Permutation Importance Research Articles

Abstract 4138472: Bioactive Oxylipins Predict Coronary Artery Calcification Severity in Heart Failure with Preserved Ejection Fraction

Background: Inflammation is a common underlying cause in atherosclerosis and heart failure with preserved ejection fraction (HFpEF). Recent data have highlighted the critical role of circulating bioactive polyunsaturated fatty acid oxidized lipids, or oxylipins, in initiating and resolving inflammation. We sought to study the role of oxylipins in predicting coronary artery calcification (CAC) in HFpEF. Hypothesis: We hypothesize that oxylipins can predict and distinguish CAC severity in HFpEF patients. Methods: We collected venous blood samples during right heart catheterization from 71 HFpEF patients. We categorized patients by CT-based CAC severity: none (control; n=16), mild (<25% CAC, n=7), moderate (26-50% CAC, n=22), and severe (>50% CAC, n=26). We measured 143 oxylipins in all study participants using mass spectrometry. Along with unpaired T-tests, machine learning analysis was used to determine lipid permutation importance and evaluate predictive accuracy for CAC severity. Results: Demographic variables showed no significant differences between groups. Increased total cholesterol (p=0.031) and low HDL (p=0.011) were associated with CAC severity. Pro-inflammatory 12(S)-HHTrE (a degradation product of TxA2) levels were elevated in severe CAC vs control (p=0.031). The inactivated anti-inflammatory 5,6 DiHETrE levels were higher in moderate CAC vs control (p=0.032) and severe CAC vs control (p=0.021). Out of 143 oxylipins, the panel of LTB5, 11(R)-HEDE, 20-HETE, 5-HEPE, 11(12)-EpETrE, 14(15)-EpETE, 19,20-DiHDoPE, PD1, and 5-oxoETE were the most significant predictors of severe vs non-severe (none, mild, moderate) CAC, with 88.88% predictive accuracy. Conclusion: In HFpEF patients, CAC severity can be predicted by a novel panel of circulating, bioactive venous oxylipins.

Supervised machine learning-based categorization and prediction of uranium adsorption capacity on various process parameters

Existing uranium poses significant dangers to the environment and the general population's health. Within the scope of this study, machine learning techniques were utilized to assess the characteristics that influence uranium adsorption. The adsorbent material, pH, temperature, solid-liquid ratio (SLR), and initial concentration (Ci) values were imported as the feature to evaluate using ML models in this study. The finding concluded that the Convolutional Neural Network (CNN) model performed excellently by attaining an accuracy of 98%, which is very high in classifying low and high levels of adsorption predictions. The Random Forest model was also the best performer for Qe prediction, with the lowest MSE value of 334.45 and the highest Squared-R score of 0.92 for the uranium adsorption process. Permutation Importance analysis indicated that initial concentration had the strongest impact on Qe. Knowing that a good interaction between uranyl ions and adsorbent surface would result in more active sites for trapping radioactive metal, optimizing this parameter can be quite helpful for designing new materials capable of capturing uranium species strongly from solutions. ML model excels at predicting Qe in uranium adsorption and can directly accommodate large datasets and non-linear input parameter-Qe interactions. As noted, the innovation poised to drive ML research has been made possible by recent technological developments in machine learning and can improve while offering new ways of addressing difficult environmental problems.

Generating a Landslide Susceptibility Map Using Integrated Meta-Heuristic Optimization and Machine Learning Models

A landslide susceptibility assessment is one of the critical steps in planning for landslide disaster prevention. Advanced machine learning methods can be used as data-driven approaches for landslide susceptibility zonation with several landslide conditioning factors. Despite there being a number of studies on landslide susceptibility assessment, the literature is limited in several contexts, such as parameter optimization, an examination of the factors in detail, and study area. This study addresses these lacks in the literature and aims to develop a landslide susceptibility map of Kentucky, US. Four machine learning methods, namely artificial neural network (ANN), k-nearest neighbor (KNN), support vector machine (SVM), and stochastic gradient boosting (SGB), were used to train the dataset comprising sixteen landslide conditioning factors after pre-processing the data in terms of data encoding, data scaling, and dimension reduction. The hyperparameters of the machine learning methods were optimized using a state-of-the-art artificial bee colony (ABC) algorithm. The permutation importance and Shapley additive explanations (SHAP) methods were employed to reduce the dimension of the dataset and examine the contributions of each landslide conditioning factor to the output variable, respectively. The findings show that the ABC-SGB hybrid model achieved the highest prediction performance. The SHAP summary plot developed using the ABC-SGB model shows that intense precipitation, distance to faults, and slope were the most significant factors affecting landslide susceptibility. The SHAP analysis further underlines that increases in intense precipitation, distance to faults, and slope are associated with an increase in the probability of landslide incidents. The findings attained in this study can be used by decision makers to develop the most effective resource allocation plan for preventing landslides and minimizing related damages.

Macro-Driven Stock Market Volatility Prediction: Insights from a New Hybrid Machine Learning Approach

This study comprehensively investigates stock market volatility based on over one hundred monthly macroeconomic variables, applying machine learning models. Methodological contribution integrating the random forest (RF) with the least absolute shrinkage and selection operator methods (LASSO). Importantly, the RF-LASSO model can robustly achieve the best forecasting performance under different circumstances. In addition, we focus on model explanation from different perspectives based on permutation importance and shapley additive explanation (SHAP) methods. This study illuminates novel insights into the realm of stock market volatility, harnessing the transformative potential of machine learning methodologies.

Predicting ignitability classification of thermally thick solids using hybrid GA-BPNN and PSO-BPNN algorithms

Two hybrid binary classification models combining the merits of GA (genetic algorithm) and PSO (particle swarm optimization) with BPNN (backpropagation neural network) are proposed to estimate the ignitability of thermally thick solids. A 1D numerical model verified by analytical correlation and experimental measurements is employed to yield training, validation and testing datasets. Surface absorptivity (ε), density (ρ), specific heat (Cp), thermal conductivity (k), critical temperature (Tcri), and incident heat flux (HF) which are highly related to solid ignition are selected as inputs, and ignitability serves as output. A one-hidden-layer BPNN is first constructed, and then GA and PSO are encoded to form GA-BPNN and PSO-BPNN where connection weights and biases are optimized by GA/PSO and the predicted error serves as objective function of GA/PSO. Results show that the optimal neuron number in hidden layer is 34. Compared to BPNN, GA-BPNN and PSO-BPNN feature higher accuracy but more training time. Accuracy and convergence efficiency of PSO-BPNN are higher than GA-BPNN, nevertheless GA-BPNN exhibits better robustness. Predictive accuracies of BPNN, GA-BPNN, and PSO-BPNN on testing dataset are 99.05%, 99.55%, and 99.55%, respectively. ROC (receiver operating characteristic) curves are plotted to rank the performance of the three models, revealing 0.5 is an appropriate threshold to classify ignitability. Feature importance of the models is analyzed using PI (permutation importance) and MIV (mean impact value) methods. It is found HF and Tcri are the most important two inputs followed by ε, while ρ, Cp, and k show least and approximately identical impact. Reliability of the three models is verified by predicting critical heat flux of ignition of three polymers, and good agreement with literature is observed.

Ensemble approach for predicting the diagnosis of osteoarthritis using physical activity factors.

Osteoarthritis (OA) is a common degenerative disease of the joints. Risk factors for OA include non-modifiable factors such as age and sex, as well as modifiable factors like physical activity. this study aimed to construct a soft voting ensemble model to predict OA diagnosis using variables related to individual characteristics and physical activity and identify important variables in constructing the model through permutation importance. By using the recursive feature elimination, cross-validated technique, the variables with the best predictive performance were selected among variables, and an ensemble model combining RandomForest, XGBoost, and LightGBM algorithms was constructed. The predictive performance and permutation importance of each variable were evaluated. The variables selected to construct the model were age, sex, grip strength, and quality of life, and the accuracy of the ensemble model was 0.828. The most important variable in constructing the model was age (0.199), followed by grip strength (0.053), quality of life (0.043), and sex (0.034). The performance of the model for predicting OA was relatively good. If this model is continuously used and updated, it could be used to predict OA diagnosis, and the predictive performance of the OA model may be further improved.

Investigating the Mechanisms Impacting Soil Ca, Mg, and Na in Wastewater Land Application Systems Using Machine Learning Models

ABSTRACTWastewater land application is a widely accepted solution for addressing global water crisis, particularly in arid and semiarid regions, but it may cause soil Ca, Mg, and Na accumulation and result in soil degradation. The objective of this study was to investigate the underlying mechanisms impacting soil Ca, Mg, and Na in wastewater land application systems using tree‐based machine learning models. Using data collected from previous field studies, decision tree (DT), random forest (RF), and extreme gradient boosting decision trees (XGBoost) models were developed to predict soil Ca, Mg, and Na in wastewater land application systems. XGBoost models showed the best performance, with R2 and RMSE values of 0.999 and 18.9 mg/kg, 0.999 and 3.2 mg/kg, and 0.912 and 104 mg/kg on the training data and 0.989 and 345 mg/kg, 0.925 and 56.1 mg/kg, and 0.908 and 112 mg/kg on the test data for soil Ca, Mg, and Na prediction, respectively. Permutation importance analysis reveals that initial soil Ca and electrical conductivity (EC) and total irrigation amount, initial soil Mg, total precipitation and initial soil EC, and initial soil Na, total irrigation amount and wastewater Na were the top three predictive variables for soil Ca, Mg, and Na, respectively. Partial dependence analysis demonstrates how soil Ca, Mg, and Na changed with the predictive variables, and indicates that wastewater irrigation caused soil Ca, Mg, and Na accumulation. This study highlights the need for sustainable wastewater land application management to control soil sodium adsorption ratio and mitigate the risks of land degradation.

Enhancing Network Threat Detection with Random Forest-Based NIDS and Permutation Feature Importance

Network Intrusion Detection Systems (NIDS) are critical for protecting computer networks from unauthorized activities. Traditional NIDS rely on rule-based signatures, which can be limiting in detecting emerging threats. This study investigates the effectiveness of the random forest classifier in advancing NIDS capabilities through machine learning. Using the CICIDS-2017 dataset, the data are preprocessed to enhance their quality by removing redundancies. feature selection and permutation importance were employed to identify the most relevant features. The methodology involves rigorous testing and analysis of the random forest classifier’s performance, focusing on f1-score rates compared to other machine learning models. Results demonstrate that by optimizing class weights, applying a custom prediction function and leveraging 26 key features, the random forest classifier achieves an outstanding 99.8% in the weighted f1-score and 93.31% in the macro f1-score in various attack types. This research highlights the potential of machine learning to significantly enhance NIDS effectiveness, offering a robust defense mechanism against evolving cybersecurity threats in modern networks.

Understanding Public Opinion towards ESG and Green Finance with the Use of Explainable Artificial Intelligence

This study leverages explainable artificial intelligence (XAI) techniques to analyze public sentiment towards Environmental, Social, and Governance (ESG) factors, climate change, and green finance. It does so by developing a novel multi-task learning framework combining aspect-based sentiment analysis, co-reference resolution, and contrastive learning to extract nuanced insights from a large corpus of social media data. Our approach integrates state-of-the-art models, including the SenticNet API, for sentiment analysis and implements multiple XAI methods such as LIME, SHAP, and Permutation Importance to enhance interpretability. Results reveal predominantly positive sentiment towards environmental topics, with notable variations across ESG categories. The contrastive learning visualization demonstrates clear sentiment clustering while highlighting areas of uncertainty. This research contributes to the field by providing an interpretable, trustworthy AI system for ESG sentiment analysis, offering valuable insights for policymakers and business stakeholders navigating the complex landscape of sustainable finance and climate action. The methodology proposed in this paper advances the current state of AI in ESG and green finance in several ways. By combining aspect-based sentiment analysis, co-reference resolution, and contrastive learning, our approach provides a more comprehensive understanding of public sentiment towards ESG factors than traditional methods. The integration of multiple XAI techniques (LIME, SHAP, and Permutation Importance) offers a transparent view of the subtlety of the model’s decision-making process, which is crucial for building trust in AI-driven ESG assessments. Our approach enables a more accurate representation of public opinion, essential for informed decision-making in sustainable finance. This paper paves the way for more transparent and explainable AI applications in critical domains like ESG.

Explainability of artificial neural network in predicting career fulfilment among medical doctors in developing nations: Applicability and implications

BackgroundCareer fulfilment among medical doctors is crucial for job satisfaction, retention, and healthcare quality, especially in developing nations with challenging healthcare systems. Traditional career guidance methods struggle to address the complexities of career fulfilment. While recent advancements in machine learning, particularly Artificial Neural Network (ANN) models, offer promising solutions for personalized career predictions, their applicability, interpretability, and impact remain challenging. MethodThis study explores the applicability, explainability, and implications of ANN models in predicting career fulfillment among medical doctors in developing nations, considering socio-economic, psychological, and professional factors. Box plots visualized data distribution, while Heatmaps assessed data intensity and relationships. Matthew's correlation coefficient and Taylor's chart were used to evaluate model performance. Input feature contributions to ANN predictions were analyzed using permutation importance, SHAP, LIME, and Williams plots. The model was tested on a dataset tailored to medical professionals in Nigeria and China, with evaluation metrics including Mean Absolute Error, Mean Squared Error, Root Mean Squared Error, and R2 Score. ResultsThe ANN model demonstrates strong predictive accuracy, capturing relationships between input factors and outcomes. For Chinese doctors, it achieved an MSE of 0.0004 and R2 of 0.9994, while for Nigerian doctors, it recorded an MSE of 0.0003 and R2 of 0.9998. Key factors for Chinese doctors' satisfaction were IF1 and IF2, while EF1 and EF3 were crucial in preventing dissatisfaction. For Nigerian doctors, IF2 and IF3 drove satisfaction, while EF1 and EF4 were significant in avoiding dissatisfaction. ConclusionThe results highlights the ANN model's effectiveness in predicting career fulfillment among medical doctors in developing nations, offering a valuable tool for career guidance, policymaking, and improving job satisfaction, retention, and healthcare quality.

A machine learning approach towards assessing consistency and reproducibility: an application to graft survival across three kidney transplantation eras.

In South Africa, between 1966 and 2014, there were three kidney transplant eras defined by evolving access to certain immunosuppressive therapies defined as Pre-CYA (before availability of cyclosporine), CYA (when cyclosporine became available), and New-Gen (availability of tacrolimus and mycophenolic acid). As such, factors influencing kidney graft failure may vary across these eras. Therefore, evaluating the consistency and reproducibility of models developed to study these variations using machine learning (ML) algorithms could enhance our understanding of post-transplant graft survival dynamics across these three eras. This study explored the effectiveness of nine ML algorithms in predicting 10-year graft survival across the three eras. We developed and internally validated these algorithms using data spanning the specified eras. The predictive performance of these algorithms was assessed using the area under the curve (AUC) of the receiver operating characteristics curve (ROC), supported by other evaluation metrics. We employed local interpretable model-agnostic explanations to provide detailed interpretations of individual model predictions and used permutation importance to assess global feature importance across each era. Overall, the proportion of graft failure decreased from 41.5% in the Pre-CYA era to 15.1% in the New-Gen era. Our best-performing model across the three eras demonstrated high predictive accuracy. Notably, the ensemble models, particularly the Extra Trees model, emerged as standout performers, consistently achieving high AUC scores of 0.95, 0.95, and 0.97 across the eras. This indicates that the models achieved high consistency and reproducibility in predicting graft survival outcomes. Among the features evaluated, recipient age and donor age were the only features consistently influencing graft failure throughout these eras, while features such as glomerular filtration rate and recipient ethnicity showed high importance in specific eras, resulting in relatively poor historical transportability of the best model. Our study emphasises the significance of analysing post-kidney transplant outcomes and identifying era-specific factors mitigating graft failure. The proposed framework can serve as a foundation for future research and assist physicians in identifying patients at risk of graft failure.

Predictors of Medical and Dental Clinic Closure by Machine Learning Methods: Cross-Sectional Study Using Empirical Data.

Small clinics are important in providing health care in local communities. Accurately predicting their closure would help manage health care resource allocation. There have been few studies on the prediction of clinic closure using machine learning techniques. This study aims to test the feasibility of predicting the closure of medical and dental clinics (MCs and DCs, respectively) and investigate important factors associated with their closure using machine running techniques. The units of analysis were MCs and DCs. This study used health insurance administrative data. The participants of this study ran and closed clinics between January 1, 2020, and December 31, 2021. Using all closed clinics, closed and run clinics were selected at a ratio of 1:2 based on the locality of study participants using the propensity matching score of logistic regression. This study used 23 and 19 variables to predict the closure of MCs and DCs, respectively. Key variables were extracted using permutation importance and the sequential feature selection technique. Finally, this study used 5 and 6 variables of MCs and DCs, respectively, for model learning. Furthermore, four machine learning techniques were used: (1) logistic regression, (2) support vector machine, (3) random forest (RF), and (4) Extreme Gradient Boost. This study evaluated the modeling accuracy using the area under curve (AUC) method and presented important factors critically affecting closures. This study used SAS (version 9.4; SAS Institute Inc) and Python (version 3.7.9; Python Software Foundation). The best-fit model for the closure of MCs with cross-validation was the support vector machine (AUC 0.762, 95% CI 0.746-0.777; P<.001) followed by RF (AUC 0.736, 95% CI 0.720-0.752; P<.001). The best-fit model for DCs was Extreme Gradient Boost (AUC 0.700, 95% CI 0.675-0.725; P<.001) followed by RF (AUC 0.687, 95% CI 0.661-0.712; P<.001). The most significant factor associated with the closure of MCs was years of operation, followed by population growth, population, and percentage of medical specialties. In contrast, the main factor affecting the closure of DCs was the number of patients, followed by annual variation in the number of patients, year of operation, and percentage of dental specialists. This study showed that machine running methods are useful tools for predicting the closure of small medical facilities with a moderate level of accuracy. Essential factors affecting medical facility closure also differed between MCs and DCs. Developing good models would prevent unnecessary medical facility closures at the national level.

Feature Identification Using Interpretability Machine Learning Predicting Risk Factors for Disease Severity of In-Patients with COVID-19 in South Florida.

Objective: The objective of the study was to establish an AI-driven decision support system by identifying the most important features in the severity of disease for Intensive Care Unit (ICU) with Mechanical Ventilation (MV) requirement, ICU, and InterMediate Care Unit (IMCU) admission for hospitalized patients with COVID-19 in South Florida. The features implicated in the risk factors identified by the model interpretability can be used to forecast treatment plans faster before critical conditions exacerbate. Methods: We analyzed eHR data from 5371 patients diagnosed with COVID-19 from South Florida Memorial Healthcare Systems admitted between March 2020 and January 2021 to predict the need for ICU with MV, ICU, and IMCU admission. A Random Forest classifier was trained on patients' data augmented by SMOTE, collected at hospital admission. We then compared the importance of features utilizing different model interpretability analyses, such as SHAP, MDI, and Permutation Importance. Results: The models for ICU with MV, ICU, and IMCU admission identified the following factors overlapping as the most important predictors among the three outcomes: age, race, sex, BMI, diarrhea, diabetes, hypertension, early stages of kidney disease, and pneumonia. It was observed that individuals over 65 years ('older adults'), males, current smokers, and BMI classified as 'overweight' and 'obese' were at greater risk of severity of illness. The severity was intensified by the co-occurrence of two interacting features (e.g., diarrhea and diabetes). Conclusions: The top features identified by the models' interpretability were from the 'sociodemographic characteristics', 'pre-hospital comorbidities', and 'medications' categories. However, 'pre-hospital comorbidities' played a vital role in different critical conditions. In addition to individual feature importance, the feature interactions also provide crucial information for predicting the most likely outcome of patients' conditions when urgent treatment plans are needed during the surge of patients during the pandemic.

Protocol for the development and validation of a machine-learning based tool for predicting the risk of hypertriglyceridemia in critically-ill patients receiving propofol sedation.

Introduction Propofol is a widely used sedative-hypnotic agent for critically-ill patients requiring invasive mechanical ventilation (IMV). Despite its clinical benefits, propofol is associated with increased risks of hypertriglyceridemia. Early identification of patients at risk for propofol-associated hypertriglyceridemia is crucial for optimizing sedation strategies and preventing adverse outcomes. Machine learning (ML) models offer a promising approach to predict individualized patient risks of propofol-associated hypertriglyceridemia. Methods and analysis We propose the development of a ML model aimed at predicting the risk of propofol-associated hypertriglyceridemia in ICU patients receiving IMV. The study will utilize retrospective data from four Mayo Clinic sites. Nested cross-validation (CV) will be employed, with a 10-fold inner CV loop for model tuning and selection as well as an outer loop using leave-one-site-out CV for external validation. Feature selection will be conducted using Boruta and LASSO-penalized logistic regression. Data preprocessing steps include missing data imputation, feature scaling, and dimensionality reduction techniques. Six ML algorithms will be tuned and evaluated. Bayesian optimization will be used for hyperparameter selection. Global model explainability will be assessed using permutation importance, and local model explainability will be assessed using SHAP. Ethics and dissemination The proposed ML model aims to provide a reliable and interpretable tool for clinicians to predict the risk of propofol-associated hypertriglyceridemia in ICU patients. The final model will be deployed in a web-based clinical risk calculator. The model development process and performance measures obtained during nested cross-validation will be described in a study publication to be disseminated in a peer-reviewed journal. The proposed study has received ethics approval from the Mayo Clinic Institutional Review Board (IRB #23-007416).

High-resolution compound-specific mapping in works of art via data fusion of MA-XRPD with hyperspectral data (part 1: Method evaluation)

BackgroundHyperspectral imaging techniques have emerged as powerful tools for non-invasive investigation of artworks. This paper employs either reflectance imaging spectroscopy (RIS) or macroscopic X-ray fluorescence (MA-XRF) imaging in combination with macroscopic X-ray powder diffraction (MA-XRPD) for state-of-the-art chemical imaging of painted cultural heritage artefacts. While RIS can provide molecular information and MA-XRF can offer elemental distribution maps of paintings of high lateral resolution, the unique advantage of MA-XRPD lies in its ability to visualize the distributions of specific pigments and estimate in a quantitative manner the relative concentrations of the crystalline phases at the surface of artworks. However, MA-XRPD is more time-consuming and offers a lower lateral resolution than RIS and MA-XRF. ResultsThis study introduces a machine learning (ML) approach to obtain the distribution of specific compounds on the surface of artworks with a resolution that is comparable to that of RIS and MA-XRF data but with the compound specificity of MA-XRPD. The general aim is to expedite non-destructive artwork imaging analysis by fusing data from different imaging modalities via machine learning models. The effect of preprocessing techniques to enhance the predictive accuracy of the models is explored. The paper demonstrates the method's efficacy on a 16th-century illuminated manuscript, showcasing the feasibility of predicting compound-specific distribution maps. Three evaluation methods—visual examination of the predicted distribution, root mean square errors (RMSE), and feature permutation importance (FPI)—are employed to assess model performance. Fusing MA-XRF with MA-XRPD led to the best RMSE scores overall. However, fusing the RIS and MA-XRPD data blocks also yield very satisfactory and easily interpretable high-resolution compound maps. SignificanceWhile MA-XRPD allows for highly specific imaging of artworks, its time-consuming nature and limited resolution presents a bottleneck during non-invasive imaging of painted works of art. By integrating data from more time-efficient hyperspectral techniques such as MA-XRF and RIS, and employing machine learning, we expedite the process without compromising accuracy. The fusion process can also denoise the distribution maps, improving their readability for heritage professionals and art historical scholars.

Predicting non-chemotherapy drug-induced agranulocytosis toxicity through ensemble machine learning approaches.

Agranulocytosis, induced by non-chemotherapy drugs, is a serious medical condition that presents a formidable challenge in predictive toxicology due to its idiosyncratic nature and complex mechanisms. In this study, we assembled a dataset of 759 compounds and applied a rigorous feature selection process prior to employing ensemble machine learning classifiers to forecast non-chemotherapy drug-induced agranulocytosis (NCDIA) toxicity. The balanced bagging classifier combined with a gradient boosting decision tree (BBC + GBDT), utilizing the combined descriptor set of DS and RDKit comprising 237 features, emerged as the top-performing model, with an external validation AUC of 0.9164, ACC of 83.55%, and MCC of 0.6095. The model's predictive reliability was further substantiated by an applicability domain analysis. Feature importance, assessed through permutation importance within the BBC + GBDT model, highlighted key molecular properties that significantly influence NCDIA toxicity. Additionally, 16 structural alerts identified by SARpy software further revealed potential molecular signatures associated with toxicity, enriching our understanding of the underlying mechanisms. We also applied the constructed models to assess the NCDIA toxicity of novel drugs approved by FDA. This study advances predictive toxicology by providing a framework to assess and mitigate agranulocytosis risks, ensuring the safety of pharmaceutical development and facilitating post-market surveillance of new drugs.

Development and validation of novel interpretable survival prediction models based on drug exposures for severe heart failure during vulnerable period

BackgroundSevere heart failure (HF) has a higher mortality during vulnerable period while targeted predictive tools, especially based on drug exposures, to accurately assess its prognoses remain largely unexplored. Therefore, this study aimed to utilize drug information as the main predictor to develop and validate survival models for severe HF patients during this period.MethodsWe extracted severe HF patients from the MIMIC-IV database (as training and internal validation cohorts) as well as from the MIMIC-III database and local hospital (as external validation cohorts). Three algorithms, including Cox proportional hazards model (CoxPH), random survival forest (RSF), and deep learning survival prediction (DeepSurv), were applied to incorporate the parameters (partial hospitalization information and exposure durations of drugs) for constructing survival prediction models. The model performance was assessed mainly using area under the receiver operator characteristic curve (AUC), brier score (BS), and decision curve analysis (DCA). The model interpretability was determined by the permutation importance and Shapley additive explanations values.ResultsA total of 11,590 patients were included in this study. Among the 3 models, the CoxPH model ultimately included 10 variables, while RSF and DeepSurv models incorporated 24 variables, respectively. All of the 3 models achieved respectable performance metrics while the DeepSurv model exhibited the highest AUC values and relatively lower BS among these models. The DCA also verified that the DeepSurv model had the best clinical practicality.ConclusionsThe survival prediction tools established in this study can be applied to severe HF patients during vulnerable period by mainly inputting drug treatment duration, thus contributing to optimal clinical decisions prospectively.

Modeling the richness and spatial distribution of the wild relatives of Iranian pears (Pyrus L.) for conservation management

The preservation of the genetic resources of crop wild relatives (CWRs) is crucial for food production systems and is considered a vital measure for global agricultural health and food security. The identification of potential areas where CWRs can thrive is one of the first steps towards their conservation. In this study, we used a maximum entropy model (MaxEnt) to determine the habitat suitability of seven wild relatives of pears (Pyrus L.) for the first time. We aimed to identify high-priority areas for conservation and determine the hotspots for rich biodiversity in Iran. The study showed excellent predictive performance for all species studied (AUC value ≥ 90). The soil depth, solar radiation, minimum temperature of the coldest month (Bio6), and precipitation of the wettest quarter (Bio16) were the main environmental factors that influenced the habitat suitability of all seven species, according to permutation importance. The projected maps revealed that P. elaeagnifolia had the largest suitable habitat area, while P. glabra had the lowest. The results also showed that less than 5% of the suitable habitats for these seven species were in protected areas. This research highlights the need for national preservation policies and the development of cultivation and rehabilitation strategies for these threatened species.

Soil temperature prediction based on explainable artificial intelligence and LSTM

Soil temperature is a key parameter in many disciplines, and its research has important practical significance. In recent years, the prediction of soil temperature by deep learning has achieved good results. However, deep learning is difficult to popularize in practical use because of its opacity. This study aims to interpret and analyze the Long Short Term Memory Network (LSTM) model for global soil temperature prediction using SHapley Additive exPlanation (SHAP), Permutation Importance (PI) and Partial Dependence Plot (PDP). The results show that Temperature of air at 2 m above the surface of land has the greatest influence on the prediction of soil temperature, and its SHAP and PI characteristic values have significant seasonality. Meanwhile, radiation also has a certain influence on the prediction results. There was a significant positive correlation between the temperature of 2 m and the soil temperature. The explanatory insights provided in this paper enhance the transparency and confidence of the model, which promotes the applicability of soil temperature prediction models in relevant fields.

Machine learning-enabled prediction of prolonged length of stay in hospital after surgery for tuberculosis spondylitis patients with unbalanced data: a novel approach using explainable artificial intelligence (XAI)

BackgroundTuberculosis spondylitis (TS), commonly known as Pott’s disease, is a severe type of skeletal tuberculosis that typically requires surgical treatment. However, this treatment option has led to an increase in healthcare costs due to prolonged hospital stays (PLOS). Therefore, identifying risk factors associated with extended PLOS is necessary. In this research, we intended to develop an interpretable machine learning model that could predict extended PLOS, which can provide valuable insights for treatments and a web-based application was implemented.MethodsWe obtained patient data from the spine surgery department at our hospital. Extended postoperative length of stay (PLOS) refers to a hospitalization duration equal to or exceeding the 75th percentile following spine surgery. To identify relevant variables, we employed several approaches, such as the least absolute shrinkage and selection operator (LASSO), recursive feature elimination (RFE) based on support vector machine classification (SVC), correlation analysis, and permutation importance value. Several models using implemented and some of them are ensembled using soft voting techniques. Models were constructed using grid search with nested cross-validation. The performance of each algorithm was assessed through various metrics, including the AUC value (area under the curve of receiver operating characteristics) and the Brier Score. Model interpretation involved utilizing methods such as Shapley additive explanations (SHAP), the Gini Impurity Index, permutation importance, and local interpretable model-agnostic explanations (LIME). Furthermore, to facilitate the practical application of the model, a web-based interface was developed and deployed.ResultsThe study included a cohort of 580 patients and 11 features include (CRP, transfusions, infusion volume, blood loss, X-ray bone bridge, X-ray osteophyte, CT-vertebral destruction, CT-paravertebral abscess, MRI-paravertebral abscess, MRI-epidural abscess, postoperative drainage) were selected. Most of the classifiers showed better performance, where the XGBoost model has a higher AUC value (0.86) and lower Brier Score (0.126). The XGBoost model was chosen as the optimal model. The results obtained from the calibration and decision curve analysis (DCA) plots demonstrate that XGBoost has achieved promising performance. After conducting tenfold cross-validation, the XGBoost model demonstrated a mean AUC of 0.85 ± 0.09. SHAP and LIME were used to display the variables’ contributions to the predicted value. The stacked bar plots indicated that infusion volume was the primary contributor, as determined by Gini, permutation importance (PFI), and the LIME algorithm.ConclusionsOur methods not only effectively predicted extended PLOS but also identified risk factors that can be utilized for future treatments. The XGBoost model developed in this study is easily accessible through the deployed web application and can aid in clinical research.