Validation Dataset Research Articles

ADYOLOv5-Face: An Enhanced YOLO-Based Face Detector for Small Target Faces

Benefiting from advancements in generic object detectors, significant progress has been achieved in the field of face detection. Among these algorithms, the You Only Look Once (YOLO) series plays an important role due to its low training computation cost. However, we have observed that face detectors based on lightweight YOLO models struggle with accurately detecting small faces. This is because they preserve more semantic information for large faces while compromising the detailed information for small faces. To address this issue, this study makes two contributions to enhance detection performance, particularly for small faces: (1) modifying the neck part of the architecture by integrating a Gather-and-Distribute mechanism instead of the traditional Feature Pyramid Network to tackle the information fusion challenges inherent in YOLO-based models; and (2) incorporating an additional detection head specifically designed for detecting small faces. To evaluate the performance of the proposed face detector, we introduce a new dataset named XD-Face for the face detection task. In the experimental section, the proposed model is trained using the Wider Face dataset and evaluated on both Wider Face and XD-face datasets. Experimental results demonstrate that the proposed face detector outperforms other excellent face detectors across all datasets involving small faces and achieved improvements of 1.1%, 1.09%, and 1.35% in the AP50 metric on the WiderFace validation dataset compared to the baseline YOLOv5s-based face detector.

Phishing URL detection with neural networks: an empirical study

Cybercriminals create phishing websites that mimic legitimate websites to get sensitive information from companies, individuals, or governments. Therefore, using state-of-the-art artificial intelligence and machine learning technologies to correctly classify phishing and legitimate URLs is imperative. We report the results of applying deterministic and probabilistic neural network models to URL classification. Key achievements of this work are: (1) The development of a unique approach based on probabilistic neural networks that improves classification accuracy. (2) We show for the first time in URL phishing research that a machine learning model trained on a combination of open source and private datasets is successful in production. The dataset is constructed from open sources like Alexa, PhishTank, or OpenPhish and, most importantly, real-world production data from EasyDMARC. The daily validation of the model using daily reported URL data and corresponding labels, both from open-source platforms and private production, reach on average a 97% accuracy on the validation dataset, labeled by PhishTank, OpenPhish and EasdDMARC where possible mislabeled data can not be excluded and was not possible to check due to large number of URLs. Feature engineering was done without third-party dependencies. Lastly, the evaluation of both deterministic and probabilistic models shows high accuracy on short and long URLs, where short URLs are defined as having less than 50 characters.

Predicting ignitability classification of thermally thick solids using hybrid GA-BPNN and PSO-BPNN algorithms

Two hybrid binary classification models combining the merits of GA (genetic algorithm) and PSO (particle swarm optimization) with BPNN (backpropagation neural network) are proposed to estimate the ignitability of thermally thick solids. A 1D numerical model verified by analytical correlation and experimental measurements is employed to yield training, validation and testing datasets. Surface absorptivity (ε), density (ρ), specific heat (Cp), thermal conductivity (k), critical temperature (Tcri), and incident heat flux (HF) which are highly related to solid ignition are selected as inputs, and ignitability serves as output. A one-hidden-layer BPNN is first constructed, and then GA and PSO are encoded to form GA-BPNN and PSO-BPNN where connection weights and biases are optimized by GA/PSO and the predicted error serves as objective function of GA/PSO. Results show that the optimal neuron number in hidden layer is 34. Compared to BPNN, GA-BPNN and PSO-BPNN feature higher accuracy but more training time. Accuracy and convergence efficiency of PSO-BPNN are higher than GA-BPNN, nevertheless GA-BPNN exhibits better robustness. Predictive accuracies of BPNN, GA-BPNN, and PSO-BPNN on testing dataset are 99.05%, 99.55%, and 99.55%, respectively. ROC (receiver operating characteristic) curves are plotted to rank the performance of the three models, revealing 0.5 is an appropriate threshold to classify ignitability. Feature importance of the models is analyzed using PI (permutation importance) and MIV (mean impact value) methods. It is found HF and Tcri are the most important two inputs followed by ε, while ρ, Cp, and k show least and approximately identical impact. Reliability of the three models is verified by predicting critical heat flux of ignition of three polymers, and good agreement with literature is observed.

Estimation of carbon stocks in the areas of a seasonally dry tropical forest in the Brazilian semi-arid region

In addition to its ecological importance, the Caatinga biome, one of the most extensive seasonally dry tropical forests (SDTF) in the world, has a relevant socioeconomic role as it is used as a primary natural resource by local communities. However, inadequate ecosystem management practices have resulted in gradual loss of natural vegetation in this ecosystem. Carbon stock estimation is a parameter that can contribute as a support tool for managing and maintaining the few remaining natural vegetated areas. In this study, we calibrated and validated the CENTURY model to simulate carbon stocks in areas of the Caatinga, in the state of Pernambuco, and compared the predictive capacity of the CENTURY model with available estimates. In the validation dataset, the average for biomass stocks was 33.1 Mg C ha−1, this value is close to those observed in the literature for the region. The model also performed well when estimating carbon stocks in the soil (r2 = 0.79, p = 0.017). Ecosystem modeling combined with Geographic Information Systems (GIS) is a promising tool for estimating carbon stocks in the Caatinga, where field sampling campaigns are generally expensive and have scarce research funding opportunities. Furthermore, it also allows the evaluation of the effect of environmental changes on C stocks in long-term studies, which is essential for creating and implementing public policies to mitigate and adapt to the impacts of climate change on the ecosystem. However, additional efforts are needed to improve C estimates, especially in areas with a strongly negative water balance.

79 Using artificial intelligence and machine learning to automate injury prevention surveillance data

Abstract Background The Canadian Hospitals Injury Reporting and Prevention Program (CHIRPP) is a pan-Canadian injury and poisoning surveillance system that tracks patients presenting to Canadian Emergency Departments (EDs). Like many surveillance systems, data is input in a manual fashion with humans screening all charts. Natural language processing algorithms using transformer models can analyze text from triage notes of previous CHIRPP-positive patients to train models to predict future CHIRPP patients. We sought to automate these functions using machine learning and artificial intelligence. Objectives The objective of this initiative is to automate detection of injured patients using artificial intelligence. Once charts are identified, a further summarization model will automatically generate a narrative of the events surrounding the injury, as required for the surveillance system. Design/Methods ED notes were obtained from an electronic health record for all patients visiting the ED from October 2018 to December 2021 (n =203626). All charts were manually reviewed to determine if patients met the criteria for CHIRPP. CHIRPP-positive patients made up 24.8% (n = 50499) of the total presentations. Patient charts were randomly split into train (70%) and validation (30%) datasets. Pre-trained distilbert-base-uncased model from Huggingface was utilized. The best model according to AUROC was selected for further development. For note summarization, we fine-tuned a multi-task model (t5-small) from the Huggingface repository. We used human summaries from the same dataset (October 2018-December 2021) as training data (49,078 note, summary pairs). A 70/30 train/validation split was performed. Results Our model captured 90% of all positive cases while reducing the total number of charts for review by 77%, 75% and 63% for train, validation and test sets respectively. See AUC curve (figure 1). Based on 2022 volumes this is a reduction to 24,027 from 64,938 charts. For note summarization, when we compared the cosine similarities of human summaries to actual note text vs t5-small summaries to note text we saw remarkable similarity implying high model performance. Conclusion Our model has shown that high volumes of charts that were previously screened manually can utilize natural language processing to automate the task with minimal sacrifices in terms of data capture and accuracy, dramatically reducing cost while improving speed to reporting. Our summarization model also showed potential for further automating this process. This application to an injury prevention program has wide scale applicability and scalability.

Automated quantification of cerebral microbleeds in susceptibility-weighted MRI: association with vascular risk factors, white matter hyperintensity burden, and cognitive function.

To train and validate a deep learning (DL)-based segmentation model for cerebral microbleeds (CMB) on susceptibility-weighted MRI; and to find associations between CMB, cognitive impairment, and vascular risk factors. Participants in this single-institution retrospective study underwent brain MRI to evaluate cognitive impairment between January-September 2023. For training the DL model, the nnU-Net framework was used without modifications. The DL model's performance was evaluated on independent internal and external validation datasets. Linear regression analysis was used to find associations between log-transformed CMB numbers, cognitive function (mini-mental status examination [MMSE]), white matter hyperintensity (WMH) burden, and clinical vascular risk factors (age, sex, hypertension, diabetes, lipid profiles, and body mass index). Training of the DL model (n = 287) resulted in a robust segmentation performance with an average dice score of 0.73 (95% CI, 0.67-0.79) in an internal validation set, (n = 67) and modest performance in an external validation set (dice score = 0.46, 95% CI, 0.33-0.59, n = 68). In a temporally independent clinical dataset (n = 448), older age, hypertension, and WMH burden were significantly associated with CMB numbers in all distributions (total, lobar, deep, and cerebellar; all P <.01). MMSE was significantly associated with hyperlipidemia (β = 1.88, 95% CI, 0.96-2.81, P <.001), WMH burden (β = -0.17 per 1% WMH burden, 95% CI, -0.27-0.08, P <.001), and total CMB number (β = -0.01 per 1 CMB, 95% CI, -0.02-0.001, P = .04) after adjusting for age and sex. The DL model showed a robust segmentation performance for CMB. In all distributions, CMB had significant positive associations with WMH burden. Increased WMH burden and CMB numbers were associated with decreased cognitive function. CMB = cerebral microbleed; DL = deep learning, DSC = dice similarity coefficient; MMSE = mini-mental status examination; SVD = small vessel disease; SWI = susceptibility-weighted image; WMH = white matter hyperintensity.

Analysis of risk factors for painful diabetic peripheral neuropathy and construction of a prediction model based on Lasso regression

ObjectiveTo evaluate the prevalence and risk factors of painful diabetic peripheral neuropathy (PDPN) in patients with type 2 diabetic peripheral neuropathy (DPN) in Hunan Province, and establish and verify the prediction model.MethodsThis was a retrospective study involving 4908 patients, all patients were randomly divided into the training dataset(3436 cases)and the validation dataset (1472 cases) in a ratio of 7:3. Electroneurogram, clinical signs,and symptoms were used to evaluate neuropathy. Least absolute shrinkage and selection operator (LASSO) regression was used to select the optimal factors, and multifactorial logistic regression analysis was used to build a clinical prediction model. Calibration plots, decision curve analysis (DCA), and subject work characteristic curves (ROC) were used to assess the predictive effects.ResultThe prevalence of PDPN was 33.2%, and the multivariate logistic regression model showed that peripheral artery disease, duration of diabetes, smoking, and HBA1c were independent risk factors for PDPN in patients with type 2 diabetes. ROC analysis results showed that the AUC of the established prediction model was 0.872 in the training dataset and 0.843 in the validation dataset. The calibration curve and decision curve show that the model has good consistency and significant net benefit.Conclusion33.2% of DPN patients had PDPN in Hunan Province, China. Peripheral artery disease, duration of diabetes, smoking, and HBA1c are risk factors for PDPN in patients with type 2 diabetes. The prediction model is based on the above factors, which can well predict the probability of PDPN.

Precision phenotyping from routine laboratory parameters for out of hospital survival prediction in an all comers prospective PCI registry

Out-of-hospital mortality in coronary artery disease (CAD) is particularly high and established adverse event prediction tools are yet to be available. Our study aimed to investigate whether precision phenotyping can be performed using routine laboratory parameters for the prediction of out-of-hospital survival in a CAD population treated by percutaneous coronary intervention (PCI). All patients treated by PCI and discharged alive in a tertiary center between January 2016 – December 2022 that have been included prospectively in the local registry were analyzed. 115 parameters from the PCI registry and 266 parameters derived from routine laboratory testing were used. An extreme gradient-boosted decision tree machine learning (ML) algorithm was trained and used to predict all-cause and cardiovascular-cause survival. A total of 4027 patients with 4981 PCI hospitalizations were randomly included in the 70% training dataset and 1729 patients with 2160 PCI hospitalizations were randomly included in the 30% validation dataset. All-cause and cardiovascular cause mortality was 17.5% and 12.2%. The integrated area under the receiver operator characteristic curve for prediction of all-cause and cardiovascular cause mortality by the ML on the validation dataset was 0.844 and 0.837, respectively (all p < 0.001). Parameters reflecting renal function (first and maximum serum creatinine), hematologic function (mean corpuscular hemoglobin concentration, platelet distribution width), and inflammatory status (lymphocyte per monocyte ratio) were among the most important predictors. Accurate out-of-hospital survival prediction in CAD can be achieved using routine laboratory parameters. ML outperformed clinical risk scores in predicting out-of-hospital mortality in a prospective all-comers PCI population and has the potential to precisely phenotype patients.

Assessment of the relationship between central venous pressure waveform and the severity of tricuspid valve regurgitation using data science

Purpose The relationship between the height of the V wave in the central venous pressure (CVP) waveform and the severity of tricuspid regurgitation (TR) is well known. Their diagnostic ability is unconfirmed. This study explored CVP waveform variations with TR. Methods All patients who underwent preoperative echocardiography and CVP waveform measurements before surgery at our institution were included. Indices were created to capture each feature of the CVP waveform. The median value for each case was obtained and statistically analyzed according to the severity of TR. A deep learning technique, Transformer, was used to handle the complex features of CVP waveforms. Results This study included 436 cases. The values for C wave – Y descent, X descent – Y descent, and V wave – Y descent differed significantly in the Jonckheere–Terpstra test (p = 0.0018, 0.027, and 0.077, respectively). The area under the receiver operating characteristic (ROC) curve (AUC) for X descent – Y descent in two groups, none to moderate TR and severe TR, was 0.83 (95% confidence interval (CI) [0.68, 0.98]). For Transformer, the accuracy of the validation dataset was 0.97. Conclusions The shape of the CVP waveform varied with the severity of TR in a large dataset.

Machine Learning-Based In Silico Prediction of the Inhibitory Activity of Chemical Substances Against Rat and Human Cytochrome P450s.

The prediction of cytochrome P450 inhibition by a computational (quantitative) structure-activity relationship approach using chemical structure information and machine learning would be useful for toxicity research as a simple and rapid in silico tool. However, there are few in silico models focusing on the species differences between rat and human in the P450s inhibition. This study aimed to establish in silico models to classify chemical substances as inhibitors or non-inhibitors of various rat and human P450s, using only molecular descriptors. Using the in-house test results from our in vitro experiments, we used 326 substances for model construction and internal validation data. Apart from the 326 substances, 60 substances were used as external validation data set. We focused on seven rat P450s (CYP1A1, CYP1A2, CYP2B1, CYP2C6, CYP2D1, CYP2E1, and CYP3A2) and 11 human P450s (CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4). Most of the models established using XGBoost showed an area under the receiver operating characteristic curve (ROC-AUC) of 0.8 or more in the internal validation. When we set an applicability domain for the models and confirmed their generalization performance through external validation, most of the models showed an ROC-AUC of 0.7 or more. Interestingly, for CYP1A1 and CYP1A2, we discovered that a human P450 inhibitory activity model can predict rat P450 inhibitory activity and vice versa. These models are the first attempts to predict inhibitory activity against a wide variety of P450s in both rats and humans using chemical structure information. Our experimental results and in silico models would be helpful to support information for species similarities and differences in chemical-induced toxicity.

Mathematical modeling for the efficiency function of the Retiro small hydroelectric power plant turbine-generator set

ABSTRACT Many authors use an efficiency constant for the turbine-generator set to calculate energy generation in hydroelectric projects. This often does not reflect reality because the efficiency of this system is relative and is related to the potential of the net head height and the flow of turbine discharge. The present work aims to perform a mathematical modeling for an analysis of the efficiency of the Turbine-Generator Small Hydroelectric Power Plant (SHPP) Retiro. To perform this analysis, a model based on a second-degree polynomial function was employed. A multivariate regression technique was applied by the Least Square Method (LSM) to obtain the coefficients of the function. To verify the behavior of the measured and adjusted data, the Determination Coefficient (R2) of Root Mean Square Error (RMSE) algorithm was applied. The RMSE Mean of 0.022 and the RMSE Std of 0.029 suggest that the second-degree polynomial function is more consistent and stable across different training and validation datasets. The R2 value of 0.993 indicates that 99.3% of the variability in the data is explained by the model. The Omnibus, Durbin-Watson, and Jarque-Bera tests were applied to diagnose potential issues related to normality, autocorrelation, and adequacy, thereby validating the accuracy and significance of the model. The Hessian matrix technique was also used to verify the critical points of the function. The critical point corresponding to a water head of 11.47 meters and a turbine flow of 145.1 m3/s presented the highest operational efficiency. This study allows presenting a turbine efficiency function, its Hill Curve and a generator efficiency function with its corresponding Efficiency Curve. For efficiency analysis in several flow, height and power settings. The technique has proved efficient and can be an applicable tool in energy parameter studies for other hydroelectric projects with the same characteristics as SHPP Retiro.

P10.09.A BLOOD-BASED CLASSIFICATION OF IDH-STATUS OF NON-ENHANCING GLIOMAS: A MACHINE LEARNING APPROACH

Abstract BACKGROUND Non-invasive determination of IDH mutational status in patients with glioma could offer significant therapeutic opportunities. While IDH wildtype tumors (e.g., GBM) typically show enhancement on MRI, IDH mutant tumors often lack this enhancement. However, relying solely on anatomic radiology may lead to misclassification, and currently, tissue acquisition is the primary method for assessing IDH status in gliomas. This limitation hampers the development of neoadjuvant or intraoperative therapeutic strategies based on IDH status; thus, more minimally invasive methods to determine IDH status are needed. In this study, we assessed immune cell proportions, which vary by glioma subtype, from pre-surgery peripheral blood samples as an alternative method of classifying the IDH status in gliomas without enhancement. MATERIAL AND METHODS First, we employed a highly accurate large language model (GPT-4-Turbo-128k) with over 99% accuracy to read radiology notes and exclude patients with enhancing gliomas. Then, we identified twelve immune cell subtypes from whole blood using deconvolution algorithms based on DNA methylation data. These immune cell subtypes, along with patient age, were integrated into a machine learning model (random forest) to predict IDH status (wildtype vs. mutant), leveraging conditional Generative Adversarial Networks to generate synthetic data and mitigate bias in the datasets. Two independent datasets were included for training and validating the model. RESULTS The random forest model had an area under the ROC curve (AUC) of 0.90 and accurately identified IDH status in 81% of a training set of 287 gliomas (65 wildtype and 222 mutant tumors at varied time points after diagnosis). In an independent validation data set of 99 gliomas (6 wildtype and 93 mutant) from pre-surgery blood samples, the AUC was 0.93, and accuracy predicted IDH status in 92% of cases with a sensitivity of 93% and specificity of 83%. CONCLUSION We could accurately predict IDH status in pre-surgical peripheral blood samples from patients with non-enhancing gliomas. This minimally invasive approach is a promising step toward reducing risk and exploring opportunities for earlier therapies based on IDH status.

Artificial Neural Network Modeling of the Removal of Methylene Blue Dye Using Magnetic Clays: An Environmentally Friendly Approach

In this study, the effectiveness of Fe3O4-based clay as a cost-effective material for removing methylene blue (MB) dye from aqueous solutions was evaluated. The structural properties of the clay and Fe3O4-based clay were analyzed using SEM, XRF, BET, XRD, FTIR, and TGA techniques. In this research, the effects of various aspects, such as adsorbent amount, contact time, solution pH, adsorption temperature, and initial dye concentration, on the adsorption of Fe3O4-based clay are investigated. The experiments aimed at understanding the adsorption mechanism of Fe3O4-based clay have shown that the adsorption kinetics are accurately described by the pseudo-second order kinetic model, while the equilibrium data are well represented by the Langmuir isotherm model. The maximum adsorption capacity (qm) was calculated as 52.63 mg/g at 25 °C, 53.48 mg/g at 30 °C, and 54.64 mg/g at 35 °C. All variables affecting the MB adsorption process were systematically optimized in a controlled experimental framework. The effectiveness of the artificial neural network (ANN) model was refined by modifying variables such as the quantity of neurons in the latent layer, the number of inputs, and the learning rate. The model’s accuracy was assessed using the mean absolute percentage error (MAPE) for the removal and adsorption percentage output parameters. The coefficient of determination (R2) values for the dyestuff training, validation, and test sets were found to be 99.40%, 92.25%, and 96.30%, respectively. The ANN model demonstrated a mean squared error (MSE) of 0.614565 for the training data. For the validation dataset, the model recorded MSE values of 0.99406 for the training data, 0.92255 for the validation set, and 0.96302 for the test data. In conclusion, the examined Fe3O4-based clays offer potential as effective and cost-efficient adsorbents for purifying water containing MB dye in various industrial settings.

Enhanced Drag Force Estimation in Automotive Design: A Surrogate Model Leveraging Limited Full-Order Model Drag Data and Comprehensive Physical Field Integration

In this paper, a novel surrogate model for shape-parametrized vehicle drag force prediction is proposed. It is assumed that only a limited dataset of high-fidelity CFD results is available, typically less than ten high-fidelity CFD solutions for different shape samples. The idea is to take advantage not only of the drag coefficients but also physical fields such as velocity, pressure, and kinetic energy evaluated on a cutting plane in the wake of the vehicle and perpendicular to the road. This additional “augmented” information provides a more accurate and robust prediction of the drag force compared to a standard surface response methodology. As a first step, an original reparametrization of the shape based on combination coefficients of shape principal components is proposed, leading to a low-dimensional representation of the shape space. The second step consists in determining principal components of the x-direction momentum flux through a cutting plane behind the car. The final step is to find the mapping between the reduced shape description and the momentum flux formula to achieve an accurate drag estimation. The resulting surrogate model is a space-parameter separated representation with shape principal component coefficients and spatial modes dedicated to drag-force evaluation. The algorithm can deal with shapes of variable mesh by using an optimal transport procedure that interpolates the fields on a shared reference mesh. The Machine Learning algorithm is challenged on a car concept with a three-dimensional shape design space. With only two well-chosen samples, the numerical algorithm is able to return a drag surrogate model with reasonable uniform error over the validation dataset. An incremental learning approach involving additional high-fidelity computations is also proposed. The leading algorithm is shown to improve the model accuracy. The study also shows the sensitivity of the results with respect to the initial experimental design. As feedback, we discuss and suggest what appear to be the correct choices of experimental designs for the best results.

Geoinference of author affiliations using NLP-based text classification

Author affiliations are essential in bibliometric studies, requiring relevant information extraction from free-text affiliations. Precisely determining an author’s location from their affiliation is crucial for understanding research networks, collaborations, and geographic distribution. Existing geoparsing tools using regular expressions have limitations due to unstructured and ambiguous affiliations, resulting in erroneous location identification, especially for unconventional variations or misspellings. Moreover, their inefficient handling of big datasets hampers large-scale bibliometric studies. Though machine learning-based geoparsers exist, they depend on explicit location information, creating challenges when detailed geographic data is absent. To address these issues, we developed and evaluated a natural language processing model to predict the city, state, and country from an author’s free-text affiliation. Our model automates location inference, overcoming drawbacks of existing methods. Trained and tested with MapAffil, a publicly available geoparsed dataset of PubMed affiliations up to 2018, our model accurately retrieves high-resolution locations, even without explicit mentions of a city, state, or even country. Leveraging NLP techniques and the LinearSVC algorithm, our machine learning model achieves superior accuracy based on several validation datasets. This research demonstrates a practical application of text classification for inferring specific geographical locations from free-text affiliations, benefiting researchers and institutions in analyzing research output distribution.

Wave basin investigation of floating wind turbine model using underwater particle image velocimetry: investigating hydrodynamic wake structures

This paper conducts a fundamental study on the hydrodynamic performance of a floating offshore wind turbine (FOWT). This includes investigating the impacts of platform motion, turbine operation, and environmental conditions on hydrodynamic wake development behind the FOWT. Using underwater particle image velocimetry in a wave basin, the fluid flow behaviour behind the floating wind substructure model is compared for different test configurations. The FOWT is examined in three configurations: I. Zero degrees of freedom (fixed), II. Six degrees of freedom (floating), and III. Floating with turbine in operation (full system). The results reveal significant hydrodynamic wake turbulence differences between fixed and floating systems, with 95% and 89% increases in turbulent kinetic energy (TKE) and turbulence intensity (TI), respectively. Turbine loading in configuration III results in additional damping and reduces eddy shedding, resulting in a 26% and 31% decrease in TKE and TI, compared to the floating configuration II. It is found that neglecting turbine operation overestimates substructure drag terms by at least 30%, while fixing the model underestimates drag terms by 54%. This study highlights a trade-off between model accuracy and dynamic complexity when estimating viscous effects on FOWT. It also provides a valuable set of high-fidelity validation data for the development of numerical tools for FOWT analysis.

Resource-optimized cnns for real-time rice disease detection with ARM cortex-M microprocessors

This study explores the application of Artificial Intelligence (AI), specifically Convolutional Neural Networks (CNNs), for detecting rice plant diseases using ARM Cortex-M microprocessors. Given the significant role of rice as a staple food, particularly in Malaysia where the rice self-sufficiency ratio dropped from 65.2% in 2021 to 62.6% in 2022, there is a pressing need for advanced disease detection methods to enhance agricultural productivity and sustainability. The research utilizes two extensive datasets for model training and validation: the first dataset includes 5932 images across four rice disease classes, and the second comprises 10,407 images across ten classes. These datasets facilitate comprehensive disease detection analysis, leveraging MobileNetV2 and FD-MobileNet models optimized for the ARM Cortex-M4 microprocessor. The performance of these models is rigorously evaluated in terms of accuracy and computational efficiency. MobileNetV2, for instance, demonstrates a high accuracy rate of 97.5%, significantly outperforming FD-MobileNet, especially in detecting complex disease patterns such as tungro with a 93% accuracy rate. Despite FD-MobileNet’s lower resource consumption, its accuracy is limited to 90% across varied testing conditions. Resource optimization strategies highlight that even slight adjustments, such as a 0.5% reduction in RAM usage and a 1.14% decrease in flash memory, can result in a notable 9% increase in validation accuracy. This underscores the critical balance between computational resource management and model performance, particularly in resource-constrained settings like those provided by microcontrollers. In summary, the deployment of CNNs on microcontrollers presents a viable solution for real-time, on-site plant disease detection, demonstrating potential improvements in detection accuracy and operational efficiency. This study advances the field of smart agriculture by integrating cutting-edge AI with practical agricultural needs, aiming to address the challenges of food security in vulnerable regions.

Flood risk assessment using modern models integrated with GIS for susceptibility mapping of the M'zab Valley watershed (Algeria)

ABSTRACT Comprehensive flood control is needed because flooding can be catastrophic for people's lives and ways of living. This management must provide information on the hydrologic, geotechnical, environmental, social, and economic elements of floods. This work created maps for the flood susceptibility of the M'zab Valley using logistic regression (LR) and frequency ratio (FR). The main objective of the research was to strengthen the bivariate probability capabilities of FR and LR. A flood inventory map was created by extracting flood sites from multiple sources. The flood inventory was divided randomly into two parts: 30% was used for validation and the remaining 70% was used to train the models. Independent variable datasets included the distance from the river, rainfall, soil type, land use/land cover (LULC), elevation, drainage density, flow accumulation, and slope. The effect of each variable on flooding was assessed by comparing each independent variable with the dependent flood layer. The flood susceptibility map was assessed using the prediction rate approach on the validation dataset, which was not used for model construction. The accuracy assessment's findings revealed a prediction rate of FR 0.86 and LR 0.881, as well as a success rate of FR 0.924 and LR 0.924.

Calculation of virtual 3D subtraction angiographies using conditional generative adversarial networks (cGANs)

ObjectiveSubtraction angiographies are calculated using a native and a contrast-enhanced 3D angiography images. This minimizes both bone and metal artifacts and results in a pure image of the vessels. However, carrying out the examination twice means double the radiation dose for the patient. With the help of generative AI, it could be possible to simulate subtraction angiographies from contrast-enhanced 3D angiographies and thus reduce the need for another dose of radiation without a cutback in quality. We implemented this concept by using conditional generative adversarial networks.MethodsWe selected all 3D subtraction angiographies from our PACS system, which had performed between 01/01/2018 and 12/31/2022 and randomly divided them into training, validation, and test sets (66%:17%:17%). We adapted the pix2pix framework to work on 3D data and trained a conditional generative adversarial network with 621 data sets. Additionally, we used 158 data sets for validation and 164 for testing. We evaluated two test sets with (n = 72) and without artifacts (n = 92). Five (blinded) neuroradiologists compared these datasets with the original subtraction dataset. They assessed similarity, subjective image quality, and severity of artifacts.ResultsImage quality and subjective diagnostic accuracy of the virtual subtraction angiographies revealed no significant differences compared to the original 3D angiographies. While bone and movement artifact level were reduced, artifact level caused by metal implants differed from case to case between both angiographies without one group being significant superior to the other.ConclusionConditional generative adversarial networks can be used to simulate subtraction angiographies in clinical practice, however, new artifacts can also appear as a result of this technology.

Machine learning and explainable artificial intelligence to predict pathologic stage in men with localized prostate cancer.

Though several nomograms exist, machine learning (ML) approaches might improve prediction of pathologic stage in patients with prostate cancer. To develop ML models to predict pathologic stage that outperform existing nomograms that use readily available clinicopathologic variables. Patients with prostate adenocarcinoma who underwent surgery were identified in the National Cancer Database. Seven ML models were trained to predict organ-confined (OC) disease, extracapsular extension, seminal vesicle invasion (SVI), and lymph node involvement (LNI). Model performance was measured using area under the curve (AUC) on a holdout testing data set. Clinical utility was evaluated using decision curve analysis (DCA). Performance metrics were confirmed on an external validation data set. The ML-based extreme gradient boosted trees model achieved the best performance with an AUC of 0.744, 0.749, 0.816, 0.811 for the OC, ECE, SVI, and LNI models, respectively. The MSK nomograms achieved an AUC of 0.708, 0.742, 0.806, 0.802 for the OC, ECE, SVI, and LNI models, respectively. These models also performed the best on DCA. Findings were consistent on both a holdout internal validation data set as well as an external validation data set. Our ML models better predicted pathologic stage relative to existing nomograms at predicting pathologic stage. Accurate prediction of pathologic stage can help oncologists and patients determine optimal definitive treatment options for patients with prostate cancer.