RF Classifier Research Articles

DDoS Attack Detection with Machine Learning

Nowadays, Distributed Denial of Service (DDoS) attacks are a major issue in internet security. These attacks target servers or network infrastructure. Similar to an unanticipated traffic jam on highway (lagging/crash) that prevent normal traffic reach to destination. DDoS may prevent users to access any system services. Researchers and scientists have developed numerous methods and algorithms to improve the performance of DDoS detection. In this paper, a DDoS detection method utilizing machine learning is proposed. There are three type of supervised machine learning classification methods which are K-Nearest Neighbor, Multilayer Perceptron and Random Forest, are applied in the proposed work to assess the accuracy of the model in training and testing processes. RF classification provides robustness and interpretability, MLP offers deep learning capabilities for complex patterns, and K-NN delivers simplicity and adaptability for instance-based learning. Together, these methods can contribute to a comprehensive DDoS attack detection system using machine learning. There are two types of classification setups: binary and multi-class classification. Binary classification involves identifying traffic as either a DDoS attack or normal using the NSL-KDD dataset. Multi-class classification, on the other hand, distinguishes between various types of DDoS attacks (such as DoS, Probe, U2R, and Sybil) and normal traffic using the NSL-KDD dataset. Feature engineering is also involved in this experiment to convert the categorical features into numerical values for detecting DDoS attack. Our model's performance was effective compared to other machine learning methods. RF achieved the highest accuracy rates: 99.35% in binary classification and 97.71% in multi-class classification. K-NN followed with 99.15% in binary and 97.35% in multi-class classification, while MLP achieved 90.63% in binary and 84.33% in multi-class classification.

Developing molecular classifiers to detect environmental stressors, smolt stages and morbidity in coho salmon, Oncorhynchus kisutch

Aquatic species are increasingly confronted with environmental stressors because of climate change. Although molecular technologies have advanced our understanding of how organisms respond to stressors in laboratory settings, the ability to detect physiological responses to specific stressors under complex field conditions remains underdeveloped. This research applied multi-stressor challenge trials on coho salmon, employing the “Salmon Fit-Chips” genomic tool and a random forest-based classification model to develop classifiers predictive for chronic thermal and hypoxic stress, as well as salinity acclimation, smolt stage and morbidity status. The study also examined how smolts and de-smolts (smolts not having entered SW during the smolt window) responded transcriptionally to exposure to saltwater. Using RF classifiers optimized with 4 to 12 biomarkers, we identified transcriptional signatures that accurately predicted the presence of each stressor and physiological state, achieving prediction accuracy rates between 86.8 % and 100 %, regardless of other background stressors present. Stressor recovery time was established by placing fish back into non-stressor conditions after stress exposure, providing important context to stressor detections in field applications. Recovery from thermal and hypoxic stress requires about 3 and 2 days, respectively, with >3 days needed for re-acclimation to freshwater for seawater acclimated fish. The study also found non-additive (synergistic) effects of multiple stressors on mortality risk. Importantly, osmotic stress associated with de-smolts was the most important predictor of mortality. In saltwater, de-smolts exposed to salinity, high temperature, and hypoxia experienced a 9-fold increase in mortality compared to those only exposed to saltwater, suggesting a synergistic response to multiple stressors. These findings suggest that delays in hatchery releases to support release of larger fish need to be carefully scrutinized to ensure fish are not being released as de-smolts, which are highly susceptible to additional climate-induced stressors like rising temperatures and reduced dissolved oxygen levels in the marine environment.

Parcel-Based Sugarcane Mapping Using Smoothed Sentinel-1 Time Series Data

The timely and accurate mapping of sugarcane cultivation is significant to ensure the sustainability of the sugarcane industry, including sugarcane production, rural society, sugar futures, and crop insurance. Synthetic aperture radar (SAR), due to its all-weather and all-time imaging capability, plays an important role in mapping sugarcane cultivation in cloudy areas. However, the inherent speckle noise of SAR data worsens the “salt and pepper” effect in the sugarcane map. Therefore, in previous studies, an additional land cover map or optical image was still required. This study proposes a new application paradigm of time series SAR data for sugarcane mapping to tackle this limitation. First, the locally estimated scatterplot smoothing (LOESS) smoothing technique was exploited to reconstruct time series SAR data and reduce SAR noise in the time domain. Second, temporal importance was evaluated using RF MDA ranking, and basic parcel units were obtained only based on multi-temporal SAR images with high importance values. Lastly, the parcel-based classification method, combining time series smoothing SAR data, RF classifier, and basic parcel units, was used to generate a sugarcane extent map without unreasonable sugarcane spots. The proposed paradigm was applied to map sugarcane cultivation in Suixi County, China. Results showed that the proposed paradigm was able to produce an accurate sugarcane cultivation map with an overall accuracy of 96.09% and a Kappa coefficient of 0.91. Compared with the pixel-based classification result with original time series SAR data, the new paradigm performed much better in reducing the “salt and pepper” spots and improving the completeness of the sugarcane plots. In particular, the unreasonable non-vegetation spots in the sugarcane map were eliminated. The results demonstrated the efficacy of the new paradigm for mapping sugarcane cultivation. Unlike traditional methods that rely on optical remote sensing data, the new paradigm offers a high level of practicality for mapping sugarcane in large regions. This is particularly beneficial in cloudy areas where optical remote sensing data is frequently unavailable.

Contributions of brain regions to machine learning-based classifications of attention deficit hyperactivity disorder (ADHD) utilizing EEG signals

Objective The study presented focuses on the creation of a machine learning (ML) model that uses electrophysiological (EEG) data to identify kids with attention deficit hyperactivity disorder (ADHD) from healthy controls. The EEG signals are acquired during cognitive tasks to distinguish children with ADHD from their counterparts. Methodology The EEG data recorded in cognitive exercises was filtered using low pass Bessel filter and notch filters to remove artifacts, by the data set owners. To identify unique EEG patterns, we used many well-known classifiers, including Naïve Bayes (NB), Random Forest, Decision Tree (DT), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), AdaBoost and Linear Discriminant Analysis (LDA), to identify distinct EEG patterns. Input features comprised EEG data from nineteen channels, individually and in combination. Findings Study indicates that EEG-based categorization can differentiate between individuals with ADHD and healthy individuals with accuracy of 84%. The RF classifier achieved a maximum accuracy of 0.84 when particular region combinations were used. Evaluation of classification performance utilizing hemisphere-specific EEG data yielded promising outcomes, particularly in the right hemisphere channels. Novelty The study goes beyond traditional methodologies by investigating the effect of regional data on categorization results. The contributions of various brain regions to these classifications are being extensively researched. Understanding the role of different brain regions in ADHD can lead to better diagnosis and treatment options for individuals with ADHD. The study of categorization ability, utilizing EEG data specific to each hemisphere, particularly channels in the right hemisphere region, provides further granularity to the findings.

Comparison of the Performance of Random Forest and K-Nearest Neighbor in Classifying Leukemia Using Principal Component Analysis

Leukemia is the most common blood cancer in Asia, one of which is Indonesia. Leukemia can affect blood cells, bone marrow, lymph nodes and other parts of the lymphatic system. One way to detect leukemia is to use microarray technology by applying gene expression. Microarrays have a very large number of genes so it is necessary to reduce the number of genes in order to eliminate irrelevant features and increase the accuracy of the classification process. The leukemia feature/gene reduction process was carried out using PCA and the classification process was carried out using RF and KNN. The accuracy results from the RF classification method using 100 n_estimators were 78.57%, while using the KNN method the accuracy results with K=1 were 78.57%, K=3 and 5 were 85.71%, and K=7 and 9 were 71.42%. The best accuracy results use KNN with K=3 and 5.

Classification of Breast Cancer Subtypes using Microarray RNA Expression Data

Breast cancer is a heterogeneous disease that involves molecular alteration, cellular alterations, and clinical outcome for which the classification of Breast cancer remains a challenge to diagnose. Current practice uses immunohistochemistry markers and clinical variables to classify Breast cancer, but this approach has limitations due to the inclusion of other tumour subtypes and healthy individuals. Machine learning approaches based on mRNA expression data offer new possibilities for researchers to investigate the potential of molecular biomarkers as one of the diagnostic characteristics. The purpose of this study is to evaluate features (genes) rank through feature selection method for Breast cancer diagnostic test. Three feature selection methods of IG, relief and mRMR were applied and subsets of top 100, 50, 25, 10, 5 and 3 were created. Each subset was tested with SVM, LR and RF classifiers and its performance was assessed using confusion matrix. The result of this study found that the feature selection of IG, reliefF and mRMR was able to achieve highest accuracy with SVM, LR and RF classifier. mRMR with RF classifier achieved highest accuracy with the least number of top rank genes with 25 genes. Hybrid feature selection approached (mRMR + SVM) improved accuracy of top 3 highest rank genes using SVM, LR and RF classifier. Future work should aim to use other feature selection methods and classifiers to explore the classification accuracy with the least features subset in multiclass cancer dataset.

Prediction of Interactomic HUB Genes in Periodontitis With Acute Myocardial Infarction.

Acute myocardial infarction (AMI) risk correlates with C-reactive protein (CRP) levels, suggesting systemic inflammation is present well before AMI. Studying different types of periodontal disease (PD), extremely common in individuals at risk for AMI, has been one important research topic. According to recent research, AMI and PD interact via the systemic production of certain proinflammatory and anti-inflammatory cytokines, small signal molecules, and enzymes that control the onset and development of both disorders' chronic inflammatory reactions. This study uses machine learning to identify the interactome hub biomarker genes in acute myocardial infarction and periodontitis. GSE208194 and GSE222883 were chosen for our research after a thorough search using keywords related to the study's goal from the gene expression omnibus (GEO) datasets. DEGs were identified from the GEOR tool, and the hub gene was identified using Cytoscape-cytohubba. Using expression values, Random Forest, Adaptive Boosting, and Naive Bayes, widgets-generated transcriptomics data, were labelled, and divided into 80/20 training and testing data with cross-validation. ROC curve, confusion matrix, and AUC were determined. In addition, Functional Enrichment Analysis of Differentially Expressed Gene analysis was performed. Random Forest, AdaBoost, and Naive Bayes models with 99%, 100%, and 75% AUC, respectively. Compared to RF, AdaBoost, and NB classification models, AdaBoost had the highest AUC. Categorization algorithms may be better predictors than important biomarkers. Machine learning model predicts hub and non-hub genes from genomic datasets with periodontitis and acute myocardial infarction.

The Artistry of Stochastic Gradient Boosting and Gradient Boosting Classifiers in Chronic Renal Disease Classification

Chronic kidney failure is a medical disorder that impairs the kidneys' overall capacity to filter dangerous substances from your blood and maintain overall health. Anemia, weakened bones, poor diet, trauma, and decreased blood pressure are some of the factors that contribute to chronic kidney disease. Anemia, weakened bones, poor diet, trauma, and decreased blood pressure are some of the factors that contribute to chronic kidney disease. This study proposes using a several unsupervised algorithms, evaluate their performance, and determine the optimal combinations with higher accuracy. In addition to scaling features and applying the method to balance the data, this list of processes includes correctly imputed missing data points. The present study has employed six supervised algorithms, including DT, RF, SGB, XgBoost, Ada Boost and Gradient Boosting Classifiers has integrated them with techniques for selecting features based on the Pearson Correlation Coefficient. Classifying clinical data of chronic kidney diseases(CKD) and Non-CKD with an overall accuracy of 99.2% has been achieved by integrating feature reduction approaches with Stochastic Gradient Boosting and Gradient Boosting Classifier. These models were put to the test using a collection of data on chronic kidney illness from the University of California at Irvine, this had four hundred entries of data with twenty-six attributes. The outcomes of various models are examined. The model built with Stochastic Gradient Boosting and Gradient Boosting Classifier method performed the best in terms of correctness using 24 attributes for the small dataset, based on the comparison. The final phase of this study will examine how effectively the machine learning system predicts chronic kidney failure with respect to precision, recall, accuracy, & F1-Score.

The Artistry of Stochastic Gradient Boosting and Gradient Boosting Classifiers in Chronic Renal Disease Classification

Chronic kidney failure is a medical disorder that impairs the kidneys' overall capacity to filter dangerous substances from your blood and maintain overall health. Anemia, weakened bones, poor diet, trauma, and decreased blood pressure are some of the factors that contribute to chronic kidney disease. Anemia, weakened bones, poor diet, trauma, and decreased blood pressure are some of the factors that contribute to chronic kidney disease. This study proposes using a several unsupervised algorithms, evaluate their performance, and determine the optimal combinations with higher accuracy. In addition to scaling features and applying the method to balance the data, this list of processes includes correctly imputed missing data points. The present study has employed six supervised algorithms, including DT, RF, SGB, XgBoost, Ada Boost and Gradient Boosting Classifiers has integrated them with techniques for selecting features based on the Pearson Correlation Coefficient. Classifying clinical data of chronic kidney diseases(CKD) and Non-CKD with an overall accuracy of 99.2% has been achieved by integrating feature reduction approaches with Stochastic Gradient Boosting and Gradient Boosting Classifier. These models were put to the test using a collection of data on chronic kidney illness from the University of California at Irvine, this had four hundred entries of data with twenty-six attributes. The outcomes of various models are examined. The model built with Stochastic Gradient Boosting and Gradient Boosting Classifier method performed the best in terms of correctness using 24 attributes for the small dataset, based on the comparison. The final phase of this study will examine how effectively the machine learning system predicts chronic kidney failure with respect to precision, recall, accuracy, & F1-Score.

Heterocyclic-Based Analogues against Sarcine-Ricin Loop RNA from Escherichia coli: In Silico Molecular Docking Study and Machine Learning Classifiers.

Heterocyclic-based drugs have strong bioactivities, are active pharmacophores, and are used to design several antibacterial drugs. Due to the diverse biodynamic properties of well-known heterocyclic cores, such as quinoline, indole, and its derivatives, they have a special place in the chemistry of nitrogen-containing heterocyclic molecules. The objective of this study is to analyze the interaction of several heterocyclic molecules using molecular docking and machine learning approaches to find out the possible antibacterial drugs. The molecular docking analysis of heterocyclic-based analogues against the sarcin-Ricin Loop RNA from E. coli with a C2667-2'-OCF3 modification (PDB ID: 6ZYB) is discussed. Many heterocyclic-based derivatives show several residual interaction, affinity, and hydrogen bonding with sarcin-Ricin Loop RNA from E. coli with a C2667-2'-OCF3 alteration which are identified by the investigation of in silico molecular docking analysis of such heterocyclic derivatives. The dataset from the molecular docking study was used for additional optimum analysis, and the molecular descriptors were classified using a variety of machine learning classifiers, including the GB Classifier, CB Classifier, RF Classifier, SV Classifier, KNN Classifier, and Voting Classifier. The research presented here showed that heterocyclic derivatives may operate as potent antibacterial agents when combined with other compounds to produce highly efficient antibacterial agents.

Machine learning screening tools for the prediction of extraction yields of pharmaceutical compounds from wastewaters

Pharmaceutical compounds have become an increasingly important source of pollutants in wastewaters being conventional treatments ineffective in removing them, so they are commonly discharged into the environment. Pharmaceuticals can be successfully removed using liquid-liquid extraction, and COSMO-RS can be used to predict interactions and identify the most promising solvents. However, COSMOtherm models cannot account for key process parameters, which reduces the accuracy of these computational models. Therefore, there is a need for alternative computational approaches to accurately predict the extraction yields of pharmaceuticals which can incorporate both processing and interaction variables. This work used machine learning to predict the extraction yield of eleven pharmaceuticals using eight solvents. Six regression models and two classification models were explored. The best performance was obtained with ANN regressor (test MAE: 4.510, test R2: 0.884) and RF classifier (test accuracy: 0.938, test recall: 0.974). The RF regression analysis and classification also showed key extraction yield features: solvent-to-feed ratio, n–octanol–water partition coefficient, hydrogen bond and Van der Waals contributions to excess enthalpy, and pH distance to nearest pKa. Machine learning showed as an excellent tool for screening and selecting the most promising solvents and process conditions to remove pharmaceuticals from wastewater.

Prognostic Evaluation, Prediction and Regimen of Diseases Using SVM, NB and RF Classifiers

Prognostic Evaluation, Prediction and Regimen of Diseases Using SVM, NB and RF Classifiers

Enhancing Spam Email Classification Using Effective Preprocessing Strategies and Optimal Machine Learning Algorithms

Objective: This article proposes a content-based spam email classification by applying various text pre-processing techniques. NLP techniques have been applied to pre-process the content of an email to get the optimal performance of spam email classification using machine learning. Method: Several combinations of pre-processing methods, such as stopping, removing tags, converting to lower case, removing punctuation, removing special characters, and natural language processing, were applied to the extracted content from the email with machine learning algorithms like NB, SVM, and RF to classify an email as ham or spam. The standard datasets like Enron and SpamAssassin, along with the personal email dataset collected from Yahoo Mail, are used to evaluate the performance of the models. Findings: Applying stemming in pre-processing to the RF classifier yielded the best results, achieving 99.2% accuracy on the SpamAssassin dataset and 99.3% accuracy on the Enron dataset. Lemmatization followed closely with 99% accuracy. In real-world testing on a personal Yahoo email dataset, the proposed method significantly improved accuracy from 89.82% to 97.28% compared to the email service provider's built-in classifier. Additionally, the study found that SVM performs accurately when stop words are retained. Novelty: This article introduces a unique perspective by highlighting the fine-tuning of pre-processing techniques. The focus is on removing tags and certain special characters, while retaining those that improve spam email classification accuracy. Unlike prior works that primarily emphasize algorithmic approaches and pre-defined processing functions, our research delves into the intricacies of data preparation, showcasing its significant impact on spam email classifiers. These findings emphasize the crucial role of pre-processing and contribute to a more nuanced understanding of effective strategies for robust spam detection. Keywords: Spam, Classification, Pre-processing, NLP, Machine Learning

A Robust Approach for Ulcer Classification/Detection in WCE Images

Wireless Capsule Endoscopy (WCE) is a medical diagnostic technique recognized for its minimally invasive and painless nature for the patients. It uses remote imaging techniques to explore various segments of the gastrointestinal (GI) tract, particularly the hard-to-reach small intestine, making it an effective alternative to traditional endoscopic techniques. However, physicians face a significant challenge when it comes to analyzing a large number of endoscopic images due to the effort and time required. It is therefore imperative to implement aided-diagnostic systems capable of automatically detecting suspicious areas for subsequent medical assessment. In this paper, we present a novel approach to identify gastrointestinal tract abnormalities from WCE images, with a particular focus on ulcerated areas. Our approach involves the use of the Median Robust Extended Local Binary Pattern (MRELBP) descriptor, which effectively overcomes the challenges faced when WCE image acquisition, such as variations in illumination and contrast, rotation, and noise. Using machine learning algorithms, we conducted experiments on the extensive Kvasir-Capsule dataset, and subsequently compared our results with recent relevant studies. Noteworthy is the fact that our approach achieved an accuracy of 97.04% with the SVM (RBF) classifier and 96.77% with the RF classifier.

XAI-empowered IoT multi-sensor system for real-time milk adulteration detection

In response to the critical issue of milk adulteration jeopardizing both the nutritional integrity of milk and the health of consumers, this paper presents an innovative Artificial Intelligence (AI) enabled Internet of Things (IoT) based multi-sensor system. The escalating consumption of milk as a pivotal nutritional source necessitates robust measures to ensure its safety and quality. Traditional methods of detecting adulteration have shown limitations, prompting the development of an automated and advanced approach. The proposed system integrates various sensors capable of real-time measurement, including pH, electrical conductivity (EC), temperature, gas parameters, and Volatile Organic Compounds (VOC) parameters. This comprehensive approach extends to measuring key constituents of milk samples like Fat, Protein, Solids Not Fat (SNF), Lactose, and Gravity values. To address specific adulterants—Urea, Starch, Sodium Bicarbonate, Maltodextrin, and Formaldehyde—a machine learning-based ensemble technique is employed for classification. This ensemble method outperforms conventional algorithms like RF, Light GBM, and Extra Trees Classifiers, achieving an impressive 96% accuracy rate in detecting adulterants within the milk dataset. The pivotal contribution of this study lies in the development of an IoT-based data acquisition device that seamlessly integrates with the sensor system, enabling efficient and precise measurements. Additionally, XAI is used to analyse the results obtained by the proposed model. For this, a framework called SHAP (SHapley Additive exPlanations) analysis is employed to elucidate the decision-making process of the ensemble model, enhancing the interpretability of results. By virtue of its real-time monitoring capabilities and accurate classification, the AI-enabled IoT-based multi-sensor system emerges as a promising solution for addressing milk adulteration. This innovation holds the potential to bolster milk quality control measures in the dairy industry. The system's ability to swiftly detect and categorize adulterants underscores its significance in combating the pervasive issue of compromised milk quality, thereby ensuring consumer safety and fostering industry integrity.

Classification and quantification of adulterants in Alaska pollock based on fourier transform near-infrared spectroscopy with chemometrics

Alaska pollock is one of the most economically valuable and consumed cod species, but the adulterants in the market affect its quality and value. This study used a combination of FT-NIR spectroscopy and chemometrics to construct a qualitative and quantitative analysis model. Before the establishment of detection models, different pre-processing techniques were used to eliminate background interference. Principal component analysis (PCA) and analysis of variance (ANOVA) showed the feasibility of spectral methods for the detection of Alaska pollock adulteration. The modeling results show that the RF classification model had the most suitable prediction results (accuracy: 92.18%, precision: 98.04%, recall: 92.59%, and F1-score: 95.29%), and SVR was the optimal prediction model for quantitative analysis, with an RC2 of 0.941, an RMSEC of 0.060, an RP2 of 0.946, and an RMSEP of 0.062. These results indicate that, based on the developed detection model, adulterants in Alaska pollock could be rapidly detected and accurately quantified. In addition, it provides a reliable basis for rapid and non-destructive detection of and has great potential for determining other seafood species.

Constructing the Optimal Classification Model for Benign and Malignant Breast Tumors Based on Multifeature Analysis from Multimodal Images.

The purpose of this study was to fuse conventional radiomic and deep features from digital breast tomosynthesis craniocaudal projection (DBT-CC) and ultrasound (US) images to establish a multimodal benign-malignant classification model and evaluate its clinical value. Data were obtained from a total of 487 patients at three centers, each of whom underwent DBT-CC and US examinations. A total of 322 patients from dataset 1 were used to construct the model, while 165 patients from datasets 2 and 3 formed the prospective testing cohort. Two radiologists with 10-20years of work experience and three sonographers with 12-20years of work experience semiautomatically segmented the lesions using ITK-SNAP software while considering the surrounding tissue. For the experiments, we extracted conventional radiomic and deep features from tumors from DBT-CCs and US images using PyRadiomics and Inception-v3. Additionally, we extracted conventional radiomic features from four peritumoral layers around the tumors via DBT-CC and US images. Features were fused separately from the intratumoral and peritumoral regions. For the models, we tested the SVM, KNN, decision tree, RF, XGBoost, and LightGBM classifiers. Early fusion and late fusion (ensemble and stacking) strategies were employed for feature fusion. Using the SVM classifier, stacking fusion of deep features and three peritumoral radiomic features from tumors in DBT-CC and US images achieved the optimal performance, with an accuracy and AUC of 0.953 and 0.959 [CI: 0.886-0.996], a sensitivity and specificity of 0.952 [CI: 0.888-0.992] and 0.955 [0.868-0.985], and a precision of 0.976. The experimental results indicate that the fusion model of deep features and peritumoral radiomic features from tumors in DBT-CC and US images shows promise in differentiating benign and malignant breast tumors.

Plant leaf disease detection using hybrid feature extraction techniques

Excessive yield losses are due to delayed identification of leaf diseases, resulting in lower farmer revenue. To address this, early and precise detection methods are crucial. Sunflower, a vital commercial crop in India for edible oil production, faces issues like leaf blight, downy mildew, powdery mildew, and leaf spot. Our research aims to create an algorithm that effectively identifies these diseases using hybrid feature extraction techniques. We compiled sunflower leaf images from village databases and farm visits. Initially, color images were converted to grayscale, and the noise was reduced using Gaussian filtering during image pre-processing. Disease-affected areas were segmented using edge-based approaches. Hybrid feature extraction was then employed to encompass color and texture-related attributes. The K-Nearest Neighbors (KNN) algorithm facilitated disease classification. The proposed algorithm’s performance was benchmarked against SVM, RF and DT classifiers. Remarkably, our algorithm achieved exceptional accuracy: 98% for leaf blight, 97.3% for downy mildew, 95% for powdery mildew, and 96.5% for leaf spot detection. These results underscore the algorithm’s effectiveness in combating plant diseases and enhancing agricultural productivity.

Response of salivary microbiome to temporal, environmental, and surface characteristics under in vitro exposure

The microbiome of saliva stains deposited at crime scenes and in everyday settings is valuable for forensic investigations and environmental ecology. However, the dynamics and applications of microbial communities in these saliva stains have not been fully explored. In this study, we analyzed saliva samples that were exposed to indoor conditions for up to 1 year and to different carriers (cotton, sterile absorbent cotton swab, woolen, dacron) in both indoor and outdoor environments for 1 month using high-throughput sequencing. The analysis of microbial composition and Mfuzz clustering showed that the salivary flora, specifically Streptococcus (cluster7), which was associated with microbial contamination, remained stable over short periods of time. However, prolonged exposure led to significant differences due to the invasion of environmental bacteria such as Pseudomonas and Achromobacter. The growth and colonization of environmental flora were promoted by humidity. The neutral model predictions indicated that the assembly of salivary microbial communities in outdoor environments was significantly influenced by stochastic processes, with environmental characteristics having a greater impact on community change compared to surface characteristics. By incorporating data from previous studies on fecal and vaginal secretion microbiology, we developed RF and XGBoost classification models that achieved high accuracy (>98 %) and AUC (>0.8). Additionally, a RF regression model was created to determine the time since deposition (TsD) of the stains. Time inference models yielded a mean absolute error (MAE) of 7.1 days for stains exposed for 1 year and 14.2 h for stains exposed for 14 days. These findings enhance our understanding of the changes in the microbiome of saliva stains over time, in different environments, and on different surfaces. They also have potential applications in assessing potential microbial contamination, identifying body fluids, and inferring the time of deposition.

A Study of Machine Learning Algorithms Performance Analysis in Disease Classification

Background Because there are no symptoms, it might be difficult to detect CKD in its early stages. One of the main causes of CKD is diabetes mellitus (DM), and early detection of the condition can assist individuals in obtaining prompt treatment. Because this illness has no early signs and is only discovered after the kidneys have gone through 25% damage, early-stage prediction is not very likely. This is the key factor driving the need for early CKD prediction. Objective The objective of the paper is to find the best-performing learning algorithms that can be used to predict chronic kidney disease (CKD) at an earlier stage. Methods This research aimed to compare different machine learning algorithms used in different disease predictions by various researchers. In this comparative study, machine learning algorithms like Logistic Regression, K-Nearest Neighbor, Decision Tree, Support Vector Machine, Artificial Neural Network, Random Forest, Composite Hypercube on Iterated Random Projection, Naïve Bayes, J48, Ensembling, Multi-Layer Perceptron, Deep Neural Network, Autoencoder, and Long Short-Term Memory are used in disease classification. Results Each classification model is well tested in a different dataset, and out of these models, RF, DNN, and NB classification techniques give better performance in Diabetes and CKD prediction. Conclusion The RF, DNN, and NB classification algorithms worked well and achieved 100% accuracy in predicting diseases.