XGBoost Machine Learning Research Articles

Development of an Artificial Intelligence-Based Machine Tool to Predict Outcome in Intracerebral Hemorrhage

Introduction: Spontaneous intracerebral hemorrhage is the second most common type of stroke with high morbidity and mortality. Outcome prediction is very important in this disease, to enable us tailor treatment strategies especially in a low- and middle-income countries. Today, prediction is predominantly limited to few clinical factors and may not be very accurate. We explore the application of an artificial intelligence-based platform for outcome prediction with a combination of clinical, radiological, and biochemical parameters. Methods: Data from our prospectively maintained stroke register was cleaned and processed using the XGBoost machine learning (ML) algorithm to predict outcome at discharge and 90 days using the modified Rankin scale. A total of 1,000 patients were included in the study, 129 variables were pruned to 19 significant features during the phase of preprocessing. Results: The data set was split 9:1 with 900 cases being used for training and the remaining 100 for validation. The models were evaluated based on the mean absolute error (MAE). Model-1 trained for predicting “mRS_Discharge” had a MAE of 0.34 and model-2 trained for predicting “mRS_3months” had a MAE of 0.63. Conclusion: ML algorithms can be successfully applied for the prediction of outcome in intracerebral hemorrhage.

Integration of ultrasound radiomics features and clinical factors: A nomogram model for identifying the Ki-67 status in patients with breast carcinoma.

ObjectiveThe aim of this study was to develop and validate an ultrasound-based radiomics nomogram model by integrating the clinical risk factors and radiomics score (Rad-Score) to predict the Ki-67 status in patients with breast carcinoma.MethodsUltrasound images of 284 patients (196 high Ki-67 expression and 88 low Ki-67 expression) were retrospectively analyzed, of which 198 patients belonged to the training set and 86 patients to the test set. The region of interest of tumor was delineated, and the radiomics features were extracted. Radiomics features underwent dimensionality reduction analysis by using the independent sample t test and least absolute shrinkage and selection operator (LASSO) algorithm. The support vector machine (SVM), logistic regression (LR), decision tree (DT), random forest (RF), naive Bayes (NB) and XGBoost (XGB) machine learning classifiers were trained to establish prediction model based on the selected features. The classifier with the highest AUC value was selected to convert the output of the results into the Rad-Score and was regarded as Rad-Score model. In addition, the logistic regression method was used to integrate Rad-Score and clinical risk factors to generate the nomogram model. The leave group out cross-validation (LGOCV) method was performed 200 times to verify the reliability and stability of the nomogram model.ResultsSix classifier models were established based on the 15 non-zero coefficient features. Among them, the LR classifier achieved the best performance in the test set, with the area under the receiver operating characteristic curve (AUC) value of 0.786, and was obtained as the Rad-Score model, while the XGB performed the worst (AUC, 0.615). In multivariate analysis, independent risk factor for high Ki-67 status was age (odds ratio [OR] = 0.97, p = 0.04). The nomogram model based on the age and Rad-Score had a slightly higher AUC than that of Rad-Score model (AUC, 0.808 vs. 0.798) in the test set, but no statistical difference (p = 0.144, DeLong test). The LGOCV yielded a median AUC of 0.793 in the test set.ConclusionsThis study proposed a convenient, clinically useful ultrasound radiomics nomogram model that can be used for the preoperative individualized prediction of the Ki-67 status in patients with BC.

Comparison of Classification Success Rates of Different Machine Learning Algorithms in the Diagnosis of Breast Cancer.

To identify which Machine Learning (ML) algorithms are the most successful in predicting and diagnosing breast cancer according to accuracy rates. The "College of Wisconsin Breast Cancer Dataset", which consists of 569 data and 30 features, was classified using Support Vector Machine (SVM), Naive Bayes (NB), Random Forest (RF), Decision Tree (DT), K-Nearest Neighbor (KNN), Logistic Regression (LR), Multilayer Perceptron (MLP), Linear Discriminant Analysis (LDA), XgBoost (XGB), Ada-Boost (ABC) and Gradient Boosting (GBC) ML algorithms. Before the classification process, the dataset was preprocessed. Sensitivity, accuracy, and definiteness metrics were used to measure the success of the methods. Compared to other ML algorithms used in the study, the GBC ML algorithm was found to be the most successful method in the classification of tumors with an accuracy of 99.12%. The XGB ML algorithm was found to be the lowest method with an accuracy rate of 88.10%. In addition, it was determined that the general accuracy rates of the 11 ML algorithms used in the study varied between 88-95%. When the results obtained from the ML classifiers used in the study are evaluated, the efficiency of the GBC algorithm in the classification of tumors is obvious. It can be said that the success rates obtained from 11 different ML algorithms used in the study are valuable in terms of being used to predict different cancer types.

Investment estimation of prefabricated concrete buildings based on XGBoost machine learning algorithm

The prefabricated concrete buildings (PCBs）are the booster in the process of construction industrialization and intelligent upgrading. However, its high cost has become one of the restricting factors of further application and promotion of prefabricated concrete buildings. Moreover, the existing investment estimation methods of prefabricated concrete buildings have limited predicting accuracy as well as the ability of adapting dynamic factors. Therefore, to achieve more reliable and reasonable investment estimation of prefabricated concrete buildings, this paper has proposed an investment estimation model of prefabricated concrete buildings based on XGBoost machine learning algorithm. In the proposed model, the construction project cost-significance theory (CS) and analytic hierarchy process (AHP) were used to extract the construction characteristic indices of prefabricated concrete buildings investment estimation. Then the XGBoost machine learning algorithm was implemented to build an investment estimation model of prefabricated concrete buildings that was able to quantify the uncertainty of the confidence and prediction, and to enhance the interpretability of the model. The research conducted in this paper showed that when compared with traditional machine learning methods such as Support vector machine (SVM), Back Propagation Neural Network (BPNN) and Random Forest (RF), XGBoost had better generalization and interpretable ability. The discussion provided in this paper further demonstrated the reliability and feasibility of the proposed model, and provided reliable basis for the investment decision-making of prefabricated concrete building projects.

CROP YIELD PREDICTION USING DEEP XGBOOST ALGORITHM

Predicting crop yield is a complex task since it depends on multiple factors. Although many models have been developed so far in the literature, the performance of current models is not satisfactory, and hence, they must be improved. In this study, we developed deep learning-based models to evaluate how the underlying algorithms perform with respect to different performance criteria. The algorithms evaluated in our study are the XGBoost machine learning (ML) algorithm, Convolutional Neural Networks (CNN),XGBoost, and Recurrent Neural Networks (RNN). For the case study, we predicted crop yield based on the environmental, soil, silt, nitrogen, clay, ocd, ocs, pHH2O, sand, soc, ceo, water and crop parameters has been a potential research topic. Our proposed method has high performance by preforming feature selection on predicting the crop yield. Our proposed method used for both the soyabeans and corn crop for predicting their yield.

Association between vitamin D3 levels and frailty in the elderly: A large sample cross-sectional study.

BackgroundFrailty is recognized as a cornerstone of geriatric medicine. Accurately screening and identifying frailty can promote better quality and personalized medical services for the elderly. Previous studies have shown that the association between vitamin D and frailty in the elderly population is still controversial. More research is needed to explore the association between them.Materials and methodsWe used three waves of data from the National Health and Nutrition Examination Survey (NHANES). Based on the widely accepted AAH FRAIL Scale, we measured and evaluated the participants’ frailty from five aspects: fatigue, resistance, ambulation, illness, and loss of weight. All possible relevant variables are included. Machine learning XGboost algorithm, the Least Absolute Shrinkage Selection Operator (LASSO) regression and univariate logistic regression were used to screen variables, and multivariate logistic regression and generalized additive model (GAM) were used to build the model. Finally, subgroup analysis and interaction test were performed to further confirm the association.ResultsIn our study, XGboost machine learning algorithm explored the relative importance of all included variables, which confirmed the close association between vitamin D and frailty. After adjusting for all significant covariates, the result indicated that for each additional unit of 25-hydroxyvitamin D3, the risk of frailty was reduced by 1.3% with a statisticaldifference. A smooth curve was constructed based on the GAM. It was found that there was a significant negative correlation between 25-hydroxyvitamin D3 and the risk of frailty.ConclusionThere may be a negative correlation between 25-hydroxyvitamin D3 and the risk of frailty. However, more well-designed studies are needed to verify this relationship.

Machine learning models including insulin resistance indexes for predicting liver stiffness in United States population: Data from NHANES.

BackgroundPrevention and treatment of liver fibrosis at an early stage is of great prognostic importance, whereas changes in liver stiffness are often overlooked in patients before the onset of obvious clinical symptoms. Recognition of liver fibrosis at an early stage is therefore essential.ObjectiveAn XGBoost machine learning model was constructed to predict participants' liver stiffness measures (LSM) from general characteristic information, blood test metrics and insulin resistance-related indexes, and to compare the fit efficacy of different datasets for LSM.MethodsAll data were obtained from the National Health and Nutrition Examination Survey (NHANES) for the time interval January 2017 to March 2020. Participants' general characteristics, Liver Ultrasound Transient Elastography (LUTE) information, indicators of blood tests and insulin resistance-related indexes were collected, including homeostasis model assessment of insulin resistance (HOMA-IR) and metabolic score for insulin resistance (METS-IR). Three datasets were generated based on the above information, respectively named dataset A (without the insulin resistance-related indexes as predictor variables), dataset B (with METS-IR as a predictor variable) and dataset C (with HOMA-IR as a predictor variable). XGBoost regression was used in the three datasets to construct machine learning models to predict LSM in participants. A random split was used to divide all participants included in the study into training and validation cohorts in a 3:1 ratio, and models were developed in the training cohort and validated with the validation cohort.ResultsA total of 3,564 participants were included in this study, 2,376 in the training cohort and 1,188 in the validation cohort, and all information was not statistically significantly different between the two cohorts (p > 0.05). In the training cohort, datasets A and B both had better predictive efficacy than dataset C for participants' LSM, with dataset B having the best fitting efficacy [±1.96 standard error (SD), (-1.49,1.48) kPa], which was similarly validated in the validation cohort [±1.96 SD, (-1.56,1.56) kPa].ConclusionsXGBoost machine learning models built from general characteristic information and clinically accessible blood test indicators are practicable for predicting LSM in participants, and a dataset that included METS-IR as a predictor variable would improve the accuracy and stability of the models.

Characteristics associated with the perception of high-impact disease (PsAID ≥4) in patients with recent-onset psoriatic arthritis. Machine learning-based model

ObjectivesTo evaluate which patient and disease characteristics are associated with the perception of high-impact disease (PsAID ≥4) in recent-onset psoriatic arthritis. MethodsWe performed a multicenter observational prospective study (2-year follow-up, regular annual visits). The study population comprised patients aged ≥18 years who fulfilled the CASPAR criteria and less than 2 years since the onset of symptoms. The dataset was generated using data for each patient at the 3 visits (baseline, first year, and second year of follow-up) matched with the PsAID values at each of the 3 visits. PsAID was categorized into two groups (<4 and ≥4). We trained a logistic regression model and random forest–type and XGBoost machine learning algorithms to analyze the association between the outcome measure and the variables selected in the bivariate analysis. A k-fold cross-validation with k = 5 was performed. ResultsThe sample comprised 158 patients. Of the patients who attended the clinic, 45.8% scored PsAID ≥4 at baseline; 27.1%, at the first follow-up visit, and in 23.0%, at the second follow-up visit. The variables associated with PsAID ≥4 were, in decreasing order of importance: HAQ, pain, educational level, and physical activity. Higher HAQ (logistic regression coefficient 10.394; IC95% 7.777,13.011), higher pain (5.668; 4.016, 7.320), lower educational level (-2.064; -3.515, -0.613) and high level of physical activity (1.221; 0.158, 2.283) were associated with a higher frequency of PsAID ≥4. The mean values of the measures of validity of the algorithms were all ≥85%. ConclusionsDespite the higher weight given to pain when scoring PsAID, we observed a greater influence of physical function on disease impact.

Optimal PV system capacity ratio and power limit value selection based on a novel fast calculation method of IGBT junction temperature and IGBT annual damage analysis

In order to solve the problem of long calculation time of insulated gate bipolar transistor (IGBT) junction temperature, the XGBoost machine learning algorithm is used to calculate IGBT junction temperature in the annual damage assessment process. The XGBoost machine learning algorithm can greatly reduce the calculation time of IGBT junction temperature while ensuring the accuracy, which provides conditions for finding the optimal PV system capacity ratio and power limit value by heuristic algorithm such as differential evolution algorithm later. For PV system capacity ratio and power limit, it is necessary to consider the annual damage of the PV inverter, the increase of power generation due to capacity ratio and the power generation loss due to power limit. This paper proposes an optimization goal that considers the above factors, and uses the differential evolution algorithm to obtain the optimal PV system capacity ratio and power limit value.

Memory Forensics-Based Malware Detection Using Computer Vision and Machine Learning

Malware has recently grown exponentially in recent years and poses a serious threat to individual users, corporations, banks, and government agencies. This can be seen from the growth of Advanced Persistent Threats (APTs) that make use of advance and sophisticated malware. With the wide availability of computer-automated tools such as constructors, email flooders, and spoofers. Thus, it is now easy for users who are not technically inclined to create variations in existing malware. Researchers have developed various defense techniques in response to these threats, such as static and dynamic malware analyses. These techniques are ineffective at detecting new malware in the main memory of the computer and otherwise require considerable effort and domain-specific expertise. Moreover, recent techniques of malware detection require a long time for training and occupy a large amount of memory due to their reliance on multiple factors. In this paper, we propose a computer vision-based technique for detecting malware that resides in the main computer memory in which our technique is faster or memory efficient. It works by taking portable executables in a virtual environment to extract memory dump files from the volatile memory and transform them into a particular image format. The computer vision-based contrast-limited adaptive histogram equalization and the wavelet transform are used to improve the contrast of neighboring pixel and to reduce the entropy. We then use the support vector machine, random forest, decision tree, and XGBOOST machine learning classifiers to train the model on the transformed images with dimensions of 112 × 112 and 56 × 56. The proposed technique was able to detect and classify malware with an accuracy rate of 97.01%. Its precision, recall, and F1-score were 97.36%, 95.65%, and 96.36%, respectively. Our finding shows that our technique in preparing dataset with more efficient features to be trained by the Machine Learning classifiers has resulted in significant performance in terms of accuracy, precision, recall, F1-score, speed and memory consumption. The performance has superseded most of the existing techniques in its unique approach.

Dynamic Sepsis Prediction for Intensive Care Unit Patients Using XGBoost-Based Model With Novel Time-Dependent Features

Sepsis is a systemic inflammatory response caused by pathogens such as bacteria. Because its pathogenesis is not clear, the clinical manifestations of patients vary greatly, and the alarming incidence and mortality pose a great threat to patients and medical systems, especially in the ICU (Intensive Care Unit). The traditional judgment criteria have the problem of low specificity. Artificial intelligence models could greatly improve the accuracy of sepsis prediction and judgment. Based on the XGBoost machine learning framework taking demographic, vital signs, laboratory tests and medical intervention data as input, this paper proposes a novel model for dynamically predicting sepsis and assessing risk. To realize the model, two methods for feature construction are introduced. For the observed time-series data of vital signs and laboratory tests, the time-dependent method performs to construct the time-dependent characteristics after the statistical screening. For the clinical intervention data, the statistical counting method is applied to construct count-dependent characteristics. Moreover, a new objective function is proposed for the XGBoost framework, and the first-order and second-order gradients of the objective function are also given for model training. Compared with the state-of-the-art methods at present, the proposed model has the best performance, with AUROC improved by 5.4% on the MIMIC-III dataset and 2.1% on PhysioNet Challenge 2019 dataset. The data processing and training methods of this model can be conveniently applied in different electronic health record systems and has a wide application prospect.

Constructing prediction models for excessive daytime sleepiness by nomogram and machine learning: A large Chinese multicenter cohort study.

ObjectiveAlthough risk factors for excessive daytime sleepiness (EDS) have been reported, there are still few cohort-based predictive models for EDS in Parkinson’s disease (PD). This 1-year longitudinal study aimed to develop a predictive model of EDS in patients with PD using a nomogram and machine learning (ML).Materials and methodsA total of 995 patients with PD without EDS were included, and clinical data during the baseline period were recorded, which included basic information as well as motor and non-motor symptoms. One year later, the presence of EDS in this population was re-evaluated. First, the baseline characteristics of patients with PD with or without EDS were analyzed. Furthermore, a Cox proportional risk regression model and XGBoost ML were used to construct a prediction model of EDS in PD.ResultsAt the 1-year follow-up, EDS occurred in 260 of 995 patients with PD (26.13%). Baseline features analysis showed that EDS correlated significantly with age, age of onset (AOO), hypertension, freezing of gait (FOG). In the Cox proportional risk regression model, we included high body mass index (BMI), late AOO, low motor score on the 39-item Parkinson’s Disease Questionnaire (PDQ-39), low orientation score on the Mini-Mental State Examination (MMSE), and absence of FOG. Kaplan–Meier survival curves showed that the survival prognosis of patients with PD in the high-risk group was significantly worse than that in the low-risk group. XGBoost demonstrated that BMI, AOO, PDQ-39 motor score, MMSE orientation score, and FOG contributed to the model to different degrees, in decreasing order of importance, and the overall accuracy of the model was 71.86% after testing.ConclusionIn this study, we showed that risk factors for EDS in patients with PD include high BMI, late AOO, a low motor score of PDQ-39, low orientation score of MMSE, and lack of FOG, and their importance decreased in turn. Our model can predict EDS in PD with relative effectivity and accuracy.

From human business to machine learning—methods for automating real estate appraisals and their practical implications

Until recently, in most countries, the use of Automated Valuation Models (AVMs) in the lending process was only allowed for support purposes, and not as the sole value-determining tool. However, this is currently changing, and regulators around the world are actively discussing the approval of AVMs. But the discussion is generally limited to AVMs that are based on already established methods such as an automation of the traditional sales comparison approach or linear regressions. Modern machine learning approaches are almost completely excluded from the debate. Accordingly, this study contributes to the discussion on why AVMs based on machine learning approaches should also be considered. For this purpose, an automation of the sales comparison method by using filters and similarity functions, two hedonic price functions, namely an OLS model and a GAM model, as well as a XGBoost machine learning approach, are applied to a dataset of 1.2 million residential properties across Germany. We find that the machine learning method XGBoost offers the overall best performance regarding the accuracy of estimations. Practical application shows that optimization of the established methods—OLS and GAM—is time-consuming and labor-intensive, and has significant disadvantages when being implemented on a national scale. In addition, our results show that different types of methods perform best in different regions and, thus, regulators should not only focus on one single method, but consider a multitude of them.

Using Bayesian Optimization and Wavelet Decomposition in GPU for Arterial Blood Pressure Estimation.

Continuous monitoring of arterial blood pressure (ABP) of patients in hospital is currently carried out in an invasive way, which could represent a risk for them. In this paper, a noninvasive methodology to optimize ABP estimators using electrocardiogram and photoplethysmography signals is proposed. For this, the XGBoost machine learning model, optimized with Bayesian techniques, is executed in a Graphics Processing Unit, which drastically reduces execution time. The methodology is evaluated using the MIMIC-III Waveform Database. Systolic and diastolic pressures are estimated with mean absolute error values of 15.85 and 11.59 mmHg, respectively, similar to those of the state of the art. The main advantage of the proposed methodology with respect to others of the current state of the art is that it allows the optimization of the estimator model to be performed automatically and more efficiently at the computational level for the data available. Clinical Relevance- This approach has the advantage of using noninvasive methods to continuously monitor patient's arterial blood pressure, reducing the risk for patients.

MLMD—A Malware-Detecting Antivirus Tool Based on the XGBoost Machine Learning Algorithm

This paper focuses on training machine learning models using the XGBoost and extremely randomized trees algorithms on two datasets obtained using static and dynamic analysis of real malicious and benign samples. We then compare their success rates—both mutually and with other algorithms, such as the random forest, the decision tree, the support vector machine, and the naïve Bayes algorithms, which we compared in our previous work on the same datasets. The best performing classification models, using the XGBoost algorithm, achieved 91.9% detection accuracy and 98.2% sensitivity, 0.853 AUC, and 0.949 F1 score on the static analysis dataset, and 96.4% accuracy and 98.5% sensitivity, 0.940 AUC, and 0.977 F1 score on the dynamic analysis dataset. Then, we exported the best performing machine learning models and used them in our proposed MLMD program, automating the process of static and dynamic analysis and allowing the trained models to be used for classification on new samples.

Development and validation of machine learning models for prediction of seizure outcome after pediatric epilepsy surgery.

There is substantial variability in reported seizure outcome following pediatric epilepsy surgery, and lack of individualized predictive tools that could evaluate the probability of seizure freedom postsurgery. The aim of this study was to develop and validate a supervised machine learning (ML) model for predicting seizure freedom after pediatric epilepsy surgery. This is a multicenter retrospective study of children who underwent epilepsy surgery at five pediatric epilepsy centers in North America. Clinical information, diagnostic investigations, and surgical characteristics were collected, and used as features to predict seizure-free outcome 1 year after surgery. The dataset was split randomly into 80% training and 20% testing data. Thirty-five combinations of five feature sets with seven ML classifiers were assessed on the training cohort using 10-fold cross-validation for model development. The performance of the optimal combination of ML classifier and feature set was evaluated in the testing cohort, and compared with logistic regression, a classical statistical approach. Of the 801 patients included, 61.3% were seizure-free 1 year postsurgery. During model development, the best combination was XGBoost ML algorithm with five features from the univariate feature set, including number of antiseizure medications, magnetic resonance imaging lesion, age at seizure onset, video-electroencephalography concordance, and surgery type, with a mean area under the curve (AUC) of .73 (95% confidence interval [CI] = .69-.77). The combination of XGBoost and univariate feature set was then evaluated on the testing cohort and achieved an AUC of .74 (95% CI = .66-.82; sensitivity=.87, 95% CI = .81-.94; specificity=.58, 95% CI = .47-.71). The XGBoost model outperformed the logistic regression model (AUC=.72, 95% CI = .63-.80; sensitivity=.72, 95% CI = .63-.82; specificity=.66, 95% CI = .53-.77) in the testing cohort (p= .005). This study identified important features and validated an ML algorithm, XGBoost, for predicting the probability of seizure freedom after pediatric epilepsy surgery. Improved prognostication of epilepsy surgery is critical for presurgical counseling and will inform treatment decisions.

Direct Prediction of Physicochemical Properties and Toxicities of Chemicals from Analytical Descriptors by GC-MS.

With advances in machine learning (ML) techniques, the quantitative structure–activity relationship (QSAR) approach is becoming popular for evaluating chemicals. However, the QSAR approach requires that the chemical structure of the target compound is known and that it should be convertible to molecular descriptors. These requirements lead to limitations in predicting the properties and toxicities of chemicals distributed in the environment as in the PubChem database; the structural information on only 14% of compounds is available. This study proposes a new ML-based QSAR approach that can predict the properties and toxicities of compounds using analytical descriptors of mass spectrum and retention index obtained via gas chromatography–mass spectrometry without requiring exact structural information. The model was developed based on the XGBoost ML method. The root-mean-square errors (RMSEs) for log Ko-w, log (molecular weight), melting point, boiling point, log (vapor pressure), log (water solubility), log (LD50) (rat, oral), and log (LD50) (mouse, oral) are 0.97, 0.052, 51, 23, 0.74, 1.1, 0.74, and 0.6, respectively. The model performed well on a chemical standard mixture measurement, with similar results to those of model validation. It also performed well on a measurement of contaminated oil with spectral deconvolution. These results indicate that the model is suitable for investigating unknown-structured chemicals detected in measurements. Any online user can execute the model through a web application named Detective-QSAR (http://www.mixture-platform.net/Detective_QSAR_Med_Open/). The analytical descriptor-based approach is expected to create new opportunities for the evaluation of unknown chemicals around us.

Impact from the evolution of private vehicle fleet composition on traffic related emissions in the small-medium automotive city

Understanding the emission characteristics in the evolution of private vehicle fleet composition has become a key issue to be addressed to develop appropriate emission mitigation strategies in transportation sector. In this study, the influence of such evolution on on-road emissions was investigated based on a comprehensive dataset encompassing vehicle fleet composition, demographic, economic, and energy features from a representative small-medium automotive city in North America. The decoupling analysis was carried out to assess the dynamic linkage between environmental pressure exerted by the transportation sector and economic growth at both city level and national level in North America. We also developed an approach that supports the long-term traffic-related air pollutant prediction and investigated the potential influence on urban air quality. A sharp upward trajectory was observed in the quantity of SUVs from 2001 to 2018, gradually replacing the dominance of the quantity of four-door cars. There was a significant shift in the GHG emissions emitted from vehicle types used for passenger transport: emissions from SUVs and trucks rose by 374.0% and 69.3%, respectively, whereas emissions from four-door cars, two-door cars, station wagons, and vans all decreased. The changes in vehicle composition, along with the steady trend in GHG emissions from private fleet and decrease in on-road air pollutant concentrations found in Regina, were a response to the establishment of federal fuel economy standards and improved fuel economy. Relative decoupling was observed in aggregate for Regina and Canada in most of the years while both experienced economic downturns and increases in environmental pressure in the form of emissions from 2014 to 2015. The predicted results also demonstrate the high capability of XGboost machine learning algorithm in predicting on-road air pollutant concentrations of CO, PM2.5, and NOX.

Machine learning (ML)–enabled, circulating tumor cell–based classification of patients for non-prerequisite adjuvant therapy.

1547 Background: Oncology implicates highest precision using next generation diagnostics and progressive therapies assisted by predictive tools. If validated clinically, machine learning (ML) can provide better insights in precision oncology. Furthermore, it longitudinally may stratify the progression of cancer disease burden in a real time. We have developed, Circulating Tumor Cells (CTCs) driven ML model as a predictor for the treatment decision strategy for both surgery and adjuvant therapy in head and neck squamous cell carcinoma (HNSCC) patients. Methods: In this study, a total of 380 HNSCC patients who underwent either surgery alone or surgery plus adjuvant therapy were accounted for. CTCs in patients were stratified based on clinicopathological parameters and using OncoDiscover platform having anti EpCAM antibody system regulated by the Drug Controller of India. Following this, we explored the predictive performance of the ML model on the usefulness of adjuvant therapy in HNSCC patients after the surgery. The available data was randomly divided into two subsets. First, 75%, of the original data was applied for Training the ML, and rest 25% of the data was used as a Test set. Survival curves were generated by Kaplan–Meier method and calculated through the Log rank test. Results: XGBoost machine learning classifier was superior to Random Forest and SVM-based analyses in predicting the usefulness of adjuvant therapy post-surgery using CTC alone or in combination with other clinical parameters in HNSCC patients. Machine learning algorithms were compared for predicting the accuracy of patients stratification. The results for each model were: XGBoost model (Accuracy = 0.84, ROC value = 0.73, Kappa = 0.43); Random Forest model (Accuracy = 0.81 ROC value = 0.70, Kappa = 0.41); SVM model (Accuracy = 0.76, ROC value = 0. 69, Kappa = 0.40). The ROC value of the XGBoost model was highest (0.73) while the ROC value for the SVM model was lower (0.69). We observed that when CTCs were combined with clinicopathological parameters, the accuracy, kappa values and AUC-ROC drastically improved in predicting the usefulness of adjuvant therapy post-surgery. A similar trend was observed when CTCs were combined with clinicopathological parameters in predicting the line of chemotherapy, post-surgery. Conclusions: ML-enabled, CTCs driven predictions can be highly accurate and ascertain the patient treatments. CTCs can be a positive predictor for selecting patient’s treatment regimen in both surgery as well in type of treatment (e.g. surgery alone or surgery + adjuvant therapy). It can also implicate to classify the patients and determine who necessitates an additional adjuvant therapy. Further investigations in this direction are necessary to predict the treatment options based on ML that may improve the overall survival of cancer patients.

EASY-APP:An artificial intelligence model and application for early and easy prediction of severity in acute pancreatitis.

BackgroundAcute pancreatitis (AP) is a potentially severe or even fatal inflammation of the pancreas. Early identification of patients at high risk for developing a severe course of the disease is crucial for preventing organ failure and death. Most of the former predictive scores require many parameters or at least 24 h to predict the severity; therefore, the early therapeutic window is often missed.MethodsThe early achievable severity index (EASY) is a multicentre, multinational, prospective and observational study (ISRCTN10525246). The predictions were made using machine learning models. We used the scikit‐learn, xgboost and catboost Python packages for modelling. We evaluated our models using fourfold cross‐validation, and the receiver operating characteristic (ROC) curve, the area under the ROC curve (AUC), and accuracy metrics were calculated on the union of the test sets of the cross‐validation. The most critical factors and their contribution to the prediction were identified using a modern tool of explainable artificial intelligence called SHapley Additive exPlanations (SHAP).ResultsThe prediction model was based on an international cohort of 1184 patients and a validation cohort of 3543 patients. The best performing model was an XGBoost classifier with an average AUC score of 0.81 ± 0.033 and an accuracy of 89.1%, and the model improved with experience. The six most influential features were the respiratory rate, body temperature, abdominal muscular reflex, gender, age and glucose level. Using the XGBoost machine learning algorithm for prediction, the SHAP values for the explanation and the bootstrapping method to estimate confidence, we developed a free and easy‐to‐use web application in the Streamlit Python‐based framework (http://easy‐app.org/).ConclusionsThe EASY prediction score is a practical tool for identifying patients at high risk for severe AP within hours of hospital admission. The web application is available for clinicians and contributes to the improvement of the model.