Permutation Importance Research Articles

A Novel Epilepsy Detection Method Based on Feature Extraction by Deep Autoencoder on EEG Signal.

Electroencephalogram (EEG) signals are the gold standard tool for detecting epileptic seizures. Long-term EEG signal monitoring is a promising method to realize real-time and automatic epilepsy detection with the assistance of computer-aided techniques and the Internet of Medical Things (IoMT) devices. Machine learning (ML) algorithms combined with advanced feature extraction methods have been widely explored to precisely recognize EEG signals, while among which, little attention has been paid to high computing costs and severe information losses. The lack of model interpretability also impedes the wider application and deeper understanding of ML methods in epilepsy detection. In this research, a novel feature extraction method based on an autoencoder (AE) is proposed in the time domain. The architecture and mechanism are elaborated. In this method, specified features are defined and calculated on the basis of signal reconstruction quantification of the AE. The EEG recognition is performed to validate the effectiveness of the proposed detection method, and the prediction accuracy reached 97%. To further investigate the superiority of the proposed AE-based feature extraction method, a widely used feature extraction method, PCA, is allocated for comparison. In order to understand the underlying working mechanism, permutation importance and SHapley Additive exPlanations (SHAP) are conducted for model interpretability, and the results further confirm the reasonability and effectiveness of the extracted features by AE reconstruction. With high computing efficiency in the time domain and an extensively satisfactory accuracy, the proposed epilepsy detection method exhibits great superiority and potential in almost real-time and automatic epilepsy monitoring.

Near perfect classification of cardiac biomarker Troponin-I in human serum assisted by SnS2-CNT composite, explainable ML, and operating-voltage-selection-algorithm

The high worldwide mortality and disproportionate impact of cardiovascular diseases have emerged as the most significant global health burden, unfortunately, unmet by the traditional detection methods. Therefore, developing a rapid, sensitive, selective, and rugged biosensor for the precise classification/quantification of cardiac biomarkers is a stepping stone for the future generation of cardiac healthcare. We demonstrate a facile, time-efficient, and scalable biosensor for classifying the FDI approved gold standard cardiac biomarker Troponin-I (cTnI) in untreated human serum matrix, built-on 2-D SnS2 and 1-D MWCNT composite transducer and decision-tree based explainable machine learning (ML) algorithm. The proposed methodology is further enhanced using an inimitable Operating-Voltage-Selection-Algorithm (OVSA), which boosts ML accuracy to ∼100%. The near-perfect classification is realized by strategically incorporating this two-step algorithm-first the OVSA, then the heuristic and ML approaches on the selected dataset. Dynamic concentrations of the biomarker (100 fg/mL to 100 ng/mL) are estimated with high sensitivity, ∼71 (ΔR/R) (ng/mL)−1cm−2 and low limit of detection (0.02 fg/mL), aiding to the prediction and prognosis of acute myocardial infarction. The hyperparameter tuning and feature engineering improve the decision process of the ML algorithm, fostering robustness against data variability. Feature importance indices, namely the Gini index, Permutation Importance, and SHAP values, portray ‘Voltage’ as the most important feature, further justifying our insight into the OVSA. The biosensor's specificity, selectivity, reproducibility and stability are effectively demonstrated with the sampling to result reporting time of just 20 min, establishing it as a potential candidate for clinical testing.

Interpretable machine learning to identify important predictors of birth weight: A prospective cohort study.

Predicting birth weight and identifying its risk factors are clinically important. This study aims to use interpretable machine learning to predict birth weight and identity important predictors. This prospective cohort study was conducted in Tongzhou Maternal and Child Health Care Hospital of Beijing, China, recruiting pregnant women between June 2018 and February 2019. We used 24 features to predict infant birth weight, including gestational age, mother's age, parity, history of macrosomia delivery, pre-pregnancy body mass index (BMI), height, father's BMI, lifestyle (diet, physical activity, smoking), and biomarker (fasting glucose and lipids) features. Study outcome was birth weight of infant. We used 8 supervised learning models including 4 individual [linear regression, ridge regression, lasso regression, support vector machines regression (SVR)], and 4 ensemble estimators (random forest, AdaBoost, gradient boosted trees, and voting ensemble for regression) to predict birth weight. Model accuracy was measured by root mean squared error (RMSE) of 10-fold cross validation on the training set and RMSE of prediction on the test set. We used permutation importance algorithm to understand the prediction from the models and what affected them. This study included 4,754 mother-child dyads. RMSEs were lower in voting ensemble for regression, linear regression, and SVR than random forest, AdaBoost, and gradient boosted tree. The 5 most important predictors for infant birth weight were gestational age, fetal sex, preterm birth, mother's height, and pre-pregnancy BMI. After adding ultrasound-measured indicators of fetal growth into predictors, mother's height and pre-pregnancy BMI remained the most important predictors in predicting the outcome. Mother's height and pre-pregnancy BMI were identified as important predictors for infant birth weight. Interpretable machine learning is a promising tool in the prediction of birth weight.

Abstract 15753: Comparative Transcriptomics Analysis of Histone (De)acetylases and Cardiac Ion Channels in Human Induced Pluripotent Stem-Cell-Derived Cardiomyocytes vs. the Adult Human Heart

Epigenetic regulation is critical for cardiac electrophysiology and pathology. Epigenetic modulators, such as histone deacetylases (HDACs) and histone acetyltransferases (HATs) are known master regulators of gene expression. Recently, novel pharmacological agents, HDAC inhibitors, have been developed as treatments for cancer and immune diseases. The effects of HDAC inhibitors on cardiac ion channels (ICs) are of great interest. To exert specific gene modulation, we used small interfering RNAs against the known HDACs, including sirtuins, and deployed them in human induced pluripotent stem-cell-derived cardiomyocytes (hiPSC-CMs). Follow-up RNAseq data (n = 61) were compared to identically processed and normalized RNAseq data from human left ventricle (LV) from the GTEx database (n = 84). Gene expression of cardiac ICs displayed similar patterns, with some differences. For example, hiPSC-CMs showed upregulated CACNA1C, SLC8A1 and downregulated KCNJ2 and RYR2 compared to the adult LV, most of which are known distinctions (Fig. 1A). Correlative analysis (Fig. 1B) and partial least square regression models helped visualize links between HDACs/HATs, key transcription factors (TFs) and cardiac ICs. Powerful TFs, including MEF2A, GATA4, 6 exerted positive effect on ICs in hiPSC-CM and the adult LV. In the hiPSC-CMs, HDAC1, HDAC10 and SIRT6 were found to be the strongest predictors of the expression of individual cardiac ICs, as revealed by permutation importance. Further studies will involve determination of the role of different cell types using single-cell sequencing data from the adult LV. Our analysis offers new insights about the role of epigenetic modifiers on cardiac electrophysiology and informs the utility of hiPSC-CM as a scalable, experimental model for cardiotoxicity testing of HDAC inhibitors.

Climate Change Influences the Spread of African Swine Fever Virus.

Climate change is an inevitable and urgent issue in the current world. African swine fever virus (ASFV) is a re-emerging viral animal disease. This study investigates the quantitative association between climate change and the potential spread of ASFV to a global extent. ASFV in wild boar outbreak locations recorded from 1 January 2019 to 29 July 2022 were sampled and investigated using the ecological distribution tool, the Maxent model, with WorldClim bioclimatic data as the predictor variables. The future impacts of climate change on ASFV distribution based on the model were scoped with Representative Concentration Pathways (RCP 2.6, 4.5, 6.0, and 8.5) scenarios of Coupled Model Intercomparison Project 5 (CMIP5) bioclimatic data for 2050 and 2070. The results show that precipitation of the driest month (Bio14) was the highest contributor, and annual mean temperature (Bio1) was obtained as the highest permutation importance variable on the spread of ASFV. Based on the analyzed scenarios, we found that the future climate is favourable for ASFV disease; only quantitative ratios are different and directly associated with climate change. The current study could be a reference material for wildlife health management, climate change issues, and World Health Organization sustainability goal 13: climate action.

Predicting and optimizing syngas production from fluidized bed biomass gasifiers: A machine learning approach

Biomass gasification is one of the primary thermal conversion processes where fluidized bed reactors are often used to produce syngas with low heating values. However, there has not yet been an effective model to predict gasification yield with broad applicability. In this study, machine learning was adopted to realize the prediction of syngas compositions and lower heating values (LHV) using various lignocellulosic biomass feedstocks at a wide range of operating conditions. Three machine learning techniques, i.e., Random Forest (RF), Support Vector Machine (SVM) and Artificial Neural Network (ANN) were adopted after determining hyperparameters optimization. Pearson correlation and permutation importance were used for the sensitivity analysis. RF and ANN were found to have high prediction accuracy with R2 and RMSE results (RF: R2=0.809–0.946, RMSE=1.39–11.54%; ANN: R2=0.565–0.924, RMSE=1.46–10.56%). Monte Carlo filtering (MCF) was integrated into the three machine learning algorithms to forecast the desired products by predicting the important features of the operating conditions and biomass characteristics. Considering the desired H2/CO > 1.1 and LHV > 5.86 MJ/m3, the RF-MCF was a more suitable approach with R2=0.791–0.902 for H2, CO and LHV features. The machine learning approach can be widely adapted in various scenarios predicting output features as well as MCF for finding the significant variables for optimization.

Sentinel-2 Images Based Modeling of Grassland Above-Ground Biomass Using Random Forest Algorithm: A Case Study on the Tibetan Plateau

Accurate information on grassland above-ground biomass (AGB) is critical to better understanding the carbon cycle and conserve grassland resources. As a climate-sensitive key ecological function area, it is important to accurately estimate the grassland AGB of the Tibetan Plateau. Sentinel-2 (S2) images have advantages in reducing mixed pixels and the scale effect for remote sensing, while the data volume is correspondingly larger. In order to improve the estimation accuracy while reducing the data volume required for AGB estimation and improving the computational efficiency, this study used the Recursive Feature Elimination (RFE) algorithm to find the optimal feature set and compared the performance of the Cubist, Gradient Boosting Regression Tree (GBRT), random forest (RF) and eXtreme Gradient Boosting (XGBoost) algorithms for estimating AGB. In this study, ten S2 bands, ten S2-derived vegetation indexes, 218 pieces of AGB field survey data, four types of meteorological data and three types of topographic data were used as the alternative input features for the AGB estimation model. The impurity and permutation importance were used as the feature importance calculation method input to the RFE, and the Cubist, GBRT, RF and XGBoost algorithms were used to construct the AGB estimation models. The results showed that the RF algorithm based on the monthly average temperature (T), elevation, Normalized Difference Phenology Index (NDPI), Normalized Difference Infrared Index (NDII) and Palmer Drought Severity Index (PDSI) performed best (R2 = 0.8838, RMSE = 35.05 g/m2, LCCC = 2.44, RPPD = 0.91). The above findings suggest that the RF model based on the features related to temperature, altitude, humidity and leaf water content is beneficial to estimate the grassland AGB on the Tibetan Plateau.

Predicting suitable habitat for the endangered plant Cephalotaxus oliveri Mast. in China

SummaryThe coniferous shrub Oliver’s plum yew (Cephalotaxus oliveri) is endemic to southern China with potential medicinal use for cancer treatment and ecological value in sustaining China’s threatened subtropical forest ecosystems. Comprehensive understanding of the current spatial patterns of this vulnerable species vis-à-vis climatic conditions is crucial for its sustained economic use and conservation. Based on 100 reliable occurrence records and nine environmental variables, MaxEnt and QGIS programs were used to predict the potential geographical distribution of C. oliveri in China. Combined with percentage contribution and permutation importance, the jackknife statistical method was used to test and evaluate pertinent factors restricting the potential distribution of C. oliveri. The response curves of critical bioclimatic factors were employed to determine the potential species range. The current core potential distribution areas were concentrated in China’s central and south-west regions. Temperature was identified as the crucial determinant of species distribution patterns, particularly the mean temperature of the coldest quarter. Precipitation was a necessary but not critical secondary factor. These findings should inform the ex situ conservation and cultivation of C. oliveri in China and its introduction to other parts of the world for similar purposes.

Benchmark study of feature selection strategies for multi-omics data

BackgroundIn the last few years, multi-omics data, that is, datasets containing different types of high-dimensional molecular variables for the same samples, have become increasingly available. To date, several comparison studies focused on feature selection methods for omics data, but to our knowledge, none compared these methods for the special case of multi-omics data. Given that these data have specific structures that differentiate them from single-omics data, it is unclear whether different feature selection strategies may be optimal for such data. In this paper, using 15 cancer multi-omics datasets we compared four filter methods, two embedded methods, and two wrapper methods with respect to their performance in the prediction of a binary outcome in several situations that may affect the prediction results. As classifiers, we used support vector machines and random forests. The methods were compared using repeated fivefold cross-validation. The accuracy, the AUC, and the Brier score served as performance metrics.ResultsThe results suggested that, first, the chosen number of selected features affects the predictive performance for many feature selection methods but not all. Second, whether the features were selected by data type or from all data types concurrently did not considerably affect the predictive performance, but for some methods, concurrent selection took more time. Third, regardless of which performance measure was considered, the feature selection methods mRMR, the permutation importance of random forests, and the Lasso tended to outperform the other considered methods. Here, mRMR and the permutation importance of random forests already delivered strong predictive performance when considering only a few selected features. Finally, the wrapper methods were computationally much more expensive than the filter and embedded methods.ConclusionsWe recommend the permutation importance of random forests and the filter method mRMR for feature selection using multi-omics data, where, however, mRMR is considerably more computationally costly.

PUMD: a PU learning-based malicious domain detection framework

Domain name system (DNS), as one of the most critical internet infrastructure, has been abused by various cyber attacks. Current malicious domain detection capabilities are limited by insufficient credible label information, severe class imbalance, and incompact distribution of domain samples in different malicious activities. This paper proposes a malicious domain detection framework named PUMD, which innovatively introduces Positive and Unlabeled (PU) learning solution to solve the problem of insufficient label information, adopts customized sample weight to improve the impact of class imbalance, and effectively constructs evidence features based on resource overlapping to reduce the intra-class distance of malicious samples. Besides, a feature selection strategy based on permutation importance and binning is proposed to screen the most informative detection features. Finally, we conduct experiments on the open source real DNS traffic dataset provided by QI-ANXIN Technology Group to evaluate the PUMD framework’s ability to capture potential command and control (C&C) domains for malicious activities. The experimental results prove that PUMD can achieve the best detection performance under different label frequencies and class imbalance ratios.

Predicting mechanical properties of carbon nanotube-reinforced cementitious nanocomposites using interpretable ensemble learning models

Compressive and flexural strength are important characteristics that indicate the efficiency of utilizing carbon nanotube (CNT) in cementitious nanocomposites. Preparing numerous test samples and conducting mechanical tests at different ages is costly and time-consuming, so a predictive model appears essential. Due to the complexity of recent building materials, experimental and statistical models do not function well for such complex materials, so machine learning techniques were adopted in this study. From the results of experiments performed in the literature, separate comprehensive data sets for compressive and flexural strength were collected, and water to cement ratio (W/C), CNT type, CNT content, CNT length, CNT diameter, sand to cement ratio for mortars (S/C), surfactant type, dispersion method, curing days (age) and compressive or flexural strengths of the control sample (C0 or F0) considered as input variables. Four ensemble learning-based algorithms, including random forest, AdaBoost, Gradient Boost, and XGBoost, were developed alongside a standalone model, the decision tree. To assess the sensitivity of input variables, four methodologies, including the Gini importance, permutation importance, F score, and mean of relative Shapely additive explanation (SHAP) values, were obtained, and compressive or flexural strengths of the control sample and CNT content identified overall as the most influential variables. The results revealed that XGBoostprovides more reliable and accurate results for compressive and flexural characteristics and better maps the relationships between input variables. In addition, SHAP analysis was performed to explain the contribution of each input according to its value on the output of XGBoost.

Using machine learning regression models to predict the pellet quality of pelleted feeds

Pelleted feeds are widely used in monogastric animal production systems because they not only improve animal performance (increasing digestibility and feed consumption) but are convenient to store and handle. However, pellet quality can be affected by many factors. While previous studies have reported the effect of a single or several factors on pellet quality, no studies have investigated how pellet quality can be affected by the large number of factors that vary during feed manufacturing. Therefore, the current study reports using machine learning regression models to predict pellet quality using commercial feed mill data. A dataset consisting of 2471 observations describing the pellet manufacturing process, the feed formulation, and environmental conditions (e.g. outdoor temperature) were collected from two feed mill lines for 8 months. Sixteen features (13 continuous, 3 categorical) were used for building the regression models, and the output was the pellet durability index (PDI) of the pelleted feeds. Twelve regression algorithms including Linear Regression (LR), Least Absolute Shrinkage and Selection Operator (LASSO) regression, Ridge Regression (RR), Support Vector Regression (SVR), Linear Support Vector Regression (LSVR), Random Forest (RF), Decision Tree (DT), Gradient Boosting Regression (GBR), Adaptive Boosting Regression (ABR), Multi-Layer Perceptron (MLP) neural network, K-Nearest Neighbor (KNN), and Stacking Regression (SR) were examined in this study. Feature importance analysis using permutation importance was performed to identify what features were most relevant for each model. Average outdoor temperature, bakery byproduct and wheat inclusion levels, as well as production line all had high permutation importance values, while the fat added into the mixer (with controls at the mill already in place to limit it) was less important than most features. The cleaned dataset was preprocessed and then split into a training (80 % of total samples, n = 1147) and a testing (20 % of total samples, n = 287) set. A 5-fold cross-validation process was applied and learning curves were used to verify the presence of overfitting for each algorithm before and after tuning the hyperparameters on the training set. The models that exhibited overfitting were excluded from the final results and only models with tuned hyperparameters were evaluated on the testing set. The SVR algorithm was selected as the best overall model for predicting PDI, as it had the lowest mean absolute and mean squared prediction errors (MAE = 3.280, MSPE = 16.192), and the second highest concordance correlation coefficient (CCC = 0.636). In conclusion, this study shows that feed mill features describing manufacturing parameters, feed formulation, and environmental data can be successfully used to build machine learning regression models for pellet quality prediction.

Developing and validating a scoring system for measuring frailty in patients with hip fracture: a novel model for predicting short-term postoperative mortality

ObjectivesFrailty is common among patients with hip fracture and may, in part, contribute to the increased risk of mortality and morbidity after hip fracture surgery. This study aimed to develop...

Explainable machine learning improves interpretability in the predictive modeling of biological stream conditions in the Chesapeake Bay Watershed, USA

Anthropogenic alterations have resulted in widespread degradation of stream conditions. To aid in stream restoration and management, baseline estimates of conditions and improved explanation of factors driving their degradation are needed. We used random forests to model biological conditions using a benthic macroinvertebrate index of biotic integrity for small, non-tidal streams (upstream area ≤200 km2) in the Chesapeake Bay watershed (CBW) of the mid-Atlantic coast of North America. We utilized several global and local model interpretation tools to improve average and site-specific model inferences, respectively. The model was used to predict condition for 95,867 individual catchments for eight periods (2001, 2004, 2006, 2008, 2011, 2013, 2016, 2019). Predicted conditions were classified as Poor, FairGood, or Uncertain to align with management needs and individual reach lengths and catchment areas were summed by condition class for the CBW for each period. Global permutation and local Shapley importance values indicated percent of forest, development, and agriculture in upstream catchments had strong impacts on predictions. Development and agriculture negatively influenced stream condition for model average (partial dependence [PD] and accumulated local effect [ALE] plots) and local (individual condition expectation and Shapley value plots) levels. Friedman's H-statistic indicated large overall interactions for these three land covers, and bivariate global plots (PD and ALE) supported interactions among agriculture and development. Total stream length and catchment area predicted in FairGood conditions decreased then increased over the 19-years (length/area: 66.6/65.4% in 2001, 66.3/65.2% in 2011, and 66.6/65.4% in 2019). Examination of individual catchment predictions between 2001 and 2019 showed those predicted to have the largest decreases in condition had large increases in development; whereas catchments predicted to exhibit the largest increases in condition showed moderate increases in forest cover. Use of global and local interpretative methods together with watershed-wide and individual catchment predictions support conservation practitioners that need to identify widespread and localized patterns, especially acknowledging that management actions typically take place at individual-reach scales.

Analyzing Suicide Risk From Linguistic Features in Social Media: Evaluation Study.

BackgroundEffective suicide risk assessments and interventions are vital for suicide prevention. Although assessing such risks is best done by health care professionals, people experiencing suicidal ideation may not seek help. Hence, machine learning (ML) and computational linguistics can provide analytical tools for understanding and analyzing risks. This, therefore, facilitates suicide intervention and prevention.ObjectiveThis study aims to explore, using statistical analyses and ML, whether computerized language analysis could be applied to assess and better understand a person’s suicide risk on social media.MethodsWe used the University of Maryland Suicidality Dataset comprising text posts written by users (N=866) of mental health–related forums on Reddit. Each user was classified with a suicide risk rating (no, low, moderate, or severe) by either medical experts or crowdsourced annotators, denoting their estimated likelihood of dying by suicide. In language analysis, the Linguistic Inquiry and Word Count lexicon assessed sentiment, thinking styles, and part of speech, whereas readability was explored using the TextStat library. The Mann-Whitney U test identified differences between at-risk (low, moderate, and severe risk) and no-risk users. Meanwhile, the Kruskal-Wallis test and Spearman correlation coefficient were used for granular analysis between risk levels and to identify redundancy, respectively. In the ML experiments, gradient boost, random forest, and support vector machine models were trained using 10-fold cross validation. The area under the receiver operator curve and F1-score were the primary measures. Finally, permutation importance uncovered the features that contributed the most to each model’s decision-making.ResultsStatistically significant differences (P<.05) were identified between the at-risk (671/866, 77.5%) and no-risk groups (195/866, 22.5%). This was true for both the crowd- and expert-annotated samples. Overall, at-risk users had higher median values for most variables (authenticity, first-person pronouns, and negation), with a notable exception of clout, which indicated that at-risk users were less likely to engage in social posturing. A high positive correlation (ρ>0.84) was present between the part of speech variables, which implied redundancy and demonstrated the utility of aggregate features. All ML models performed similarly in their area under the curve (0.66-0.68); however, the random forest and gradient boost models were noticeably better in their F1-score (0.65 and 0.62) than the support vector machine (0.52). The features that contributed the most to the ML models were authenticity, clout, and negative emotions.ConclusionsIn summary, our statistical analyses found linguistic features associated with suicide risk, such as social posturing (eg, authenticity and clout), first-person singular pronouns, and negation. This increased our understanding of the behavioral and thought patterns of social media users and provided insights into the mechanisms behind ML models. We also demonstrated the applicative potential of ML in assisting health care professionals to assess and manage individuals experiencing suicide risk.

Machine learning approach for predicting and evaluating California bearing ratio of stabilized soil containing industrial waste

The California bearing ratio (CBR) is one of the important indexes, which is used to represent the strength of subgrade or subbases of pavement. In general, the CBR can be determined through experiments both in the laboratory and field. However, the determination of the CBR is time and cost-consuming as well as low accuracy due to the disturbance of samples and limitations of preparation in the laboratory. Thus, this study estimates the CBR of stabilized soil using twelve machine learning techniques (6 single models and 6 hybrid models). The single models include artificial neural network (ANN), gradient boosting (GB), extreme gradient boosting (XGB), random forest (RF), support vector machine (SVM), and K-nearest neighbors (KNN), while the six hybrid models are a combination of these single models and random restart hill-climbing optimization (RRHC). Twelve models are constructed based on eleven input variables, including cement, Atterberg's limits, optimum moisture content (OMC), maximum dry density (MDD), and dust and ashes. To evaluate the performance of the proposed models, four popular statistical indexes namely mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and determination coefficient (R2) were used. The results indicate that among twelve models, four models using GB, RRHC_XGB, RRHC_RF, and RF had a high prediction accuracy (R2 > 0.98) and outperformed than other remaining models. Among these four models, the model using RF has the highest prediction accuracy (R2 = 0.9817, RMSE = 2.3970, MAE = 1.1682, MAPE = 0.0666). According to the result of feature importance analysis using Sklearn permutation importance, SHapley Additive exPlanation (SHAP), individual conditional expectation (ICE), and partial dependence plots-2D, cement and plasticity index (PI) are two most important variables affecting the CBR prediction of stabilized soil. Where, PI was found to be a very crucial factor, which improved the prediction ability of the models compared to the results of previous studies. The results also reveal that when cement content is larger than 2%, there is an insignificant influence of the cement on the CBR of stabilized soil. The value of PI smaller than 15% has a vital impact on the CBR for any values of cement, dust, and ashes. Furthermore, the results also indicate that dust and ash content have less effect on the CBR of stabilized soil. In summary, it can be concluded that this study provides an insightful assessment of the CBR prediction of stabilized soil, and the results of this study can fill the gap in the literature and provide practical knowledge and application on the CBR of stabilized soil.

Modeling the distribution of the endangered Jemez Mountains salamander (Plethodon neomexicanus) in relation to geology, topography, and climate.

The Jemez Mountains salamander (Plethodon neomexicanus; hereafter JMS) is an endangered salamander restricted to the Jemez Mountains in north‐central New Mexico, United States. This strictly terrestrial and lungless species requires moist surface conditions for activities such as mating and foraging. Threats to its current habitat include fire suppression and ensuing severe fires, changes in forest composition, habitat fragmentation, and climate change. Forest composition changes resulting from reduced fire frequency and increased tree density suggest that its current aboveground habitat does not mirror its historically successful habitat regime. However, because of its limited habitat area and underground behavior, we hypothesized that geology and topography might play a significant role in the current distribution of the salamander. We modeled the distribution of the JMS using a machine learning algorithm to assess how geology, topography, and climate variables influence its distribution. The best habitat suitability model indicates that geology type and maximum winter temperature (November to March) were most important in predicting the distribution of the salamander (23.5% and 50.3% permutation importance, respectively). Minimum winter temperature was also an important variable (21.4%), suggesting this also plays a role in salamander habitat. Our habitat suitability map reveals low uncertainty in model predictions, and we found slight discrepancies between the designated critical habitat and the most suitable areas for the JMS. Because geological features are important to its distribution, we recommend that geological and topographical data are considered, both during survey design and in the description of localities of JMS records once detected.

Predictive Analysis of Errors During Robot-Mediated Gamified Training.

This paper presents our approach to predicting future error-related events in a robot-mediated gamified physical training activity for stroke patients. The ability to predict future error under such conditions suggests the existence of distinguishable features and separated class characteristics between the casual gameplay state and error prune state in the data. Identifying such features provides valuable insight to creating individually tailored, adaptive games as well as possible ways to increase rehabilitation success by patients. Considering the time-series nature of sensory data created by motor actions of patients we employed a predictive analysis strategy on carefully engineered features of sequenced data. We split the data into fixed time windows and explored logistic regression models, decision trees, and recurrent neural networks to predict the likelihood of a patient making an error based on the features from the time window before the error. We achieved an 84.4% F1-score with a 0.76 ROC value in our best model for predicting motion accuracy related errors. Moreover, we computed the permutation importance of the features to explain which ones are more indicative of future errors.

Environmental variable importance for under-five mortality in Malaysia: A random forest approach

BackgroundEnvironmental factors have been associated with adverse health effects in epidemiological studies. The main exposure variable is usually determined via prior knowledge or statistical methods. It may be challenging when evidence is scarce to support prior knowledge, or to address collinearity issues using statistical methods. This study aimed to investigate the importance level of environmental variables for the under-five mortality in Malaysia via random forest approach. MethodWe applied a conditional permutation importance via a random forest (CPI-RF) approach to evaluate the relative importance of the weather- and air pollution-related environmental factors on daily under-five mortality in Malaysia. This study spanned from January 1, 2014 to December 31, 2016. In data preparation, deviation mortality counts were derived through a generalized additive model, adjusting for long-term trend and seasonality. Analyses were conducted considering mortality causes (all-cause, natural-cause, or external-cause) and data structures (continuous, categorical, or all types [i.e., include all variables of continuous type and all variables of categorical type]). The main analysis comprised of two stages. In Stage 1, Boruta selection was applied for preliminary screening to remove highly unimportant variables. In Stage 2, the retained variables from Boruta were used in the CPI-RF analysis. The final importance value was obtained as an average value from a 10-fold cross-validation. ResultSome heat-related variables (maximum temperature, heat wave), temperature variability, and haze-related variables (PM10, PM10-derived haze index, PM10- and fire-derived haze index, fire hotspot) were among the prominent variables associated with under-five mortality in Malaysia. The important variables were consistent for all- and natural-cause mortality and sensitivity analyses. However, different most important variables were observed between natural- and external-cause under-five mortality. ConclusionHeat-related variables, temperature variability, and haze-related variables were consistently prominent for all- and natural-cause under-five mortalities, but not for external-cause.

Machine Learning Analysis of Handgun Transactions to Predict Firearm Suicide Risk

Evidence suggests that limiting access to firearms among individuals at high risk of suicide can be an effective means of suicide prevention, yet accurately identifying those at risk to intervene remains a key challenge. Firearm purchasing records may offer a large-scale and objective data source for the development of tools to predict firearm suicide risk. To test whether a statewide database of handgun transaction records, coupled with machine learning techniques, can be used to forecast firearm suicide risk. This prognostic study used the California database of 4 976 391 handgun transaction records from 1 951 006 individuals from January 1, 1996, to October 6, 2015. Transaction-level random forest classification was implemented to predict firearm suicide risk, and the relative predictive power of features in the algorithm was estimated via permutation importance. Analyses were performed from December 1, 2020, to May 19, 2022. The main outcome was firearm suicide within 1 year of a firearm transaction, derived from California death records (1996-2016). With the use of California's Dealer's Records of Sale (1996-2015), 41 handgun, transaction, purchaser, and community-level predictor variables were generated. There are a total of 4 976 391 transactions in the California's Dealer's Record of Sale database representing 1 951 006 individuals (1 525 754 men [78.2% of individuals]; mean [SD] age, 43.4 [13.9] years). Firearm suicide within 1 year occurred in 0.07% of handgun transactions (3278 transactions among 2614 individuals). A total of 38.6% of observed firearm suicides were among transactions classified in the highest-risk ventile (379 of 983 transactions), with 95% specificity. Among the small number of transactions with a random forest score above 0.95, more than two-thirds (24 of 35 [68.6%]) were associated with a purchaser who died by firearm suicide within 1 year. Important features included known risk factors, such as older age at first purchase, and previously unreported predictors, including distance to firearms dealer and month of purchase. This prognostic study presented the first large-scale machine learning analysis of individual-level handgun transaction records. The results suggested the potential utility of such records in identifying high-risk individuals to aid suicide prevention efforts. It also identified handgun, individual, and community characteristics that have strong predictive relationships with firearm suicide and may warrant further study.