designs show promise for ensuring the temporal characteristics of handwritten data are captured with subtle features. The handwritten input is converted into a format suitable for deep learning using feature extraction techniques, which helps with efficient model training. A thorough assessment of the model's performance is conducted using common metrics including specificity, sensitivity, and accuracy. The goal is to determine whether the model can correctly forecast Alzheimer's disease based just on the unique features present in handwritten samples.

Web Application for Company Stock Price Prediction

This study contributes to predicting stock prices. In this paper, we have provided a novel machine learning and deep learning technique for stock prediction, which capitalizes on the complex emotional patterns of investors that can be extracted from Twitter data. By joining two separate datasets, such as Twitter sentiments and the associated stock prices, we have applied complex algorithms and evaluated this method, which classifies a user’s sentiment into a positive and negative one. It was possible to address the complexity of investor behavior within the highly unstable stock market environment. One of the most important problems that are encountered by developers is gaining the possibility to properly capture stock price fluctuations in the temporal dimension. It implies the necessity to utilize such reliable evaluation metrics as precision-recall factors for classification purpose and root mean squared error for regression purpose. by using these metrics, the performance of the predictive model can be evaluated in detail, which allows to draw meaningful permanent conclusions within the context of how well it reflects the complex relationship between investor sentiment and market dynamics. The results of the evaluation conducted as part of the process of validating the proposed hypothesis show that, indeed, the predictive accuracy improves significantly from the use of machine learning and deep learning algorithms if paired with Twitter users’ emotional sentiment data. This combination enhances the predictive capacity of the model and provides meaningful insights into how the changes in stock prices are impacted by the sentiment of investors. Moreover, in the process of this study, the explanation and detail consideration of the algorithms for feature engineering and data processing were provided, which stands crucial for increasing the predictive model’s accuracy The study has a valuable implication for both the financial analysis and technical spheres due to the resolution of these methods and presents a model for further research activities in stock prediction. To conclude, this interdisciplinary strategy that combines sentiment analysis and stock prediction not only advances our understanding of the investor behavior but also brings the opportunity for novel applications in the emerging field at the juncture of technology and finance. As a result, the current research sheds light on the intricate connection between the dynamics of the stock market and investor sentiment, thus offering new insights that can better inform predictive modeling research in the financial industry.

Testing the Performance of LSTM and ARIMA Models for In-Season Forecasting of Canopy Cover (CC) in Cotton Crops

Testing the Performance of LSTM and ARIMA Models for In-Season Forecasting of Canopy Cover (CC) in Cotton Crops

Construction and validation of nomogram prognostic model for predicting survival in hepatoblastoma patients: a population-based study.

For patients with hepatoblastoma (HB), current staging system is not accurate in predicting survival outcomes. The aim of this study was to develop two accurate survival prediction models to guide clinical decision making. A retrospective analysis of 424 HB patients was performed from 2004 to 2015 using the Surveillance, Epidemiology and End Results (SEER) database. Univariate and multivariate Cox regression analysis was used to screen for variables. The identified variables were used to build survival prediction model. The performance of the nomogram models was assessed based on the concordance index (C-index), calibration plot, and receiver operating characteristic (ROC) curve. The Cox regression analysis identified six variables affecting overall survival (OS) in HB patients, including race, tumor size, lymph node involvement, distant metastases, surgery and chemotherapy. And the Cox regression analysis identified five variables including race, lymph node involvement, distant metastases, surgery, and chemotherapy that affect cancer-specific survival (CCS) in HB patients. In the training cohort, the C-index of the nomogram in predicting the OS was 0.791 [95% confidence intervals (95% CI) 0.717-0.865], CSS was 0.805(95% CI 0.728-0.882). In the validation cohort, the C-index of the nomogram in predicting the OS was 0.712 (95% CI 0.511-0.913), the CSS was 0.751 (95% CI 0.566-0.936). In the training cohort, the area under the receiver operator characteristics curve (AUC) values of the nomogram in prediction of the 1-, 3-, and 5-year OS were 0.842 (95% CI 0.739-0.944), 0.759 (95% CI 0.670-0.849), and 0.770 (95% CI 0.686-0.852), respectively. In the validation cohort, the AUC values for prediction of the 1-, 3-, and 5-year OS were 0.920 (95% CI 0.806-1.034), 0.863 (95% CI 0.750-0.976), and 0.844 (95% CI 0.721-0.967), respectively. Two nomogram models were developed and validated in this study which provided accurate prediction of the OS and CSS in HB patients. The constructed models can be used for predicting survival outcomes and guide treatment for HB patients.

Seismic retrofit of structures using steel frames with viscoelastic hinges

In this paper a new seismic retrofit system composed of a steel frame with viscoelastic hinges is proposed and its applicability and efficiency are evaluated in a theoretical framework. First, the mechanical behavior of viscoelastic hinges and the system are studied and the related formulas are derived. The analytical model of the proposed seismic damper is established first, and the damper is subjected to cyclic loads to compare its hysteretic behavior with the one from the formulation. In order to further evaluate the efficiency of this system, the validated analytical model is used in seismic retrofit of a case study structure to reach the collapse prevention performance under the maximum considered earthquake hazard level. The seismic performance of the analysis model is compared before and after retrofit in terms of maximum interstory drift ratio, residual displacement, and energy dissipation of structural members and dampers. The results show that the proposed damper can be used to reach the stipulated drift limit state. Furthermore, the inelastic energy dissipated by the structural members is reduced drastically and the structure can be effectively protected against irreversible damages due to inelastic deformations.

Group-Based Federated Knowledge Distillation Intrusion Detection

Intrusion detection based on federated learning allows the sharing of more high-quality attack samples to improve the intrusion detection performance of local models while preserving the privacy of local data. Most research on federated learning intrusion detection requires local models to be homogeneous. However, in practical scenarios, local models often include both homogeneous and heterogeneous models due to differences in hardware capabilities and business requirements among nodes. Additionally, there is still room for improvement in the accuracy of recognizing novel attacks in existing researches. To address the challenges mentioned above, we propose a Group-based Federated Knowledge Distillation Intrusion Detection approach. First, through a step-by-step grouping method, we achieve the grouping effect of intra-group homogeneity and inter-group heterogeneity, laying the foundation for reducing the aggregation difficulty in intra-group homogenous aggregation and inter-group heterogeneous aggregation. Second, in intra-group homogenous aggregation, a dual-objective optimization model is employed to quantify the learning quality of local models. Weight coefficients are assigned based on the learning quality to perform weighted aggregation. Lastly, in inter-group heterogeneous aggregation, the group leader model’s learning quality is used to classify and aggregate local soft labels, generating global soft labels. Group leader models utilize global soft labels for knowledge distillation to acquire knowledge from heterogeneous models. Experimental results on NSL-KDD and UNSW-NB datasets demonstrate the superiority of our proposed method over other algorithms.

Individual random effects model for differences in trait distribution among respondents

The homogeneity hypothesis is a common assumption in classic measurement. However, the item response theory model assumes that different respondents with same ability have the same option probabilities, which may not hold. The aim of this study is to propose a new individual random effect model that accounts for the differences in option probabilities among respondents with same latent traits by using within-person variance. The performance of the new model is evaluated through simulation studies and real data using the PRESUPP scale of PISA. The model parameters are estimated by the MCMC method. The results show that the individual random effect model can provide more accurate parameter estimates and obtain a scale parameter to describe the distribution of respondents’ abilities, under different within-person variances. The new model has lower RMSE and better model fit than the classic IRT model.

A deep learning framework for quality control process in the motor oil industry

Given the advancements in the modern world, using Multivariate-Multistage Quality Control (MVMSQC) patterns in continuous production industries is deemed crucial and essential. This study examines the importance and necessity of Multivariate-Multistage Quality Control in manufacturing industries, focusing on motor oil production. Motor oil quality significantly influences engine performance. Thus, the primary objective of this research is to enhance accuracy in detecting faults in quality variables, utilizing deep learning algorithms for visual quality control of data, a facet often overlooked in classical statistical methods and reducing process control time. Combining deep learning algorithms such as Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN) for controlling numerical variables, and Residual Networks (ResNet) and Dense Convolutional Networks (DenseNet) for controlling image variables have been employed. Additionally, the Honey Bee Mating Optimization Algorithm (GBC) has been utilized to tune the parameters of the LSTM-CNN and ResNet-DenseNet deep learning algorithms. Combining heuristic algorithms and deep learning algorithms enhances the performance of the final models in quality control processes. A case study in the motor oil production industry is examined to demonstrate the real-world application of the proposed model. The proposed LSTM-CNN hybrid algorithm outperforms individual CNN and LSTM algorithms in fault detection, improving performance by 15% and 8%, respectively. Furthermore, in visual components, the proposed ResNet-DenseNet hybrid algorithm has shown higher accuracy than ResNet and DenseNet algorithms, improving performance by 10% and 15%, respectively. This research significantly contributes to developing Multivariate-Multistage Quality Control in manufacturing industries by employing deep learning methods. The primary aim is to identify faults in quality components and utilize deep learning algorithms for visual quality control of data, a facet often overlooked in classical statistical methods. The research significantly advances multivariate-multistage quality control in manufacturing industries by leveraging deep learning methods. It addresses detecting complex nonlinear patterns in large-scale data with high flexibility and minimal time.