Performance of nano-modified soil–brick–cement: a machine learning and experimental study

California Bearing Ratio (CBR) is an essential parameter utilized to evaluate the strength of the soil subgrades and base course materials of different types of pavements. In this study, the Machine Learning (ML) approach has been adopted using Random Forest (RF) model to estimate the CBR of the soil based on 10 input parameters such as Plasticity Index (PI), Liquid Limit (LL), Silt Clay content (SC), Fine Sand content (FS), Coarse sand content (CS), Optimum Water Content (OWC), Organic content (O), Plastic Limit (PL), Gravel content (G), and Maximum Dry Density (MDD), which can be easily determined in the laboratory. An experimental database was collected from 214 soil samples, which were classified according to AASHTO M 145(clayey, gravel, sand, silty and clayey soils). The data was divided into 70% training and 30% test data in the model study. Model performance was evaluated using standard statistical measures such as coefficient of determination, correlations, and errors (relative error, MAE, and RMSE). Based on the analysis results shows the RF model is capable of correct prediction of the CBR of the Soil.

California Bearing Ratio (CBR) is an essential parameter utilized to evaluate the strength of the soil subgrades and base course materials of different types of pavements. In this study, the Machine Learning (ML) approach has been adopted using Random Forest (RF) model to estimate the CBR of the soil based on 10 input parameters such as Plasticity Index (PI), Liquid Limit (LL), Silt Clay content (SC), Fine Sand content (FS), Coarse sand content (CS), Optimum Water Content (OWC), Organic content (O), Plastic Limit (PL), Gravel content (G), and Maximum Dry Density (MDD), which can be easily determined in the laboratory. An experimental database was collected from 214 soil samples, which were classified according to AASHTO M 145(clayey, gravel, sand, silty and clayey soils). The data was divided into 70% training and 30% test data in the model study. Model performance was evaluated using standard statistical measures such as coefficient of determination, correlations, and errors (relative error, MAE, and RMSE). Based on the analysis results shows the RF model is capable of correct prediction of the CBR of the Soil.

California Bearing Ratio (CBR) is an essential parameter utilized to evaluate the strength of the soil subgrades and base course materials of different types of pavements. In this study, the Machine Learning (ML) approach has been adopted using Random Forest (RF) model to estimate the CBR of the soil based on 10 input parameters such as Plasticity Index (PI), Liquid Limit (LL), Silt Clay content (SC), Fine Sand content (FS), Coarse sand content (CS), Optimum Water Content (OWC), Organic content (O), Plastic Limit (PL), Gravel content (G), and Maximum Dry Density (MDD), which can be easily determined in the laboratory. An experimental database was collected from 214 soil samples, which were classified according to AASHTO M 145(clayey, gravel, sand, silty and clayey soils). The data was divided into 70% training and 30% test data in the model study. Model performance was evaluated using standard statistical measures such as coefficient of determination, correlations, and errors (relative error, MAE, and RMSE). Based on the analysis results shows the RF model is capable of correct prediction of the CBR of the Soil.

This study aims to use two types of brick aggregates as partial replacements for natural stone and checks their feasibility as road aggregates in dry lean concrete (DLC) mixes. Laboratory studies were performed by partially replacing natural stones with both types of brick aggregates, and the optimum contents were evaluated. The paper describes a detailed analysis of DLC mixes with several percentages of bricks including a number of laboratory tests such as the moisture density, strength parameters, ultrasonic pulse velocity, and resistance to acid attack. For the replacement of stone with brick aggregate, the prime focus was paid to the usage of cement content and the aggregate to cement (A/C) ratio. An A/C ratio of 10:1 is found to be suitable for 100% replacement of natural stone aggregate (NSA) with first-class brick aggregate (FCBA), whereas an A/C ratio of 11:1 may be suitable for the use of 100% overburnt brick aggregate (OBBA). However, high contents of the A/C ratio can also be used if NSA is partially replaced by OBBA or FCBA. The strength parameters of DLC mixes such as flexural strength or split tensile strength showed a reduction of about 20%–22% when replaced by 50% OBBA and FCBA, while the compressive strength was within the limit (i.e., 7 MPa after 7 days of curing). Finally, the cost analysis showed that the production cost of DLC mixes can be reduced by about 15.6%–16.2% and 29%–32%, if the stone aggregates are partially (50% by volume) or fully replaced by FCBA and OBBA, respectively. Thus, the paper indicates that not only the scarcity of stone aggregates can be minimized but the cost can also be reduced by partially or fully replacing NSA with FCBA and OBBA.

Traditional models for predicting in-hospital mortality for patients with heart failure (HF) have used logistic regression and do not account for social determinants of health (SDOH). To develop and validate novel machine learning (ML) models for HF mortality that incorporate SDOH. This retrospective study used the data from the Get With The Guidelines-Heart Failure (GWTG-HF) registry to identify HF hospitalizations between January 1, 2010, and December 31, 2020. The study included patients with acute decompensated HF who were hospitalized at the GWTG-HF participating centers during the study period. Data analysis was performed January 6, 2021, to April 26, 2022. External validation was performed in the hospitalization cohort from the Atherosclerosis Risk in Communities (ARIC) study between 2005 and 2014. Random forest-based ML approaches were used to develop race-specific and race-agnostic models for predicting in-hospital mortality. Performance was assessed using C index (discrimination), regression slopes for observed vs predicted mortality rates (calibration), and decision curves for prognostic utility. The training data set included 123 634 hospitalized patients with HF who were enrolled in the GWTG-HF registry (mean [SD] age, 71 [13] years; 58 356 [47.2%] female individuals; 65 278 [52.8%] male individuals. Patients were analyzed in 2 categories: Black (23 453 [19.0%]) and non-Black (2121 [2.1%] Asian; 91 154 [91.0%] White, and 6906 [6.9%] other race and ethnicity). The ML models demonstrated excellent performance in the internal testing subset (n = 82 420) (C statistic, 0.81 for Black patients and 0.82 for non-Black patients) and in the real-world-like cohort with less than 50% missingness on covariates (n = 553 506; C statistic, 0.74 for Black patients and 0.75 for non-Black patients). In the external validation cohort (ARIC registry; n = 1205 Black patients and 2264 non-Black patients), ML models demonstrated high discrimination and adequate calibration (C statistic, 0.79 and 0.80, respectively). Furthermore, the performance of the ML models was superior to the traditional GWTG-HF risk score model (C index, 0.69 for both race groups) and other rederived logistic regression models using race as a covariate. The performance of the ML models was identical using the race-specific and race-agnostic approaches in the GWTG-HF and external validation cohorts. In the GWTG-HF cohort, the addition of zip code-level SDOH parameters to the ML model with clinical covariates only was associated with better discrimination, prognostic utility (assessed using decision curves), and model reclassification metrics in Black patients (net reclassification improvement, 0.22 [95% CI, 0.14-0.30]; P < .001) but not in non-Black patients. ML models for HF mortality demonstrated superior performance to the traditional and rederived logistic regressions models using race as a covariate. The addition of SDOH parameters improved the prognostic utility of prediction models in Black patients but not non-Black patients in the GWTG-HF registry.

Convective storms and the associated heavy rainfall, flooding, hail, wind gusts and lightning can result in significant damage to property and loss of lives. Thus, there is a need for accurate prediction of the future location and severity of such storms (i.e. in the sub-kilometer resolution for the next hour) to assist the decision making of civil protection authorities. The typical approach to produce such nowcasts is to identify storm cells as separate entities from radar images, which provides a natural way for associating the storm attributes with its severity. Following this approach, we have implemented a set of pan-European nowcast products in the TAMIR and EDERA projects funded by the EU Civil Protection Mechanism. This has been done by combining a cell tracking method with a random forest-based machine learning (ML) model and a Kalman filter model. In the nowcast products, storm cells are identified from the OPERA radar composites. The ML model is used for predicting the storm severity level. It is trained using a large database of meteorological features and weather hazard reports during summer months (May-September) between 2018-2022. The storm features include basic cell and track properties (i.e. area and age), lightning flash and wind observations, and also indicators of convective potential from ERA5 reanalyses. The target variable for training the ML model is the storm hazard level. The hazard level estimation is done based on the distances and time delays between the storms and the associated weather hazard reports obtained from the European Severe Weather Database (ESWD). We have trained different ML models against the hazard levels estimated for each event type (e.g. heavy rain, lightning and severe wind gusts). The above models are combined with a Kalman filter-based methodology to produce probabilistic nowcasts of future storm locations together with their severity level. The added value of such nowcasts for decision making is demonstrated with case studies and relevant verification metrics. Finally, we demonstrate how the nowcasts can be combined with different exposure layers to translate them into predictions of actual storm impacts.

The software has turn into an imperious part of human’s life. In the recent computing era, many large-scale complex network systems and millions of modern technological devices produce a huge amount of data every second. Among these data, the amount of imbalanced data is relatively excessive. The machine learning model is miss leaded by these imbalanced data. Software Defect Prediction (SDP) is a standout amongst the most helping exercises during the testing phase. The estimated cost of finding and fixing defects is approximately billions of pounds per year. To reduce this problem, software defect prediction has come forth but need fine tuning to have expected efficiency. In this chapter, we have proposed a new model based on machine learning approach to predict software defect and identify the key factors that may help the software engineer to identify the most defect-prone part of the system. The proposed model works as follows. First, need to remove highly correlated features and turn all the feature in the same scale using the scaling feature approach. Second, we have used Synthetic Minority Over-Sampling Technique (SMOTE), Adaptive Synthetic (ADASYN) and Hybrid sampling method to balance highly imbalanced datasets. Third, Random Forest Importance and Chi-square algorithms are chosen to find out the factors which have high effect on software defect. Cross validation is used to remove overriding problem. Scikit-learn library is used for machine learning algorithms. Pandas library is used for data processing. Matplotlib, and PyPlot are used for graph and data visualization respectively. The hybrid sampling method and Random Forest (RF) algorithms achieved the highest prediction accuracy about 93.26% by showing its superiority.

Applied machine learning to predict stress hotspots II: Hexagonal close packed materials

Cardiovascular diseases are among the foremost common serious diseases’ poignant human health. Disorder is also prevented or relieved by early diagnosis, and this might scale back mortality rates. Distinctive risk factors mistreatment machine learning models may be a promising approach. The model that comes with totally different strategies to realize effective prediction of heart disease. For this planned model to be successful, economical information collection, information Pre-processing and information Transformation methods to form correct info for the coaching model. The model has a combined dataset. Appropriate options are hand-picked by using the Relief and LASSO techniques. New hybrid classifiers like Random Forest based Machine Learning are developed by group action the normal classifiers with fabric and boosting methods, that are employed in the coaching process. Some machine learning algorithms to calculate the accuracy, sensitivity, error rate, precision. The results are shown singly to supply comparisons. Supported the result analysis will conclude that our planned model created the very best accuracy whereas mistreatment RFBM and Relief feature choice method.

The modulation of protein-protein interactions (PPIs) by small chemical compounds is challenging. PPIs play a critical role in most cellular processes and are involved in numerous disease pathways. As such, novel strategies that assist the design of PPI inhibitors are of major importance. We previously reported that the knowledge-based DLIGAND2 scoring tool was the best-rescoring function for improving receptor-based virtual screening (VS) performed with the Surflex docking engine applied to several PPI targets with experimentally known active and inactive compounds. Here, we extend our investigation by assessing the vs. potential of other types of scoring functions with an emphasis on docking-pose derived solvent accessible surface area (SASA) descriptors, with or without the use of machine learning (ML) classifiers. First, we explored rescoring strategies of Surflex-generated docking poses with five GOLD scoring functions (GoldScore, ChemScore, ASP, ChemPLP, ChemScore with Receptor Depth Scaling) and with consensus scoring. The top-ranked poses were post-processed to derive a set of protein and ligand SASA descriptors in the bound and unbound states, which were combined to derive descriptors of the docked protein-ligand complexes. Further, eight ML models (tree, bagged forest, random forest, Bayesian, support vector machine, logistic regression, neural network, and neural network with bagging) were trained using the derivatized SASA descriptors and validated on test sets. The results show that many SASA descriptors are better than Surflex and GOLD scoring functions in terms of overall performance and early recovery success on the used dataset. The ML models were superior to all scoring functions and rescoring approaches for most targets yielding up to a seven-fold increase in enrichment factors at 1% of the screened collections. In particular, the neural networks and random forest-based ML emerged as the best techniques for this PPI dataset, making them robust and attractive vs. tools for hit-finding efforts. The presented results suggest that exploring further docking-pose derived SASA descriptors could be valuable for structure-based virtual screening projects, and in the present case, to assist the rational design of small-molecule PPI inhibitors.

Background Crohn’s disease (CD) and intestinal tuberculosis (ITB) are chronic granulomatous inflammatory disorders characterized by a compromised mucosal immunity. Even with diverging etiologies, CD and ITB presents an uncanny resemblance in clinical manifestation resulting in diagnostic dilemma. The gut microbiota regulates myriad of gut mucosal immunological processes. Present study aims to decipher gut microbial dysbiosis in the two disorders and utilize the CD and ITB-specific gut dysbiosis to construct a machine learning (ML)-based predictive model, which can aid in their differential diagnosis. Methods Fecal samples from healthy controls (n=12) and from patients with CD (n=23) and ITB (n=25) were subjected to 16S (V3-V4) amplicon sequencing. Processing of raw reads, construction of ASV feature tables, diversity, core microbiome analysis and ML classifier construction was done using QIIME2-2021.4. Differential abundance analysis (DAA) between the groups was carried out using Deseq2, after adjusting for the subject-specific confounders. Results The α and β diversity indices in CD and ITB groups were significantly reduced than HC group (p = 0.011 and 0.012 resp.), with no significant differences between the two diseases (Fig.1A, 1B). On comparison with HC, CD and ITB groups showed reduction in members of Firmicutes and Bacteroidetes, with enhancement of Actinobacteria and Proteobacteria (Fig.1C and 1D). DAA (FDR q &amp;lt;0.1, FC &amp;gt;2.5) between CD and ITB groups revealed expansion of Succinivibrio dextrinisolvens, Odoribacter splanchnicus, Megasphaera massiliensis, Bacteroides uniformis and B.xylanisolvens in CD group, while Clostridium sp., Haemophilus parainfluenzae and Bifidobacterium sp. were elevated in ITB (Fig.2A). Random Forest-based ML model constructed on the basis of raw microbiome reads and using 80% of the samples to train the model, showed predictive accuracy of 0.78 (AUC=93%). (Fig.2B) Conclusion Our study shows that CD and ITB witnesses significant changes in gut microbial structure. With no significant differences in microbial diversity between two diseases, the signature of gut dysbiosis is distinct between CD and ITB. Exploitation of these differences to construct ML models can potentiate differential diagnosis of CD and ITB.

Hypertension and depression have a well-recognized bidirectional association, with comorbidity reducing antihypertensive medication adherence and elevating adverse health outcomes; depression affects 30% to 40% of hypertensive patients, far higher than the general population. However, validated, hypertension-specific depression screening tools based on machine learning (ML) are currently lacking. This study aimed to develop, validate, and visualize an ML-based depression screening tool for adult hypertensive patients. A cross-sectional study was conducted using 2005–2020 National Health and Nutrition Examination Survey data of 6687 hypertensive patients aged ≥20 years (depression defined by Patient Health Questionnaire-9 score ≥ 10). Least absolute shrinkage and selection operator regression was used for feature selection, data were split into a 7:3 training/test set, and 7 ML models were constructed. Model performance was assessed by area under the receiver operating characteristic curve, calibration, and clinical utility (decision curve analysis); SHapley Additive exPlanations values ensured interpretability, and a web-based tool was built from the optimal model. The random forest model was identified as the optimal screening tool (area under the receiver operating characteristic curve = 0.761), with good calibration and favorable net clinical benefit (threshold probability: 0.07–0.39). Twelve key predictors were established, with sleep disturbances as the top factor, followed by poverty-to-income ratio and age. An interactive online screening tool (https://chunyuzhang28.shinyapps.io/Hypertension/) was developed, enabling rapid depression risk prediction for hypertensive patients via input of the 12 predictors. This random forest-based ML screening tool is a valid and practical tool for clinical depression risk assessment in adult hypertensive patients. Clinicians can use it to stratify depression risk in hypertensive patients, identify high-risk subgroups (e.g., those with sleep disturbances or low poverty-to-income ratio), and implement targeted early intervention and integrated management strategies for hypertension and depression, thus optimizing clinical decision-making for comorbidity care.

Respiratory sensory irritation is the basis for many occupational exposure limits. Irritation thresholds have hitherto mainly been identified in humans by questionnaires and in mice by measuring the inhaled concentration causing 50% respiratory depression (RD50). Both methods are ethically questionable. We investigated an alternative New Approach Methodology (NAM) approach, namely the neuronal SH-SY5Y cell model expressing the sensory receptor TRPV1 in combination with random forest-based machine learning. The intracellular Ca2+ concentration was monitored during acute exposure to different concentrations of 34 organic chemicals. Potency and efficacy were determined with and without the TRPV1 antagonist capsazepine (CZ). Fifteen of the chemicals induced TRPV1 activation at some concentrations, however only phenol appeared as a true TRPV1 agonist. Using machine learning, the parameters EC20, Emax, concentration at Emax, the two first components from principal component analyses, and pH were analysed against previously published RD50 data to classify each chemical as non-irritant/irritant or as non-irritant/ intermediate/irritant. The best 2-class model (accuracy 0.90, outlier frequency 4.8%) was the one using experiments with CZ present, suggesting that the irritancy was not mediated by TRPV1 activation. The best 3-class model (accuracy 0.77, outlier frequency 5.7%) was the one using data without CZ, indicating that TRPV1 activation may play a role for intermediate irritation. The three false negative chemicals, as predicted by the NAMs, were the most irritating chemicals according to RD50 determined in vivo, indicating that other processes may also be important.

Purpose: We are aiming to build a supervised machine learning-based classifier, in order to preoperatively distinguish focal cortical dysplasia (FCD) from glioneuronal tumors (GNTs) in patients with epilepsy.Methods: This retrospective study was comprised of 96 patients who underwent epilepsy surgery, with the final neuropathologic diagnosis of either an FCD or GNTs. Seven classical machine learning algorithms (i.e., Random Forest, SVM, Decision Tree, Logistic Regression, XGBoost, LightGBM, and CatBoost) were employed and trained by our dataset to get the classification model. Ten features [i.e., Gender, Past history, Age at seizure onset, Course of disease, Seizure type, Seizure frequency, Scalp EEG biomarkers, MRI features, Lesion location, Number of antiepileptic drug (AEDs)] were analyzed in our study.Results: We enrolled 56 patients with FCD and 40 patients with GNTs, which included 29 with gangliogliomas (GGs) and 11 with dysembryoplasic neuroepithelial tumors (DNTs). Our study demonstrated that the Random Forest-based machine learning model offered the best predictive performance on distinguishing the diagnosis of FCD from GNTs, with an F1-score of 0.9180 and AUC value of 0.9340. Furthermore, the most discriminative factor between FCD and GNTs was the feature “age at seizure onset” with the Chi-square value of 1,213.0, suggesting that patients who had a younger age at seizure onset were more likely to be diagnosed as FCD.Conclusion: The Random Forest-based machine learning classifier can accurately differentiate FCD from GNTs in patients with epilepsy before surgery. This might lead to improved clinician confidence in appropriate surgical planning and treatment outcomes.

In this paper, a dual port high gain hybrid (silicon graphene) antenna is designed and investigated. Some of the important qualities of the proposed radiating structure include the following: (1) XGBoost and Random Forest-based machine learning (ML) techniques are applied to predict the reflection coefficient (|S11|) features of the designed radiator to reduce the simulation time; (2) the use of a cross slot provides the circularly polarized waves between 4.92 and 5.38 THz; (3) a 1∗2 array structure produces the high gain, i.e., 10.0 dBi within the operating frequency range; and (4) the opposite orientation of the aperture provides a polarization diversity concept, which, in turn, improves the isolation level to more than 30 dB. After comparing the results obtained from CST and HFSS, as well as those predicted from the ML algorithm, it is confirmed that the designed radiator operates from 4.71 to 5.85 THz. The good value of far-field and diversity parameters affirms the fitness of the designed radiator for wireless power transfer in the THz regime.