Accuracy Of Machine Learning Models Research Articles

Explainable Machine Learning Model for Predicting First-Time Acute Exacerbation in Patients with Chronic Obstructive Pulmonary Disease.

Background: The study developed accurate explainable machine learning (ML) models for predicting first-time acute exacerbation of chronic obstructive pulmonary disease (COPD, AECOPD) at an individual level. Methods: We conducted a retrospective case–control study. A total of 606 patients with COPD were screened for eligibility using registry data from the COPD Pay-for-Performance Program (COPD P4P program) database at Changhua Christian Hospital between January 2017 and December 2019. Recursive feature elimination technology was used to select the optimal subset of features for predicting the occurrence of AECOPD. We developed four ML models to predict first-time AECOPD, and the highest-performing model was applied. Finally, an explainable approach based on ML and the SHapley Additive exPlanations (SHAP) and a local explanation method were used to evaluate the risk of AECOPD and to generate individual explanations of the model’s decisions. Results: The gradient boosting machine (GBM) and support vector machine (SVM) models exhibited superior discrimination ability (area under curve [AUC] = 0.833 [95% confidence interval (CI) 0.745–0.921] and AUC = 0.836 [95% CI 0.757–0.915], respectively). The decision curve analysis indicated that the GBM model exhibited a higher net benefit in distinguishing patients at high risk for AECOPD when the threshold probability was <0.55. The COPD Assessment Test (CAT) and the symptom of wheezing were the two most important features and exhibited the highest SHAP values, followed by monocyte count and white blood cell (WBC) count, coughing, red blood cell (RBC) count, breathing rate, oral long-acting bronchodilator use, chronic pulmonary disease (CPD), systolic blood pressure (SBP), and others. Higher CAT score; monocyte, WBC, and RBC counts; BMI; diastolic blood pressure (DBP); neutrophil-to-lymphocyte ratio; and eosinophil and lymphocyte counts were associated with AECOPD. The presence of symptoms (wheezing, dyspnea, coughing), chronic disease (CPD, congestive heart failure [CHF], sleep disorders, and pneumonia), and use of COPD medications (triple-therapy long-acting bronchodilators, short-acting bronchodilators, oral long-acting bronchodilators, and antibiotics) were also positively associated with AECOPD. A high breathing rate, heart rate, or systolic blood pressure and methylxanthine use were negatively correlated with AECOPD. Conclusions: The ML model was able to accurately assess the risk of AECOPD. The ML model combined with SHAP and the local explanation method were able to provide interpretable and visual explanations of individualized risk predictions, which may assist clinical physicians in understanding the effects of key features in the model and the model’s decision-making process.

Error Analysis of Machine Learning Predictions on Asthma Hospital Encounters

To guide the allocation of finite preventive care resources for asthma management, we recently developed the world’s most accurate machine learning model to predict which asthmatic patients will have poor outcomes of incurring asthma hospital encounters (AHEs) in the following 12 months, where an AHE is an emergency department visit or an inpatient stay for asthma. In this abstract, we present an error analysis of our model’s predictions.

Quantum Machine Learning Approach for Studying Atmospheric Cluster Formation

Quantum chemical (QC) calculations can yield direct insight into an atmospheric cluster formation mechanism and cluster formation rates. However, such calculations are extremely computationally demanding as more than millions of cluster configurations might exist and need to be computed. We present an efficient approach to produce high quality QC data sets for applications in cluster formation studies and how to train an accurate quantum machine learning model on the generated data. Using the two-component sulfuric acid─water system as a proof of concept, we demonstrate that a kernel ridge regression machine learning model with Δ-learning can be trained to accurately predict the binding energies of cluster equilibrium configurations with mean absolute errors below 0.5 kcal mol–1. Additionally, we enlarge the training data set with nonequilibrium configurations and show the possibility of predicting the binding energies of new structures of clusters several molecules larger than those in the training set. Applying the trained machine learning model leads to a drastic reduction in the number of relevant clusters that need to be explicitly evaluated by QC methods. The presented approach is directly transferable to clusters of arbitrary composition and will lead to faster and more efficient exploration of the configurational space of new cluster systems.

Evaluating the Impact of Scaling Considering the Extrapolation Domain on the Prediction Performance of Machine Learning Algorithms

In this study, we used a benchmark dataset to evaluate the impact of scaling with the extrapolation domain on the prediction performance of machine learning algorithms. We pseudo-divided the data into the interpolation domain (training data) and the extrapolation domain (test data) using a combination of UMAP (Uniform Manifold Approximation and Projection) and material domain knowledge. In anticipation of bridging interpolation and extrapolation domains in nonlinear machine learning algorithms, we evaluated how the scaling considering the extrapolation domain affects prediction performance in the extrapolation domain. For this evaluation, we used three nonlinear algorithms widely used in the MI (Materials Informatics) domain: XGB (XGBoost) regression, GP (Gaussian Process) regression, and SVR (Support Vector Regression). In this study, by defining the pseudo extrapolation domain, we established the approach for evaluating the prediction accuracy of machine learning models in the extrapolation domain, which is considered difficult to evaluate quantitatively. We also demonstrated that this method, which uses scaling that considers the extrapolation domain, is an effective method for improving prediction accuracy in the extrapolation domain while maintaining prediction accuracy in the interpolation domain.

Accuracy of Machine Learning Models to Predict In-hospital Cardiac Arrest

Purpose/Aims Despite advances in healthcare, the incidence of in-hospital cardiac arrest (IHCA) has continued to rise for the past decade. Identifying those patients at risk has proven challenging. Our objective was to conduct a systematic review of the literature to compare the IHCA predictive performance of machine learning (ML) models with the Modified Early Warning Score (MEWS). Design The systematic review was conducted following the Preferred Reporting Items of Systematic Review and Meta-Analysis guidelines and registered on PROSPERO CRD42020182357. Method Data extraction was completed using the Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies checklist. The risk of bias and applicability were evaluated using the Prediction model Risk of Bias Assessment Tool. Results Nine articles were included in this review that developed or validated IHCA prediction models and compared them with the MEWS. The studies by Jang et al and Kim et al showed that their ML models outperformed MEWS to predict IHCA with good to excellent predictive performance. Conclusions The ML models presented in this systematic review demonstrate a novel approach to predicting IHCA. All included studies suggest that ML models had similar or better predictive performance compared with MEWS. However, there is substantial variability in performance measures and concerns for risk of bias.

Improving machine learning performance on small chemical reaction data with unsupervised contrastive pretraining.

Machine learning (ML) methods have great potential to transform chemical discovery by accelerating the exploration of chemical space and drawing scientific insights from data. However, modern chemical reaction ML models, such as those based on graph neural networks (GNNs), must be trained on a large amount of labelled data in order to avoid overfitting the data and thus possessing low accuracy and transferability. In this work, we propose a strategy to leverage unlabelled data to learn accurate ML models for small labelled chemical reaction data. We focus on an old and prominent problem—classifying reactions into distinct families—and build a GNN model for this task. We first pretrain the model on unlabelled reaction data using unsupervised contrastive learning and then fine-tune it on a small number of labelled reactions. The contrastive pretraining learns by making the representations of two augmented versions of a reaction similar to each other but distinct from other reactions. We propose chemically consistent reaction augmentation methods that protect the reaction center and find they are the key for the model to extract relevant information from unlabelled data to aid the reaction classification task. The transfer learned model outperforms a supervised model trained from scratch by a large margin. Further, it consistently performs better than models based on traditional rule-driven reaction fingerprints, which have long been the default choice for small datasets, as well as those based on reaction fingerprints derived from masked language modelling. In addition to reaction classification, the effectiveness of the strategy is tested on regression datasets; the learned GNN-based reaction fingerprints can also be used to navigate the chemical reaction space, which we demonstrate by querying for similar reactions. The strategy can be readily applied to other predictive reaction problems to uncover the power of unlabelled data for learning better models with a limited supply of labels.

Explainable graph neural networks for organic cages.

The development of accurate and explicable machine learning models to predict the properties of topologically complex systems is a challenge in materials science. Porous organic cages, a class of polycyclic molecular materials, have potential application in molecular separations, catalysis and encapsulation. For most applications of porous organic cages, having a permanent internal cavity in the absence of solvent, a property termed “shape persistence” is critical. Here, we report the development of Graph Neural Networks (GNNs) to predict the shape persistence of organic cages. Graph neural networks are a class of neural networks where the data, in our case that of organic cages, are represented by graphs. The performance of the GNN models was measured against a previously reported computational database of organic cages formed through a range of [4 + 6] reactions with a variety of reaction chemistries. The reported GNNs have an improved prediction accuracy and transferability compared to random forest predictions. Apart from the improvement in predictive power, we explored the explicability of the GNNs by computing the integrated gradient of the GNN input. The contribution of monomers and molecular fragments to the shape persistence of the organic cages could be quantitatively evaluated with integrated gradients. With the added explicability of the GNNs, it was possible not only to accurately predict the property of organic materials, but also to interpret the predictions of the deep learning models and provide structural insights for the discovery of future materials.

On the effectiveness of Gated Echo State Networks for data exhibiting long-term dependencies

In the context of recurrent neural networks, gated architectures such as the GRU have contributed to the development of highly accurate machine learning models that can tackle long-term dependencies in the data. However, the training of such networks is performed by the expensive algorithm of gradient descent with backpropagation through time. On the other hand, reservoir computing approaches such as Echo State Networks (ESNs) can produce models that can be trained efficiently thanks to the use of fixed random parameters, but are not ideal for dealing with data presenting long-term dependencies. We explore the problem of employing gated architectures in ESNs from both theoretical and empirical perspectives. We do so by deriving and evaluating a necessary condition for the non-contractivity of the state transition function, which is important to overcome the fading-memory characterization of conventional ESNs. We find that using pure reservoir computing methodologies is not sufficient for effective gating mechanisms, while instead training even only the gates is highly effective in terms of predictive accuracy.

Machine Learning Model Generation With Copula-Based Synthetic Dataset for Local Differentially Private Numerical Data

With the development of IoT technology, personal data are being collected in many places. These data can be used to create new services, but consideration must be given to the individual’s privacy. We can safely collect personal data while adding noise by applying differential privacy. However, because such data are very noisy, the accuracy of machine learning trained by the data greatly decreased. In this study, our objective is to build a highly accurate machine learning model using these data. We focus on the decision tree machine learning algorithm, and, instead of applying it as is, we use a preprocessing technique wherein pseudodata are generated using a copula while removing the effect of noise added by differential privacy. In detail, the proposed novel protocol consists of three steps: generating a covariance matrix from the differentially private numerical data, generating a discrete cumulative distribution function from differentially private numerical data, and generating copula-based numerical samples. Simulation results using synthetic and real datasets verify the utility of the proposed method not only for the decision tree algorithm but also for other machine learning algorithms such as deep neural networks. This method will help create machine learning models, such as recommendation systems, using differential privacy data.

Comparison of the Performance of Machine Learning Algorithms in Predicting Heart Disease

Introduction: Heart disease is often associated with conditions such as clogged arteries due to the sediment accumulation which causes chest pain and heart attack. Many people die due to the heart disease annually. Most countries have a shortage of cardiovascular specialists and thus, a significant percentage of misdiagnosis occurs. Hence, predicting this disease is a serious issue. Using machine learning models performed on multidimensional dataset, this article aims to find the most efficient and accurate machine learning models for disease prediction.Material and Methods: Several algorithms were utilized to predict heart disease among which Decision Tree, Random Forest and KNN supervised machine learning are highly mentioned. The algorithms are applied to the dataset taken from the UCI repository including 294 samples. The dataset includes heart disease features. To enhance the algorithm performance, these features are analyzed, the feature importance scores and cross validation are considered.Results: The algorithm performance is compared with each other, so that performance based on ROC curve and some criteria such as accuracy, precision, sensitivity and F1 score were evaluated for each model. As a result of evaluation, Accuracy, AUC ROC are 83% and 99% respectively for Decision Tree algorithm. Logistic Regression algorithm with accuracy and AUC ROC are 88% and 91% respectively has better performance than other algorithms. Therefore, these techniques can be useful for physicians to predict heart disease patients and prescribe them correctly.Conclusion: Machine learning technique can be used in medicine for analyzing the related data collections to a disease and its prediction. The area under the ROC curve and evaluating criteria related to a number of classifying algorithms of machine learning to evaluate heart disease and indeed, the prediction of heart disease is compared to determine the most appropriate classification. As a result of evaluation, better performance was observed in both Decision Tree and Logistic Regression models.

Prediction and associated factors of hypothyroidism in systemic lupus erythematosus: a cross-sectional study based on multiple machine learning algorithms

Objectives The prevalence of hypothyroidism in systemic lupus erythematosus (SLE) is significantly higher than that in the common public. While SLE itself can affect multiple organs, abnormal thyroid function may exacerbate organ damage in patients with SLE. We aimed to predict abnormal thyroid function and to examine the associated factors with multiple machine learning approaches. Methods In a cross-sectional study, 255 patients diagnosed with SLE at the rheumatology department in Xiangya Hospital between June 2012 and December 2016 were investigated. Feature engineering was used for filtering out principle clinical parameters, and five different machine learning methods were used to build prediction models for SLE with hypothyroidism. Results Feature engineering selected 11 variables with which to build machine learning models. Among them, random forest modelling obtained the best prediction performance, with an accuracy rate of 88.37 and an area under the receiver operating characteristic curve of 0.772. The weights of anti-SSB antibody and anti-dsDNA antibody were 1.421 and 1.011, respectively, indicating a strong association with hypothyroidism in SLE. Conclusions Random Forest model performed best and is recommended for selecting vital indices and assessing clinical complications of SLE, it indicated that anti-SSB and anti-dsDNA antibodies may play principal roles in the development of hypothyroidism in SLE patients. It’s feasible to build an accurate machine learning model for early diagnosis or risk factors assessment in SLE using clinical parameters, which would provide a reference for the research work of SLE in China.

Differential mechanisms of posterior cingulate cortex downregulation and symptom decreases in posttraumatic stress disorder and healthy individuals using real-time fMRI neurofeedback.

BackgroundIntrinsic connectivity networks, including the default mode network (DMN), are frequently disrupted in individuals with posttraumatic stress disorder (PTSD). The posterior cingulate cortex (PCC) is the main hub of the posterior DMN, where the therapeutic regulation of this region with real‐time fMRI neurofeedback (NFB) has yet to be explored.MethodsWe investigated PCC downregulation while processing trauma/stressful words over 3 NFB training runs and a transfer run without NFB (total n = 29, PTSD n = 14, healthy controls n = 15). We also examined the predictive accuracy of machine learning models in classifying PTSD versus healthy controls during NFB training.ResultsBoth the PTSD and healthy control groups demonstrated reduced reliving symptoms in response to trauma/stressful stimuli, where the PTSD group additionally showed reduced symptoms of distress. We found that both groups were able to downregulate the PCC with similar success over NFB training and in the transfer run, although downregulation was associated with unique within‐group decreases in activation within the bilateral dmPFC, bilateral postcentral gyrus, right amygdala/hippocampus, cingulate cortex, and bilateral temporal pole/gyri. By contrast, downregulation was associated with increased activation in the right dlPFC among healthy controls as compared to PTSD. During PCC downregulation, right dlPFC activation was negatively correlated to PTSD symptom severity scores and difficulties in emotion regulation. Finally, machine learning algorithms were able to classify PTSD versus healthy participants based on brain activation during NFB training with 80% accuracy.ConclusionsThis is the first study to investigate PCC downregulation with real‐time fMRI NFB in both PTSD and healthy controls. Our results reveal acute decreases in symptoms over training and provide converging evidence for EEG‐NFB targeting brain networks linked to the PCC.

Estimation and Interpretation of Machine Learning Models with Customized Surrogate Model

Machine learning has the potential to predict unseen data and thus improve the productivity and processes of daily life activities. Notwithstanding its adaptiveness, several sensitive applications based on such technology cannot compromise our trust in them; thus, highly accurate machine learning models require reason. Such models are black boxes for end-users. Therefore, the concept of interpretability plays the role if assisting users in a couple of ways. Interpretable models are models that possess the quality of explaining predictions. Different strategies have been proposed for the aforementioned concept but some of these require an excessive amount of effort, lack generalization, are not agnostic and are computationally expensive. Thus, in this work, we propose a strategy that can tackle the aforementioned issues. A surrogate model assisted us in building interpretable models. Moreover, it helped us achieve results with accuracy close to that of the black box model but with less processing time. Thus, the proposed technique is computationally cheaper than traditional methods. The significance of such a novel technique is that data science developers will not have to perform strenuous hands-on activities to undertake feature engineering tasks and end-users will have the graphical-based explanation of complex models in a comprehensive way—consequently building trust in a machine.

Experimental and numerical model for mechanical properties of concrete containing fly ash: Systematic review

Increasing use and need for cement have forced researchers to build alternative building materials that are environmentally friendly in nature and help manage less waste. Using waste materials to produce the structural concrete needs theoretical models for implement application for construction. Fly ash as a byproduct of coal-fired electric generating plants showed many advantages in improving the properties of concrete due to its pozzolanic feature. Therefore, this study intends to review previous studies and propose new models to determine the elastic moduli, compressive and tensile strengths of concrete produced by fly ash as a replacement of cement. For this aim, wide-range experimental results were evaluated and provided from previous studies. The disadvantage of the previous study is not covering all elastic moduli, compressive and tensile strengths together. Therefore, a total of 263 concrete mixtures were also produced in this study and the gap of the previous investigations was filled. Therefore, highly accurate machine learning models were used in MATLAB software to predict the mechanical properties of concrete produced by fly ash including radial basis function, multilayer perceptron, support vector regression, adaptive-network-based fuzzy inference system and deep neural network. Additionally, the experimental results were compared with existing models and new highly accurate models were developed to determine the modulus of elasticity, compressive and tensile strengths of concrete produced by various fly ash contents. Experimental results showed that the optimal content for fly ash was obtained by 10% in terms of the maximum mechanical properties of concrete. In addition, used Artificial Neural Networks, particularly deep neural networks showed a highly accurate prediction. Moreover, the proposed models by a high agreement with experimental results (R2>0.98) could be used as highly efficient and accurate tools to determine the mechanical properties of fly ash concrete.

Predicting Risk of Stroke From Lab Tests Using Machine Learning Algorithms: Development and Evaluation of Prediction Models.

BackgroundStroke, a cerebrovascular disease, is one of the major causes of death. It causes significant health and financial burdens for both patients and health care systems. One of the important risk factors for stroke is health-related behavior, which is becoming an increasingly important focus of prevention. Many machine learning models have been built to predict the risk of stroke or to automatically diagnose stroke, using predictors such as lifestyle factors or radiological imaging. However, there have been no models built using data from lab tests.ObjectiveThe aim of this study was to apply computational methods using machine learning techniques to predict stroke from lab test data.MethodsWe used the National Health and Nutrition Examination Survey data sets with three different data selection methods (ie, without data resampling, with data imputation, and with data resampling) to develop predictive models. We used four machine learning classifiers and six performance measures to evaluate the performance of the models.ResultsWe found that accurate and sensitive machine learning models can be created to predict stroke from lab test data. Our results show that the data resampling approach performed the best compared to the other two data selection techniques. Prediction with the random forest algorithm, which was the best algorithm tested, achieved an accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and area under the curve of 0.96, 0.97, 0.96, 0.75, 0.99, and 0.97, respectively, when all of the attributes were used.ConclusionsThe predictive model, built using data from lab tests, was easy to use and had high accuracy. In future studies, we aim to use data that reflect different types of stroke and to explore the data to build a prediction model for each type.

Gradient domain machine learning with composite kernels: improving the accuracy of PES and force fields for large molecules

The generalization accuracy of machine learning models of potential energy surfaces (PES) and force fields (FF) for large polyatomic molecules can be improved either by increasing the number of training points or by improving the models. In order to build accurate models based on expensive ab initio calculations, much of recent work has focused on the latter. In particular, it has been shown that gradient domain machine learning (GDML) models produce accurate results for high-dimensional molecular systems with a small number of ab initio calculations. The present work extends GDML to models with composite kernels built to maximize inference from a small number of molecular geometries. We illustrate that GDML models can be improved by increasing the complexity of underlying kernels through a greedy search algorithm using Bayesian information criterion as the model selection metric. We show that this requires including anisotropy into kernel functions and produces models with significantly smaller generalization errors. The results are presented for ethanol, uracil, malonaldehyde and aspirin. For aspirin, the model with composite kernels trained by forces at 1000 randomly sampled molecular geometries produces a global 57-dimensional PES with the mean absolute accuracy 0.177 kcal mol−1 (61.9 cm−1) and FFs with the mean absolute error 0.457 kcal mol−1 Å−1.

Efficient prediction for high precision CO-N2 potential energy surface by stacking ensemble DNN.

High-dimensional potential energy surface (PES) for van der Waals systems with spectroscopic accuracy, is of great importance for quantum dynamics and an extremely challenge job. CO-N2 is a typical van der Waals system and its high-precision PES may help elucidate weak interaction mechanisms. Taking CO-N2 potential energies calculated by CCSD(T)-F12b/aug-cc-pVQZ as the benchmark, we establish an accurate, robust, and efficient machine learning model by using only four molecular structure descriptors based on 7966 benchmark potential energies. The highest accuracy is obtained by a stacking ensemble DNN (SeDNN). Its evaluation parameters MAE, RMSE, and R2 reach 0.096, 0.163, 0.9999 cm-1 , respectively, and the spectroscopic accuracy for vibration spectrum is achieved with predicted PES, which shows SeDNN superior goodness-of-fit and prediction performance. An elaborated PES with the reported global minimum has been predicted with the model, which perfectly reproduces CCSD(T) potential energies and the analytical MLR PES [PCCP, 2018, 20, 2036]. The critical points (global minimum, TSI, TSII, and their barriers), potential curve, and entire PES profile are remarkably consistent with CCSD(T) calculations. To further improve the usability of constructing PESs in practice, the size of the training set (energy points) for the model is reduced to 50%, 30%, and 20% of the database, respectively. The results show that even training with the smallest training set (1593 points), the PES only differs 2.555 cm-1 with the analytic MLR PES. Therefore, the proposed SeDNN is promisingly an alternative efficient tool to construct subtle PES for van der Waals systems.

Boosting biomedical document classification through the use of domain entity recognizers and semantic ontologies for document representation: The case of gluten bibliome

The increasing number of scientific research documents published keeps growing at an unprecedented rate, making it increasingly difficult to access practical information within a target domain. This situation is motivating a growing interest in applying text mining techniques for the automatic processing of text resources to structure the information that helps researchers to find information of interest and infer knowledge of practical use. However, the automatic processing of research documents requires the previous existence of large, manually annotated text corpora to develop robust and accurate text mining processing methods and machine learning models. In this context, semi-automatic extraction techniques based on structured data and state-of-the-art biomedical tools appear to have significant potential to enhance curator productivity and reduce the costs of document curation. In this line, this work proposes a semi-automatic machine learning workflow and a NER + Ontology boosting technique for the automatic classification of biomedical literature. The practical relevance of the proposed approach has been proven in the curation of 4,115 gluten-related documents extracted from PubMed and contrasted against the word embedding alternative. Comparing the results of the experiments, the proposed NER + Ontology technique is an effective alternative to other state-of-the-art document representation techniques to process the existing biomedical literature.

Prediction of irrigation groundwater quality parameters using ANN, LSTM, and MLR models.

Forecasting the irrigation groundwater parameters helps plan irrigation water and crop, and it is commonly expensive because it needs various parameters, mainly in developing nations. Therefore, the present research's core objective is to create accurate and reliable machine learning models for irrigation parameters. To accomplish this determination, three machine learning (ML) models, viz. long short-term memory (LSTM), multi-linear regression (MLR), and artificial neural network (ANN), have been trained. It is validated with mean squared error (MSE) and correlation coefficients (r), root mean square error (RMSE), and mean absolute error (MAE). These machine learning models have been used and applied for predicating the six irrigation water quality parameters such as sodium absorption ratio (SAR), percentage of sodium (%Na), residual sodium carbonate (RSC), magnesium hazard (MH), Permeability Index (PI), and Kelly ratio (KR). Therefore, the two scenario performances of ANN, LSTM, and MLR have been developed for each model to predict irrigation water quality parameters. The first and second scenario performance was created based on all and second reduction input variables. The ANN, LSTM, and MLR models have discovered that excluding for ANN and MLR models shows high accuracy in first and second scenario models, respectively. These model's accuracy was checked based on the mean squared error (MSE), correlation coefficients (r), and root mean square error (RMSE) for training and testing processes serially. The RSC values are highly accurate predicated values using ANN and MLR models. As a result, machine learning models may improve irrigation water quality parameters, and such types of results are essential to farmers and crop planning in various irrigation processes.

Modeling hydrogen solubility in alcohols using machine learning models and equations of state

Knowledge of hydrogen (H2) solubility in alcohols is important for designing and performing various processes in chemical plants. Accurate predictions of H2 solubility in alcohols can affect the quality and applications of pharmaceuticals, perfume, cosmetics, flavor, and many others. In this work, deep echo state network (DeepESN), extreme gradient boosting (XGBoost), extreme learning machine (ELM), and multivariate adaptive regression splines (MARS) as four advanced machine learning models were utilized for predicting the H2 solubility in alcohols. To this end, a complete set of H2 solubility data (673 experimental data points) for 26 different alcohols or alcohol-containing solvents is gathered over a wide range of operating pressure (0.101–110.3 MPa) and temperature (213.15–524.9 K). The XGBoost model was obtained as the best model for estimation H2 solubility in alcohols based on graphical and statistical analyses having a root mean square error of 0.0022 and coefficient of determination of 0.9946. Four well-known equations of state (EOSs) were also utilized to estimate H2 solubility in alcohols, among which Redlich-Kwong EOS had the best performance. However, the accuracy of machine learning models was much higher than the EOSs. Based on sensitivity analysis, pressure, temperature, and molecular weight of alcohols have the highest impact on the solubility of H2 in alcohols, respectively. Eventually, the Leverage approach was utilized to recognize the applicability domain of the XGBoost model and probable outlier data, the results of which show that this model has high credit for estimating the solubility of H2 in alcohols. The outcome of this study can help to design the hydrogenation process in chemical plants, and the XGBoost model can act as an efficient predictor for predicting H2 solubility in alcohols.