Enhanced weed and crop species classification using optimized machine learning and ensemble techniques

There is growing interest in canine behavioral research specifically for working dogs. Here we take advantage of a dataset of a Transportation Safety Administration olfactory detection cohort of 628 Labrador Retrievers to perform Machine Learning (ML) prediction and classification studies of behavioral traits and environmental effects. Data were available for four time points over a 12 month foster period after which dogs were accepted into a training program or eliminated. Three supervised ML algorithms had robust performance in correctly predicting which dogs would be accepted into the training program, but poor performance in distinguishing those that were eliminated (~ 25% of the cohort). The 12 month testing time point yielded the best ability to distinguish accepted and eliminated dogs (AUC = 0.68). Classification studies using Principal Components Analysis and Recursive Feature Elimination using Cross-Validation revealed the importance of olfaction and possession-related traits for an airport terminal search and retrieve test, and possession, confidence, and initiative traits for an environmental test. Our findings suggest which tests, environments, behavioral traits, and time course are most important for olfactory detection dog selection. We discuss how this approach can guide further research that encompasses cognitive and emotional, and social and environmental effects.

Understanding individual genetic factors has led to significant improvements in disease prediction, diagnosis and tailored treatment, and data well beyond what the human mind can comprehend can only be managed through machine learning (ML). It focuses on the use of ML techniques for analyzing and interpreting genomic data, particularly in their contributions to the identification of disease-associated genetic variants. ML uses one or a combination of supervised and unsupervised learning models, deep learning architectures, and ensemble methods to guide the extraction of meaningful information from complex genetic data. The feature selection techniques, including recursive feature elimination (RFE) and principal component analysis (PCA), improve model performance by minimizing dimensions and removing noise. Specific disease susceptibility can be predicted with high accuracy from the genetic markers using classification algorithms, such as random forests (RF), support vector machines (SVM), and neural networks. Clustering techniques (k-means, hierarchical clustering, etc.) are also used to identify hidden genetic subgroups associated with particular diseases. With the development of next-generation sequencing (NGS) and genome-wide association studies (GWAS), big data sets have emerged encouraging the application of ML-based approaches for instance to allow fast and accurate variant classification. CNNs (convolutional neural networks) are effective in modelling sequence-based data, while RNNs (recurrent neural networks) have demonstrated utility in recognizing pathogenic mutations. Naturallanguage processing (NLP) is another approach that further improves the variant annotation process by extracting valuable information about variants from biomedical publications. Future Work In this paper reinforcement learning in used for genetic diagnosis and treatment planning. Explainable AI methods make sure that the model is transparent and reliable and mitigate the black-box issue of deep learning models. However, challenges remain such as data heterogeneity, class imbalance, and explaining ML models. Some federated learning frameworks can aid genome privacy-preserving analysis by allowing collaborative model training without the aggregation of sensitive patient data. These approaches come with ethical questions including potential biases in genetic information resources and fair access to ML solutions in healthcare, that need to be constantly monitored. Further research could aim to improve model generalizability across different populations and develop hybrid ML methods that utilize both statistical and deep learning approaches to provide optimal predictive accuracy. This impressive potential for ML to revolutionize precision medicine lies in its ability to achieve disease prediction, targeted therapeutics development, and hence combat diseases at branch level.

Root discrimination of species is a pre-condition for studying belowground competition processes between crop and weed species. In this experiment, we tested Fourier transform mid-infrared (FT MIR)-attenuated total reflection (ATR) spectroscopy to discriminate roots of closely related crop and weed species grown in the greenhouse: maize/barnyard grass, barley/wild oat, wheat/blackgrass (Poaceae), and sugar beet/common lambsquarters (Chenopodiaceae). Fresh (moist) and dried root segments as well as ground roots were analyzed by FT MIR-ATR spectroscopy. Root absorption spectra showed species specific peak distribution and peak height. A clear separation according to species was not possible with fresh root segments. Dried root segments (including root basis, middle section, and root tip) of maize/barnyard grass and sugar beet/common lambsquarters formed completely separated species clusters. Wheat and blackgrass separated in species specific clusters when root tips were removed from cluster analysis. A clear separation of dried root segments according to species was not possible in the case of barley and wild oat. Cluster analyses of ground roots revealed a 100% separation of all tested crop and weed species combinations. Spectra grouped in Poaceae and Chenopodiaceae clusters. Within the Poaceae cluster, C3 and C4 species differed significantly in heterogeneity. Thus, root spectra reflected the degree of kinship. To quantify species proportion in root mixtures, a two- and a three-species model for species quantification in root mixtures of maize, barnyard grass, and wild oat was calculated. The models showed low standard errors of prediction (RMSEP) and high residual predictive deviation values in an external test set validation. Hence, FT MIR-ATR spectroscopy seems to be a promising tool for root research even between closely related plant species.

Early diagnosis and treatment is very essential in monitoring Brain tumor using MRI images. Convolutional Neural Networks (CNN) and Machine Learning (ML) classifiers have been widely used but there is not much work on how feature selection techniques would affect the performance of the CNN. Secondly, there is a need for investigation concerning small dataset adaptability and ML-CNN comparisons. To improve the classification accuracy, we integrate Univariate, Recursive Feature Elimination (RFE), Recursive Feature Elimination with Cross Validation (RFECV) with CNN in this study. Preprocessing, feature extraction & selection was carried out on the dataset consisting of 253 MRI images and they are classified using CNN and ML models (Logistic Regression, Decision Tree, Random Forest, Naïve Bayes). With the results 96%, CNN with Univariate Feature Selection performed better than ML classifiers, and other selection techniques. The results demonstrate that feature selection is necessary to get the best performance out of CNN models operating on small datasets. Future studies should be based on different deep learning architectures to improve classification and application in other datasets.

BACKGROUNDThe International Study Group of Pancreatic Surgery has established the definition and grading system for postpancreatectomy acute pancreatitis (PPAP). There are no established machine learning models for predicting PPAP following pancreaticoduodenectomy (PD).AIMTo explore the predictive model of PPAP, and test its predictive efficacy to guide the clinical work.METHODSClinical data from consecutive patients who underwent PD between 2016 and 2024 were retrospectively collected. An analysis of PPAP risk factors was performed, various machine learning algorithms [logistic regression, random forest, gradient boosting decision tree, extreme gradient boosting, light gradient boosting machine, and category boosting (CatBoost)] were utilized to develop predictive models. Recursive feature elimination was employed to select several variables to achieve the optimal machine algorithm.RESULTSThe study included 381 patients, of whom 88 (23.09%) developed PPAP. PPAP patients exhibited a significantly higher incidence of postoperative pancreatic fistula (55.68% vs 14.68%, P < 0.001), grade C postoperative pancreatic fistula (9.09% vs 1.37%, P = 0.001). The CatBoost algorithm outperformed other algorithms with a mean area under the receiver operating characteristic curve of 0.859 [95% confidence interval (CI): 0.814-0.905] in the training cohort and 0.822 (95%CI: 0.717-0.927) in the testing cohort. According to shapley additive explanations analysis, pancreatic texture, main pancreatic duct diameter, body mass index, estimated blood loss, and surgery time were the most important variables based on recursive feature elimination. The CatBoost algorithm based on selected variables demonstrated superior performance, with an area under the receiver operating characteristic curve of 0.837 (95%CI: 0.788-0.886) in the training cohort and 0.812 (95%CI: 0.697-0.927) in the testing cohort.CONCLUSIONWe developed the first machine learning-based predictive model for PPAP following PD. This predictive model can assist surgeons in anticipating and managing this complication proactively.

The goal of securing a network is to protect the information flowing through the network and to ensure the security of intellectual as well as sensitive data for the underlying application. To accomplish this goal, security mechanism such as Intrusion Detection System (IDS) is used, that analyzes the network traffic and extract useful information for inspection. It identifies various patterns and signatures from the data and use them as features for attack detection and classification. Various Machine Learning (ML) techniques are used to design IDS for attack detection and classification. All the features captured from the network packets do not contribute in detecting or classifying attack. Therefore, the objective of our research work is to study the effect of various feature selection techniques on the performance of IDS. Feature selection techniques select relevant features and group them into subsets. This paper implements Chi-Square, Information Gain (IG), and Recursive Feature Elimination (RFE) feature selection techniques with ML classifiers namely Support Vector Machine, Naive Bayes, Decision Tree Classifier, Random Forest Classifier, k-nearest neighbours, Logistic Regression, and Artificial Neural Networks. The methods are experimented on NSL-KDD dataset and comparative analysis of results is presented.

Parkinson's disease (PD) is a progressive neurodegenerative disorder that significantly impacts quality of life and healthcare systems. Early detection is crucial for timely interventions that can mitigate disease progression and improve patient outcomes. This study leverages advanced machine learning (ML) techniques to detect PD using speech features as non-invasive biomarkers. A dataset containing 754 features derived from sustained vowel phonations of 252 individuals (188 PD patients, 64 healthy controls) was analyzed. The dataset, originally collected by Istanbul University and publicly hosted via the UCI ML repository, was accessed through Kaggle for preprocessing and analysis. To identify the most predictive features, we employed recursive feature elimination (RFE), random forest importance, lasso regression, and the boruta algorithm—ensuring robust feature selection while reducing dimensionality. The XGBoost model, optimised using synthetic minority oversampling technique (SMOTE) for class balancing, achieved an accuracy of 96.69%, a recall of 96%, and an F1-score of 98%. Model robustness was validated through 5-fold cross-validation, yielding an average accuracy of 89.54%. These findings establish a scalable, costeffective, and non-invasive framework for early PD detection, demonstrating the potential of speech analysis and ML in neurodegenerative disease management.

Sub-Saharan Africa continues to shoulder the heaviest burden of malaria. The 2024 World Health Organisation (WHO) malaria report highlighted that Africa contributed an alarming 94% of the global cases and 95% of the deaths. In the WHO African region, progress towards elimination and management of malaria is hindered by weak health systems and a lack of traditional diagnostic methods such as microscopy and malaria Rapid Diagnostic Tests (mRDTs). The primary aim of this paper is to develop a machine learning (ML) ensemble model for malaria diagnosis using clinical and demographic data, tailored for resource-limited settings. A retrospective study was conducted using 637 patient records from Gutu Mission Hospital and Gweru Provincial Hospital in Zimbabwe. Clinical symptoms (fever, chills, abdominal pain, headache and diarrhea) and demographic features (age, gender, residence and travel history) were analysed. Data preprocessing included handling class imbalance using the Synthetic Minority Oversampling Technique (SMOTE) and feature selection using Recursive Feature Elimination (RFE). Seven individual ML models, including Logistic regression (LR), Random Forest (RF), Decision Trees (DT), Gradient Boosting (GB), K-Nearest Neighbour (KNN), Naive Bayes (NB) and XGBoost were trained and evaluated on the malaria dataset. The individual models were further combined to build, train and evaluate ensemble models such as Bagging, Stacking, Soft Voting and AdaBoost. Model performance was assessed using accuracy, precision, confusion matrices, recall, F1 score, and AUC-ROC metrics. Clinical symptoms (chills: p-value=0.001, fever: p=0.003, diarrhea: p=0.01, abdominal pain: p&amp;lt; 0.001) were statistically significant predictors of malaria. Of the demographic factors, only travel history (p=0.02) showed a significant association with malaria. Among the seven individual ML models, GB achieved the highest predictive performance (Accuracy =0.94), followed by RF (Accuracy =0.94) and XGBoost (Accuracy =0.93). The stacking ensemble model outperformed all individual ML models and other ensemble models (bagging, soft voting and AdaBoost), achieving accuracy =0.96, precision =0.95, recall =0.98, F1 score =0.96 and AUC-ROC =0.98. This study demonstrates that ML, particularly ensemble models, can be used to improve malaria diagnosis significantly. The integration of these models into a web-based application could provide a scalable and accessible diagnostic tool for healthcare workers in resource-limited settings.

Stroke is a disease that occurs in the brain and can cause both vocal and global brain dysfunction. Stroke research mainly aims to predict risk and mortality. Machine learning can be used to diagnose and predict diseases in the healthcare field, especially in stroke prediction. However, collecting medical record data to predict a disease usually makes much noise because not all variables are important and relevant to the prediction process. In this case, dimensionality reduction is essential to remove noisy (i.e., irrelevant) and redundant features. This study aims to predict stroke using Recursive Feature Elimination as feature selection, Principal Component Analysis as feature extraction, and a combination of Recursive Feature Elimination and Principal Component Analysis. The dataset used in this research is stroke prediction from Kaggle. The research methodology consists of pre-processing, SMOTE, 10-fold Cross-Validation, feature selection, feature extraction, and machine learning, which includes SVM, Random Forest, Naive Bayes, and Linear Discriminant Analysis. From the results obtained, the SVM and Random Forest get the highest accuracy value of 0.8775 and 0.9511 without using PCA and RFE, Naive Bayes gets the highest value of 0.7685 when going through PCA with selection of 20 features followed by RFE feature selection with selection of 5 features, and LDA gets the highest accuracy with 20 features from feature selection and continued feature extraction with a value of 0. 7963. It can be concluded in this study that SVM and Random Forest get the highest accuracy value without PCA and RFE techniques, while Naive Bayes and LDA show better performance using a combination of PCA and RFE techniques. The implication of this research is to know the effect of RFE and PCA on machine learning to improve stroke prediction.

Feature extraction-reduction and machine learning for fault diagnosis in PV panels

In Mild Cognitive Impairment (MCI), identifying a high risk of conversion to Alzheimer’s Disease Dementia (AD) is a primary goal for patient management. Machine Learning (ML) algorithms are widely employed to pursue data-driven diagnostic and prognostic goals. An agreement on the stability of these algorithms –when applied to different biomarkers and other conditions– is far from being reached. In this study, we compared the different prognostic performances of three supervised ML algorithms fed with multimodal biomarkers of MCI subjects obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. Random Forest, Gradient Boosting, and eXtreme Gradient Boosting algorithms predict MCI conversion to AD. They can also be simultaneously employed –with the voting procedure– to improve predictivity. AD prediction accuracy is influenced by the nature of the data (i.e., neuropsychological test scores, cerebrospinal fluid AD-related proteins and APOE ε4, cerebral structural MRI (sMRI) data). In our study, independent of the applied ML algorithms, sMRI data showed the lowest accuracy (0.79) compared to other classes. Multimodal data were helpful in the algorithms’ performances by combining clinical and biological measures. Accordingly, using the three ML algorithms, the highest accuracy (0.90) was reached by employing neuropsychological and AD-related biomarkers. Finally, the feature selection procedure indicated that the most critical variables in the respective classes were the ADAS-Cog-13 scale, the medial temporal lobe and hippocampus atrophy, and the ratio between phosphorylated Tau and Aβ42 proteins. In conclusion, our data support the notion that using multiple ML algorithms and multimodal biomarkers helps make more accurate and solid predictions.

CVDs are among the most preventable diseases globally, yet they continue to cause more than one in three deaths worldwide. Objectives This study investigates the use of machine learning (ML) techniques for the prediction of cardiovascular risk from basic clinical and lifestyle factors. The analysis used a publicly available data set of medical records of 70,000 patients that included variables on the patients' age, blood pressure, cholesterol levels, glucose levels, and lifestyle, including smoking and exercise. Deep analysis on the preprocessing data, we used Recursive Feature Elimination (RFE) with Random Forest estimator to explore key predictors. Multiple ML models (Logistic Regression, Support Vector Machine, Random Forest and Gradient Boosting) were trained by 5-fold cross validation for hyperparameter tuning. The accuracy, precision, recall, F1-score, and AUC-ROC metrics were used to assess the performance of the models. Based on the results, Gradient Boosting was the most accurate and overall, the most robust of the algorithms considered. Implication of all the available evidence: The current knowledge synthesis demonstrates a potential role for ML to improve clinical decision-making related to cardiovascular risk assessment and stratification for early prevention as well as a strong need for multi-centre trials to affirm these findings.

Background Current heart failure guidelines emphasize the importance of timely detection of subclinical left ventricular (LV) remodelling and dysfunction for more precise risk stratification of asymptomatic subjects. Both LV diastolic dysfunction (LVDD) and LV hypertrophy (LVH) as assessed by echocardiography are known independent prognostic markers of future cardiovascular events in the community. However, selective screening strategies of individuals at risk who would benefit most from in-depth cardiac phenotyping are lacking. Purpose We assess the utility of several Machine Learning (ML) classifiers built on clinical and biochemical features for detecting subclinical LV abnormalities. Methods We included 1407 participants (mean age, 51 years, 51% women) randomly recruited from the general population. We used echocardiographic parameters reflecting LV diastolic function and structure to define LV abnormalities (LVDD, n=239; LVH, n=135). After that four supervised ML algorithms (Random Forest (RF), Gradient Boosting (GD), Stochastic Gradient Descent (SGD) and Support Vector Machines (SV)) were built based on routine clinical, hemodynamic and laboratory data (features; n=61) to categorize LVDD and LVH (two prediction tasks). We applied a 10-fold stratified cross-validation set-up. Results ML classifiers exhibited a high area under the ROC (AUC) for predicting LVDD with values between 88.5% and 93.1% (Figure, left panel). Age, BMI, different components of blood pressure, antihypertensive treatment, routine biomarkers such as serum electrolytes, creatinine, blood sugar, leptin, uric acid, lipid profile, as well as blood cell counts were the top selected features for predicting LVDD. Prediction AUC of ML algorithms for detection of LVH was somewhat lower than for LVDD and ranged from 72.5% to 78.7% (Figure, right panel). The top selected features for LVH classifier were similar to those of LVDD, but also included social class, serum gamma-glutamyl transferase, fasting insulin, plasma renin activity and cortisol. ROC curves (sensitivity-1-specificity) Conclusions ML algorithms combining routinely measured clinical and laboratory data have shown high accuracy of LVDD and LVH prediction. These ML classifiers might be useful to preselect individuals at risk for further in depth echocardiographic examination, monitoring and implementation of preventive strategies in order to delay transition to disease symptoms.

Mental stress is a growing global health concern, closely linked to psychological, behavioral, and physiological disorders. Accurate and early prediction of mental stress is crucial for timely interventions and improved health outcomes. Despite numerous studies leveraging machine learning (ML) techniques for stress classification, many have overlooked the integration of systematic feature selection and comprehensive model evaluation, limiting generalizability and interpretability. To address these gaps, this study proposes a robust ML-based framework that combines optimized feature selection methods - Recursive Feature Elimination (RFE), Extra Trees (ET), and Boruta - with various classification algorithms including Random Forest (RF), K-Nearest Neighbors (KNN), Decision Tree (DT), Multilayer Perceptron (MLP), Support Vector Machine (SVM), Gradient Boosting, and voting classifier. The models were evaluated using 10-fold cross-validation and ranked using the TOPSIS multi-criteria decision-making approach. The experimental results demonstrate high predictive performance across models (accuracy ≥ 0.98), with RF, DT, MLP, and Gradient Boosting achieving perfect accuracy (1.00). Among all configurations, the RF-Boruta model emerged as the most optimal (TOPSIS score: 0.914558). These findings highlight the effectiveness of combining systematic feature optimization with ML classification for accurate and interpretable stress prediction, offering valuable insights for data-driven mental health interventions.

The application of machine-learning and Raman spectroscopy for the rapid detection of edible oils type and adulteration