Boruta Algorithm Research Articles

Substrate-Assisted Laser-Induced Breakdown Spectroscopy Combined with Variable Selection and Extreme Learning Machine for Quantitative Determination of Fenthion in Soybean Oil

The quality and safety of edible vegetable oils are closely related to human life and health, meaning it is of great significance to explore the rapid detection methods of pesticide residues in edible vegetable oils. This study explored the applicability potential of substrate-assisted laser-induced breakdown spectroscopy (LIBS) for quantitatively determining fenthion in soybean oils. First, we explored the impact of laser energy, delay time, and average oil film thickness on the spectral signals to identify the best experimental parameters. Afterward, we quantitatively analyzed soybean oil samples using these optimized conditions and developed a full-spectrum extreme learning machine (ELM) model. The model achieved a prediction correlation coefficient (RP2) of 0.8417, a root mean square error of prediction (RMSEP) of 167.2986, and a mean absolute percentage error of prediction (MAPEP) of 26.46%. In order to enhance the prediction performance of the model, a modeling method using the Boruta algorithm combined with the ELM was proposed. The Boruta algorithm was employed to identify the feature variables that exhibit a strong correlation with the fenthion content. These selected variables were utilized as inputs for the ELM model, with the RP2, RMSEP, and MAPEP of Boruta-ELM being 0.9631, 71.4423, and 10.06%, respectively. Then, the genetic algorithm (GA) was used to optimize the parameters of the Boruta-ELM model, with the RP2, RMSEP, and MAPEP of GA-Boruta-ELM being 0.9962, 11.005, and 1.66%, respectively. The findings demonstrate that the GA-Boruta-ELM model exhibits excellent prediction capability and effectively predicts the fenthion contents in soybean oil samples. It will be valuable for the LIBS quantitative detection and analysis of pesticide residues in edible vegetable oils.

The effect of covariates on Soil Organic Matter and pH variability: a digital soil mapping approach using random forest model

ABSTRACT This research focuses on understanding the spatial variation of Soil Organic Matter (SOM) and pH levels in the North of Morocco. The study employs a comprehensive approach to enhance predictive modelling, incorporating the Boruta algorithm for effective environmental covariates selection and optimizing model parameters through hyperparameter optimization. Utilizing a Random Forest (RF) model with remote sensing indices and topographic features, the research predicts SOM and pH to identify key contributors to their spatial variability. SOM prediction saw significant success, with a notable correlation to remote sensing indices such as the RVI, NDVI, and TNDVI. These indices, indicative of vegetation health and productivity, emerged as primary influencers of SOM. In comparison, the influence of topographic features like elevation, slope, and aspect was found to be less significant. Conversely, predicting pH was challenging due to the minimal spatial variability within the dataset. Addressing this limitation could involve dataset expansion or alternative models for low-correlated data handling. Despite the RF model’s limited efficacy in pH prediction, an observable correlation between SOM and pH was identified, consistent with prior research. Areas with higher SOM exhibited lower pH values, indicating relative soil acidification from organic matter decomposition. The study’s RF model demonstrated potential in SOM prediction using remote sensing indices, but enhancing pH prediction is essential. Future research may explore dataset expansion, diverse sampling, or testing alternative predictive models for better performance with low-correlated datasets. The study offers valuable insights for advanced predictive model development and enriches understanding of soil management practices.

Machine learning in the prediction of massive transfusion in trauma: a retrospective analysis as a proof-of-concept.

Early administration and protocolization of massive hemorrhage protocols (MHP) has been associated with decreases in mortality, multiorgan system failure, and number of blood products used. Various prediction tools have been developed for the initiation of MHP, but no single tool has demonstrated strong prediction with early clinical data. We sought to develop a massive transfusion prediction model using machine learning and early clinical data. Using the National Trauma Data Bank from 2013 to 2018, we included severely injured trauma patients and extracted clinical features available from the pre-hospital and emergency department. We subsequently balanced our dataset and used the Boruta algorithm to determine feature selection. Massive transfusion was defined as five units at 4h and ten units at 24h. Six machine learning models were trained on the balanced dataset and tested on the original. A total of 326,758 patients met our inclusion with 18,871 (5.8%) requiring massive transfusion. Emergency department models demonstrated strong performance characteristics with mean areas under the receiver-operating characteristic curve of 0.83. Extreme gradient boost modeling slightly outperformed and demonstrated adequate predictive performance with pre-hospital data only, as well as 4-h transfusion thresholds. We demonstrate the use of machine learning in developing an accurate prediction model for massive transfusion in trauma patients using early clinical data. This research demonstrates the potential utility of artificial intelligence as a clinical decision support tool.

P402 Early diagnosis of Crohn’s disease with perianal lesions as first clinical symptom: Development and validation of a new MR imaging-based model disease

Abstract Background Early diagnosis is crucial for the treatment and prognosis of Crohn's disease. Perianal fistulae or abscesses may present as the initial manifestation of Crohn's disease (CD) in patients. Despite magnetic resonance imaging (MRI) is the gold standard for the clinical assessment of perianal fistulae, there is shortage of MRI properties for clinical use in early perianal fistula diagnosis in CD. Methods Data from patients with a definitive diagnosis of PFCD and cryptoglandular disease (CGD) between September 2013 and December 2020 were examined retrospectively. The data was splited into training and test sets at a 7:3 ratio. The MRI characteristics were reviewed by two gastroenterology radiologists in consensus. The characteristics were then used to train a random forest classification model to predict a diagnosis of PFCD. The validity of the model was assessed using the area under the receiver operating characteristic curve (AUROC) and confusion matrix. Predictor importance was determined using the Shapley additive explanation (SHAP) framework. Results A total of 144 patients diagnosed with PFCD and 406 patients diagnosed with CGD were included. Utilizing the Boruta algorithm, 12 significant features which confirmed important for the diagnosis of PFCD, and a random forest classification model was developed. The model demonstrated an area under the receiver operating characteristic curve (AUROC) of 0.9294 (95% CI: 0.8706-0.9882), an accuracy of 0.9515 (95% CI: 0.9067-0.9788) and a kappa of 0.8779. SHAP values was utilized to estimate the variable importance, and the top five features were ulceration in rectal wall, thickened rectal wall, T2 hyperintensity, submucosal fistula and age. The model for predicting PFCD diagnosis based on MRI was dynamically visualised using a web display approach (https://doctoryang1971.shinyapps.io/forceplot/). Conclusion We have developed a multi-modal visual dynamic prediction model that can be effectively used to guide the early diagnosis of patients with Crohn's disease who present with perianal fistula at their first clinical visit. PFCD is likely to originate as transmural ulcers leading to submucosal inflammation or fistulae before penetrating the anal sphincter and developing into complex perianal fistulae.

Novel Associations Between Mid-Pregnancy Cardiovascular Biomarkers and Preeclampsia: An Explorative Nested Case-Control Study.

Prediction of women at high risk of preeclampsia is important for prevention and increased surveillance of the disease. Current prediction models need improvement, particularly with regard to late-onset preeclampsia. Preeclampsia shares pathophysiological entities with cardiovascular disease; thus, cardiovascular biomarkers may contribute to improving prediction models. In this nested case-control study, we explored the predictive importance of mid-pregnancy cardiovascular biomarkers for subsequent preeclampsia. We included healthy women with singleton pregnancies who had donated blood in mid-pregnancy (~ 18weeks' gestation). Cases were women with subsequent preeclampsia (n = 296, 10% of whom had early-onset preeclampsia [< 34weeks]). Controls were women who had healthy pregnancies (n = 333). We collected data on maternal, pregnancy, and infant characteristics from medical records. We used the Olink cardiovascular II panel immunoassay to measure 92 biomarkers in the mid-pregnancy plasma samples. The Boruta algorithm was used to determine the predictive importance of the investigated biomarkers and first-trimester pregnancy characteristics for the development of preeclampsia. The following biomarkers had confirmed associations with early-onset preeclampsia (in descending order of importance): placental growth factor (PlGF), matrix metalloproteinase (MMP-12), lectin-like oxidized LDL receptor 1, carcinoembryonic antigen-related cell adhesion molecule 8, serine protease 27, pro-interleukin-16, and poly (ADP-ribose) polymerase 1. The biomarkers that were associated with late-onset preeclampsia were BNP, MMP-12, alpha-L-iduronidase (IDUA), PlGF, low-affinity immunoglobulin gamma Fc region receptor II-b, and T cell surface glycoprotein. Our results suggest that MMP-12 is a promising novel preeclampsia biomarker. Moreover, BNP and IDUA may be of value in enhancing prediction of late-onset preeclampsia.

Transcriptomic Changes in the Myocardium and Coronary Artery of Donation after Circulatory Death Hearts following Ex Vivo Machine Perfusion.

Donation after circulatory death (DCD) hearts are predominantly maintained by normothermic blood perfusion (NBP). Nevertheless, it was shown that hypothermic crystalloid perfusion (HCP) is superior to blood perfusion to recondition left ventricular (LV) contractility. However, transcriptomic changes in the myocardium and coronary artery in DCD hearts after HCP and NBP have not been investigated yet. In a pig model, DCD hearts were harvested and maintained for 4 h by NBP (DCD-BP group, N = 8) or HCP with oxygenated histidine-tryptophane-ketoglutarate (HTK) solution (DCD-HTK, N = 8) followed by reperfusion with fresh blood for 2 h. In the DCD group (N = 8), hearts underwent reperfusion immediately after procurement. In the control group (N = 7), no circulatory death was induced. We performed transcriptomics from LV myocardial and left anterior descending (LAD) samples using microarrays (25,470 genes). We applied the Boruta algorithm for variable selection to identify relevant genes. In the DCD-BP group, compared to DCD, six genes were regulated in the myocardium and 1915 genes were regulated in the LAD. In the DCD-HTK group, 259 genes were downregulated in the myocardium and 27 in the LAD; and 52 genes were upregulated in the myocardium and 765 in the LAD, compared to the DCD group. We identified seven genes of relevance for group identification: ITPRIP, G3BP1, ARRDC3, XPO6, NOP2, SPTSSA, and IL-6. NBP resulted in the upregulation of genes involved in mitochondrial calcium accumulation and ROS production, the reduction in microvascular endothelial sprouting, and inflammation. HCP resulted in the downregulation of genes involved in NF-κB-, STAT3-, and SASP-activation and inflammation.

Predicting pedestrian-involved crash severity using inception-v3 deep learning model

This research leverages a novel deep learning model, Inception-v3, to predict pedestrian crash severity using data collected over five years (2016–2021) from Louisiana. The final dataset incorporates forty different variables related to pedestrian attributes, environmental conditions, and vehicular specifics. Crash severity was classified into three categories: fatal, injury, and no injury. The Boruta algorithm was applied to determine the importance of variables and investigate contributing factors to pedestrian crash severity, revealing several associated aspects, including pedestrian gender, pedestrian and driver impairment, posted speed limits, alcohol involvement, pedestrian age, visibility obstruction, roadway lighting conditions, and both pedestrian and driver conditions, including distraction and inattentiveness. To address data imbalance, the study employed Random Under Sampling (RUS) and the Synthetic Minority Oversampling Technique (SMOTE). The DeepInsight technique transformed numeric data into images. Subsequently, five crash severity prediction models were developed with Inception-v3, considering various scenarios, including original, under-sampled, over-sampled, a combination of under and over-sampled data, and the top twenty-five important variables. Results indicated that the model applying both over and under sampling outperforms models based on other data balancing techniques in terms of several performance metrics, including accuracy, sensitivity, precision, specificity, false negative ratio (FNR), false positive ratio (FPR), and F1-score. This model achieved prediction accuracies of 93.5%, 77.5%, and 85.9% for fatal, injury, and no injury categories, respectively. Additionally, comparative analysis based on several performance metrics and McNemar’s tests demonstrated that the predictive performance of the Inception-v3 deep learning model is statistically superior compared to traditional machine learning and statistical models. The insights from this research can be effectively harnessed by safety professionals, emergency service providers, traffic management centers, and vehicle manufacturers to enhance their safety measures and applications.

Application of the Boruta algorithm to assess the multidimensional determinants of malnutrition among children under five years living in southern Punjab, Pakistan

BackgroundMalnutrition causes nutrient deficiencies that have both physical and clinical consequences in severe acute malnutrition children. Globally, there were 47 million wasted children under the age of five in 2019. One in four were located in sub-Saharan Africa, with half being in South Asia. This study aims to apply the Boruta algorithm to identify the determinants of undernutrition among children under five living in Dera Ghazi Khan, one of the marginalized districts of densely populated Punjab Province in Pakistan.MethodsA multicenter cross-sectional study design was used to collect data from 185 children with severe acute malnutrition aged under five years visiting the OTPs centers located in Dera Ghazi Khan, Punjab, Pakistan. A purposive sampling technique was used to collect data using a pretested structured questionnaire from parents/caregivers regarding family sociodemographic characteristics, child nutrition, and biological and healthcare characteristics. Anthropometric measurements, including height, weight, and mid-upper arm circumference, were collected. The Boruta models were used to incorporate the children’s anthropometric, nutritional, and household factors to determine the important predictive variables for undernutrition using the Boruta package in R studio.ResultsThis study included 185 children, with a mean age of 15.36 ± 10.23 months and an MUAC of 10.19 ± 0.96 cm. The Boruta analysis identifies age, mid-upper arm circumference, weaning practices, and immunization status as important predictors of undernutrition. Income per month, exclusive breastfeeding, and immunization status were found to be key factors of undernutrition in children under the age of five.ConclusionThis study highlights age, mid-upper arm circumference, weaning practices, and immunization status as key determinants of weight-for-height and weight-for-age in children under five years. It also suggests that economic context may influence undernutrition. The findings can guide targeted strategies for combating undernutrition.

MVMSGAT: Integrating Multiview, Multi-Scale Graph Convolutional Networks with Biological Prior Knowledge for Predicting Bladder Cancer Response to Neoadjuvant Therapy

The incidence of bladder cancer is on the rise, and its molecular heterogeneity presents significant challenges for personalized cancer therapy. Transcriptome data can characterize the variability among patients. Traditional machine-learning methods often struggle with high-dimensional genomic data, falling into the ’curse of dimensionality’. To address this challenge, we have developed MVMSGAT, an innovative predictive model tailored for forecasting responses to neoadjuvant therapy in bladder cancer patients. MVMSGAT significantly enhances model performance by incorporating multi-perspective biological prior knowledge. It initially utilizes the Boruta algorithm to select key genes from transcriptome data, subsequently constructing a comprehensive graph of gene co-expression and protein–protein interactions. MVMSGAT further employs a graph convolutional neural network to integrate this information within a multiview knowledge graph, amalgamating biological knowledge maps from various scales using an attention mechanism. For validation, MVMSGAT was tested using a five-fold cross-validation approach on two specific GEO datasets, GSE169455 and GSE69795, involving a total of 210 bladder cancer samples. MVMSGAT demonstrated superior performance, with the following metrics (mean ± standard deviation): AUC-ROC of 0.8724±0.0511, accuracy of 0.7789±0.068, F1 score of 0.8529±0.0338, and recall of 0.9231±0.0719. These results underscore the potential of MVMSGAT in advancing personalized treatment and precision medicine in bladder cancer.

Improved soil carbon stock spatial prediction in a Mediterranean soil erosion site through robust machine learning techniques.

Soil serves as a reservoir for organic carbon stock, which indicates soil quality and fertility within the terrestrial ecosystem. Therefore, it is crucial to comprehend the spatial distribution of soil organic carbon stock (SOCS) and the factors influencing it to achieve sustainable practices and ensure soil health. Thus, the present study aimed to apply four machine learning (ML) models, namely, random forest (RF), k-nearest neighbors (kNN), support vector machine (SVM), and Cubist model tree (Cubist), to improve the prediction of SOCS in the Srou catchment located in the Upper Oum Er-Rbia watershed, Morocco. From an inventory of 120 sample points, 80% were used for training the model, with the remaining 20% set aside for model testing. Boruta's algorithm and the multicollinearity test identified only nine (9) factors as the controlling factors selected as input data for predicting SOCS. As a result, spatial distribution maps for SOCS were generated for all models, then compared, and further validated using statistical metrics. Among the models tested, the RF model exhibited the best performance (R2 = 0.76, RMSE = 0.52 Mg C/ha, NRMSE = 0.13, and MAE = 0.34 Mg C/ha), followed closely by the SVM model (R2 = 0.68, RMSE = 0.59 Mg C/ha, NRMSE = 0.15, and MAE = 0.34 Mg C/ha) and Cubist model (R2 = 0.64, RMSE = 0.63 Mg C/ha, NRMSE = 0.16, and MAE = 0.43 Mg C/ha), while the kNN model had the lowest performance (R2 = 0.31, RMSE = 0.94 Mg C/ha, NRMSE = 0.24, and MAE = 0.63 Mg C/ha). However, bulk density, pH, electrical conductivity, and calcium carbonate were the most important factors for spatially predicting SOCS in this semi-arid region. Hence, the methodology used in this study, which relies on ML algorithms, holds the potential for modeling and mapping SOCS and soil properties in comparable contexts elsewhere.

Exploring evapotranspiration stress in China: A blending approach employing multi-source remote sensing proxies

Study regionChina Study focusAn effective evapotranspiration stress assessment system requires reliable estimates actual evapotranspiration (ETa), and potential evapotranspiration (ETp), and should consider multiple influencing indicators. This study explores the consistency of multiple evaporation reanalysis datasets across mainland China and fuses them with multi-model ensemble method. To address the uncertainty inherent in the traditional evapotranspiration stress index (ESI) that only considers the ETa and ETp relationship, we employ Pearson Correlation Analysis and a Random Forest-based Boruta Algorithm to propose new evapotranspiration stress proxies blending multiple remote sensing indicators, named as MESI-P and MESI-B, respectively. Then, we verified their performance in depicting vegetation production capacity based on Solar-Induced Chlorophyll Fluorescence (SIF) and Gross Primary Productivity (GPP). New hydrological insights for the regionThe patterns of ETa from GLDAS, GLEAM, and Harvard Dataverse in China show consistent, but their threshold values differ. The synthesized ETa obtained after using a multi-model ensemble (MME) method is more adaptable in China, with a range of 0.45–1485.31 mm and an increase of 11.34 mm/10a. Following the same pattern as GLDAS and CRU, the synthetic ETp experiences an increasing trend (11.24 mm/10a) during 2000–2019. Compared with the traditional ESI, the MESI-P and MESI-B proposed in this study have superior signal convergence with ETa, SIF, and GPP due to the higher correlation coefficients (＞0.95) in China. Furthermore, during the calculation of MESI-B, soil moisture and solar radiation are identified as dominate factor in Northwest and Southern China respectively.

Integrated analysis of single-cell RNA-seq and chipset data unravels PANoptosis-related genes in sepsis.

The poor prognosis of sepsis warrants the investigation of biomarkers for predicting the outcome. Several studies have indicated that PANoptosis exerts a critical role in tumor initiation and development. Nevertheless, the role of PANoptosis in sepsis has not been fully elucidated. We obtained Sepsis samples and scRNA-seq data from the GEO database. PANoptosis-related genes were subjected to consensus clustering and functional enrichment analysis, followed by identification of differentially expressed genes and calculation of the PANoptosis score. A PANoptosis-based prognostic model was developed. In vitro experiments were performed to verify distinct PANoptosis-related genes. An external scRNA-seq dataset was used to verify cellular localization. Unsupervised clustering analysis using 16 PANoptosis-related genes identified three subtypes of sepsis. Kaplan-Meier analysis showed significant differences in patient survival among the subtypes, with different immune infiltration levels. Differential analysis of the subtypes identified 48 DEGs. Boruta algorithm PCA analysis identified 16 DEGs as PANoptosis-related signature genes. We developed PANscore based on these signature genes, which can distinguish different PANoptosis and clinical characteristics and may serve as a potential biomarker. Single-cell sequencing analysis identified six cell types, with high PANscore clustering relatively in B cells, and low PANscore in CD16+ and CD14+ monocytes and Megakaryocyte progenitors. ZBP1, XAF1, IFI44L, SOCS1, and PARP14 were relatively higher in cells with high PANscore. We developed a machine learning based Boruta algorithm for profiling PANoptosis related subgroups with in predicting survival and clinical features in the sepsis.

Prediction Model for in-Stent Restenosis Post-PCI Based on Boruta Algorithm and Deep Learning: The Role of Blood Cholesterol and Lymphocyte Ratio.

Percutaneous coronary intervention (PCI) is the primary treatment for acute myocardial infarction (AMI). However, in-stent restenosis (ISR) remains a significant limitation to the efficacy of PCI. The cholesterol-to-lymphocyte ratio (CLR), a novel biomarker associated with inflammation and dyslipidemia, may have predictive value for ISR. Deep learning-based models, such as the multilayer perceptron (MLP), can aid in establishing predictive models for ISR using CLR. A retrospective analysis was conducted on clinical and laboratory data from 1967 patients. The Boruta algorithm was employed to identify key features associated with ISR. An MLP model was developed and divided into training and validation sets. Model performance was evaluated using ROC curves and calibration plots. Patients in the ISR group exhibited significantly higher levels of CLR and low-density lipoprotein (LDL) compared to the non-ISR group. The Boruta algorithm identified 21 important features for subsequent modeling. The MLP model achieved an AUC of 0.95 on the validation set and 0.63 on the test set, indicating good predictive performance. Calibration plots demonstrated good agreement between predicted and observed outcomes. Feature importance analysis revealed that the number of initial stent implants, hemoglobin levels, Gensini score, CLR, and white blood cell count were significant predictors of ISR. Partial dependence plots (PDP) confirmed CLR as a key predictor for ISR. The CLR, as a biomarker that integrates lipid metabolism and inflammation, shows significant potential in predicting coronary ISR. The MLP model, based on deep learning, demonstrated robust predictive capabilities, offering new insights and strategies for clinical decision-making.

Improving crop rotation classification using a random forest model incorporating spatial heterogeneity

Accurate and timely classification of crop rotations is essential to confront the issues of agricultural management and food crisis. Crop growth conditions generally exhibit a strong spatial heterogeneity pattern, resulting in crop growth characteristics varying with locations, limiting the classification accuracy of crop rotation. To overcome this limitation, an improved method named random forest based on rotation zoning strategy (RF_RZS) that classifies crop rotations under the consideration of spatial heterogeneity is proposed. In RF_RZS, the regionalization with dynamically constrained agglomerative clustering and partitioning method is used to adaptively mine the spatial homogeneous subzones of soil organic carbon (SOC), which represent the spatial heterogeneity pattern of crop rotation given the highly correlated relationship between crop rotation and SOC. The Boruta algorithm is employed to select the optimal feature subset within each subzone. A random forest classification method is applied to categorize crop rotations within the subzones. Two integrated indexes, OA_OZ and Kc_OZ, are proposed to evaluate the comprehensive performance of RF_RZS. Furthermore, a series of traditional methods is employed for comparison and evaluation. Results demonstrate that the OA_OZ and Kc_OZ of the RF_RZS method are 0.90 and 0.88, respectively, which are increased by 3–23% and 3–27%, respectively, compared with those of traditional classification methods. This research involves mixed cropping pattern of multiple rotational crops, and the proposed methodology can provide an effective guide for the classification of rotational crops with a fragmented distribution and a complex cropping structure.

Machine Learning-Based Prediction of Pulmonary Embolism Prognosis Using Nutritional and Inflammatory Indices

Purpose This study aimed to create and assess machine learning (ML) models that utilize nutritional and inflammatory indices, focusing on the advanced lung cancer inflammation index (ALI) and neutrophil-to-albumin ratio (NAR), to improve the prediction accuracy of PE prognosis. Patients and methods We conducted a retrospective analysis of 312 patients, comprising 254 survivors and 58 non-survivors. The Boruta algorithm was used to identify significant variables, and four ML models (XGBoost, Random Forest, Logistic Regression, and SVM) were employed to analyze the clinical data and assess the performance of the models. Results The XGBoost model, with optimal hyperparameters, achieved the best performance, with an accuracy of 0.882, an F1-score of 0.889, a precision of 0.917, a sensitivity of 0.863, a specificity of 0.905, and an AUC of 0.873. Analysis of feature importance indicated that the most critical predictors across models were respiratory failure, log-transformed ALI, albumin level, age, diastolic blood pressure, and NAR. Conclusion The ML-based prediction models effectively predicted the prognosis of PE, with the XGBoost model exhibiting good performance. Respiratory failure, ALI, albumin level, age, diastolic blood pressure, and NAR were correlated with PE prognosis.

Construction of a risk prediction model for lung infection after chemotherapy in lung cancer patients based on the machine learning algorithm.

The objective of this study was to create and validate a machine learning (ML)-based model for predicting the likelihood of lung infections following chemotherapy in patients with lung cancer. A retrospective study was conducted on a cohort of 502 lung cancer patients undergoing chemotherapy. Data on age, Body Mass Index (BMI), underlying disease, chemotherapy cycle, number of hospitalizations, and various blood test results were collected from medical records. We used the Synthetic Minority Oversampling Technique (SMOTE) to handle unbalanced data. Feature screening was performed using the Boruta algorithm and The Least Absolute Shrinkage and Selection Operator (LASSO). Subsequently, six ML algorithms, namely Logistic Regression (LR), Random Forest (RF), Gaussian Naive Bayes (GNB), Multi-layer Perceptron (MLP), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN) were employed to train and develop an ML model using a 10-fold cross-validation methodology. The model's performance was evaluated through various metrics, including the area under the receiver operating characteristic curve (ROC), accuracy, sensitivity, specificity, F1 score, calibration curve, decision curves, clinical impact curve, and confusion matrix. In addition, model interpretation was performed by the Shapley Additive Explanations (SHAP) analysis to clarify the importance of each feature of the model and its decision basis. Finally, we constructed nomograms to make the predictive model results more readable. The integration of Boruta and LASSO methodologies identified Gender, Smoke, Drink, Chemotherapy cycles, pleural effusion (PE), Neutrophil-lymphocyte count ratio (NLR), Neutrophil-monocyte count ratio (NMR), Lymphocytes (LYM) and Neutrophil (NEUT) as significant predictors. The LR model demonstrated superior performance compared to alternative ML algorithms, achieving an accuracy of 81.80%, a sensitivity of 81.1%, a specificity of 82.5%, an F1 score of 81.6%, and an AUC of 0.888(95%CI(0.863-0.911)). Furthermore, the SHAP method identified Chemotherapy cycles and Smoke as the primary decision factors influencing the ML model's predictions. Finally, this study successfully constructed interactive nomograms and dynamic nomograms. The ML algorithm, combining demographic and clinical factors, accurately predicted post-chemotherapy lung infections in cancer patients. The LR model performed well, potentially improving early detection and treatment in clinical practice.

Prediction of the Efficacy of Radiotherapy in Head-and-neck Tumors Patients by Dosiomics and Radiomics

Background: The purpose is to investigate the efficacy of distant metastasis (DM) and locoregional recurrences (LRs) prediction through radiomics and dosimics in head-and-neck cancer (HNC) cases. Methods: The cases of HNC were obtained from The Cancer Imaging Archive. For the radiomics model, the features were extracted from the pretreatment computed tomography image by the pyradiomics, then the condensed features were selected as the clinically related features by the Boruta method. Finally, the random forest was used to construct the DM and LR prediction model by inputting those condensed features. For the dosiomics model, the features were extracted from the 3-dimensional dose distribution from radiation treatment plans. The radiomics and dosiomics condensed features were utilized to training and validate the prediction model’s performance. The area under the curve (AUC) value and receiver operating characteristic (ROC) curve were used to assess and compare these models. Results: Seven related features were extracted by the Boruta algorithm, which included one radiomics and two dosiomics features for the DM; two radiomics and two dosiomics features for the LR. Independent training and validation of the prediction and prognostic performance of the model have been observed. The roc-AUC values of the training model for the Rmodel and Dmodel were 0.793 and 0.797, 0.657 and 0.650 for the DM and LR; the roc-AUC values of the validation model for the Rmodel and Dmodel were 0.733 and 0.767, 0.646 and 0.6 for the DM and LR. The roc-AUC values of the training and validation for the radiomics and dosiomics integration model were 0.772 and 0.7, 0.792 and 0.762 for the DM and LR, respectively. Conclusion: Integration of radiomics and dosiomics prediction model can benefit LR in the radiotherapy patient of the head-and-neck squamous cell carcinoma, so the dosiomics should not be neglected for the related investigations.

Classification of Particulate Matter (PM2.5) Concentrations Using Feature Selection and Machine Learning Strategies

Abstract This research employed machine learning approaches to classify acceptable or non-acceptable particulate matter (PM2.5) concentrations using a dataset that was obtained from the Nairobi expressway road corridor. The dataset contained air quality data, traffic volume, and meteorological data. The Boruta Algorithm (BA) was utilized in conjunction with the Random Forests (RF) classifier to select the most appropriate features from the dataset. The findings of the BA analysis indicated that humidity was the most influential factor in determining air quality. This was closely followed by the variables of ‘day_of_week’ and the volume of traffic bound for Nairobi. The temperature of the site was determined to have a lower significance. The comparison among different machine learning classifiers for the classification of acceptable and unacceptable PM2.5 concentrations revealed that the Extreme Gradient Boosting (XGBoost) classifier displayed superior performance in terms of Sensitivity (0.774), Specificity (0.943), F1-Score (0.833), and AU-ROC (0.874). The Binary Logistic Regression (BLR) model demonstrated comparatively poorer performance in terms of Sensitivity (0.244), Specificity (0.614), F1-Score (0.455), and AU-ROC (0.508) when compared to other ML models. The prediction of PM2.5 has the potential to provide valuable insights to transport policymakers in their deliberations on urban transport policy formulation.

Artificial intelligence-based prediction models for acute myeloid leukemia using real-life data: A DATAML registry study

We designed artificial intelligence-based prediction models (AIPM) using 52 diagnostic variables from 3687 patients included in the DATAML registry treated with intensive chemotherapy (IC, N = 3030) or azacitidine (AZA, N = 657) for an acute myeloid leukemia (AML). A neural network called multilayer perceptron (MLP) achieved a prediction accuracy for overall survival (OS) of 68.5% and 62.1% in the IC and AZA cohorts, respectively. The Boruta algorithm could select the most important variables for prediction without decreasing accuracy. Thirteen features were retained with this algorithm in the IC cohort: age, cytogenetic risk, white blood cells count, LDH, platelet count, albumin, MPO expression, mean corpuscular volume, CD117 expression, NPM1 mutation, AML status (de novo or secondary), multilineage dysplasia and ASXL1 mutation; and 7 variables in the AZA cohort: blood blasts, serum ferritin, CD56, LDH, hemoglobin, CD13 and disseminated intravascular coagulation (DIC). We believe that AIPM could help hematologists to deal with the huge amount of data available at diagnosis, enabling them to have an OS estimation and guide their treatment choice. Our registry-based AIPM could offer a large real-life dataset with original and exhaustive features and select a low number of diagnostic features with an equivalent accuracy of prediction, more appropriate to routine practice.

Establishment and validation of a prognostic nomogram for severe fever with thrombocytopenia syndrome: A retrospective observational study.

Several scoring systems have been proposed to predict the risk of death due to severe fever with thrombocytopenia syndrome (STFS), but they have limitations. We developed a new prognostic nomogram for STFS-related death and compared its performance with previous scoring systems and the Acute Physiology and Chronic Health Evaluation score (APACHE II Score). A total of 292 STFS patients were retrospectively enrolled between January 2016 and March 2023. Boruta's algorithm and backward stepwise regression were used to select variables for constructing the nomogram. Time-dependent receiver operating characteristic (ROC) curves and clinical decision curves were generated to compare the strengths of the nomogram with others. Age, Sequential Organ Failure Assessment Score (SOFA score), state of consciousness, continuous renal replacement therapy (CRRT), and D-dimer were significantly correlated with mortality in both univariate and multivariate analyses (P<0.05). We developed a nomogram using these variables to predict mortality risk, which outperformed the SFTS and APACHE II scores (Training ROC: 0.929 vs. 0.848 vs. 0.792; Validation ROC: 0.938 vs. 0.839 vs. 0.851; P<0.001). In the validation set, the SFTS model achieved an accuracy of 76.14%, a sensitivity of 95.31%, a specificity of 25.00%, a precision of 77.22%, and an F1 score of 85.32%. The nomogram showed a superior performance with an accuracy of 86.36%, a precision of 88.24%, a recall of 93.75%, and an F1 score of 90.91%. Age, consciousness, SOFA Score, CRRT, and D-Dimer are independent risk factors for STFS-related death. The nomogram based on these factors has an excellent performance in predicting STFS-related death and is recommended for clinical practice.