Important Predictor Variables Research Articles

Predicting urban cold-air paths using boosted regression trees

Ensuring the transport of colder air masses from rural surroundings or inner urban open spaces into the dense urban center is a vital approach for reducing nocturnal warming of urban areas in summer, i.e. the urban heat island (UHI) effect. Hence, cold-air paths are an important mitigation aspect in urban planning. The identification of cold-air paths by field measurements, however, is time-consuming and it is difficult to cover larger spatial domains (e.g. entire cities). Therefore, we used a machine learning technique, boosted regression trees (BRT), for estimating the spatial distribution of cold-air paths in three German cities (Braunschweig, Freiburg, Stuttgart). We identified the most important predictor variables for cold-air path occurrence and tested the spatial transferability of BRT models from one training city to the other study cities.Three variables explaining differences in surface elevation indicated strongest influence on spatial distribution of cold-air paths: relative surface elevation, topographic position index, and topographic wind index. We achieved predictions for the spatial distribution of cold-air paths for cities the model was trained on, resulting in model performances of 0.85 < AUC < 0.96 (AUC: area under the receiver operating characteristic curve). Model transfers to other cities, however, achieved only poor to moderate results (0.5 < AUC < 0.8). The city of Freiburg tended to be the most suitable basis for a general model, since it covers the largest range of conditions. Transfer works best for ‘similar’ cities, i.e. cities with a comparable range of predictor variables and morphological structure. Cold-air path identification via BRT modelling appeared to be a promising method delivering helpful information for urban planning.

Soil organic carbon in peninsular Spain: Influence of environmental factors and spatial distribution

Soil organic carbon (SOC) stocks and their geographical distribution in peninsular Spain were estimated from a georeferenced database consisting of 12,724 surface samples (0–30 cm) and 3607 subsurface samples (30–50 cm), covering different climate, land use, elevation, parent material, soil type and soil pH. SOC density showed a high heterogeneity, with the lowest values in arid regions, where the average in topsoil ranged between 20 and 60 t C ha−1, under woody crops and forest respectively. Carbon stocks gradually increases as precipitation increases, and its variability is also dependent of other factors, fundamentally the presence/absence of active lime or active Al. In semi-arid zones, calcaric soils (pH ≈ 8.3) have higher contents of SOC than neutral to weakly acidic soils from siliceous materials. However, in humid regions, calcareous materials have undergo total or partial decarbonatation in the upper layer (pH < 4.0–7.5) and SOC stocks are markedly lower than in other materials. In forest soils it seems that a steady state (around 100–120 t C ha−1) (0–30 cm) has been reached in a wide range of precipitation, between 900 and 1700 mm; most of this carbon (about 80%) is labile-C. Soils from granitic rocks are acidic (pH 4.5–5.5) (Al buffering) and the mean SOC stock in the indicated precipitation range is between 170 and 200 t ha−1 (it is estimated that approximately 60% is stabilized as metal-C or mineral-C complexes). The highest values (190–240 t ha−1) are recorded in acidic soils derived from mafic rocks, which in these regions usually develop andic properties (around 73% is involved in stable metal-C or mineral C complexes). Finally, the SOC stored in neutral soils from serpentinized ultramafic rocks (without excess Ca or Al) is similar to that of the decarbonated soils derived from calcareous materials. In all regions, forest soils are a much more important SOC sink than live forest biomass (2–4 times higher in the upper 30 cm and 3–6 times greater in the upper 50 cm).Random Forest regression was used as modeling tool and digital mapping. Mean annual precipitation was estimated to be the most important predictor variable, followed by land use, lithology/soil type and soil pH. Model performance was calibrated by the internal RF validation and through cross-validation, and the results were similar. In topsoil, the mean error, root mean square error and R2 were −0.007% C, 1.48% C and 0.61, respectively. In the subsurface layer these indices were −0.020, 1.07 and 0.37, respectively. SOC stock for peninsular Spain was estimated at 3.33 Pg in the upper (0–30 cm) layer, and 0.85 Pg in the subsurface (30–50 cm) layer. Total SOC stock for 0–50 cm was 4.19 Pg, with a 95% confidence interval ranging between 3.33 and 5.03 Pg.

Biogeographical analyses to facilitate targeted conservation of orchid diversity hotspots in Costa Rica

AbstractAimWe conduct a biogeographical assessment of orchids in a global biodiversity hotspot to explore their distribution and occurrences of local hotspots while identifying geographic attributes underpinning diversity patterns. We evaluate habitat characteristics associated with orchid diversity hotspots and make comparisons to other centres of orchid diversity to test for global trends. The ultimate goal was to identify an overall set of parameters that effectively characterize critical habitats to target in local and global orchid conservation efforts.LocationCosta Rica; Mesoamerica.TaxonOrchidaceae.MethodsData from an extensive set of herbarium records were used to map orchid distributions and to identify diversity hotspots. Hotspot data were combined with geographic attribute data, including environmental and geopolitical variables, and a random forest regression model was utilized to assess the importance of each variable for explaining the distribution of orchid hotspots. A likelihood model was created based on variable importance to identify locations where suitable habitats and unidentified orchid hotspots might occur.ResultsOrchids were widely distributed and hotspots occurred primarily in mountainous regions, but occasionally at lower elevations. Precipitation and vegetation cover were the most important predictive variables associated with orchid hotspots. Variable values underpinning Costa Rican orchid hotspots were similar to those reported at other sites worldwide. Models also identified suitable habitats for sustaining orchid diversity that occurred outside of known hotspots and protected areas.Main conclusionsSeveral orchid diversity hotspots and potentially suitable habitats occur outside of known distributions and/or protected areas. Recognition of these sites and their associated geographic attributes provides clear targets for optimizing orchid conservation efforts in Costa Rica, although certain caveats warrant consideration. Habitats linked with orchid hotspots in Costa Rica were similar to those documented elsewhere, suggesting the existence of a common biogeographical trend regarding critical habitats for orchid conservation in disparate tropical regions.

Forecasting Mood in Bipolar Disorder From Smartphone Self-assessments: Hierarchical Bayesian Approach.

BackgroundBipolar disorder is a prevalent mental health condition that is imposing significant burden on society. Accurate forecasting of symptom scores can be used to improve disease monitoring, enable early intervention, and eventually help prevent costly hospitalizations. Although several studies have examined the use of smartphone data to detect mood, only few studies deal with forecasting mood for one or more days.ObjectiveThis study aimed to examine the feasibility of forecasting daily subjective mood scores based on daily self-assessments collected from patients with bipolar disorder via a smartphone-based system in a randomized clinical trial.MethodsWe applied hierarchical Bayesian regression models, a multi-task learning method, to account for individual differences and forecast mood for up to seven days based on 15,975 smartphone self-assessments from 84 patients with bipolar disorder participating in a randomized clinical trial. We reported the results of two time-series cross-validation 1-day forecast experiments corresponding to two different real-world scenarios and compared the outcomes with commonly used baseline methods. We then applied the best model to evaluate a 7-day forecast.ResultsThe best performing model used a history of 4 days of self-assessment to predict future mood scores with historical mood being the most important predictor variable. The proposed hierarchical Bayesian regression model outperformed pooled and separate models in a 1-day forecast time-series cross-validation experiment and achieved the predicted metrics, R2=0.51 and root mean squared error of 0.32, for mood scores on a scale of −3 to 3. When increasing the forecast horizon, forecast errors also increased and the forecast regressed toward the mean of data distribution.ConclusionsOur proposed method can forecast mood for several days with low error compared with common baseline methods. The applicability of a mood forecast in the clinical treatment of bipolar disorder has also been discussed.

Analisis Pengaruh Profitabilitas, Solvabilitas, Dan Nilai Pasar Terhadap Return Saham Pada Perusahaan Go Publik Di Bursa Efek Indonesia (Studi Pada Perusahaan Jasa Sub Sektor Restoran, Hotel Dan Pariwisata)

Financial ratios are employed in this research to measure the influence of profitability, solvency and price to book value on stock return in restaurant, hotel and tourism companies which are listed on Indonesia Stock Exchange from period 2014 – 2018. The objectives of this study was to provide empirical findings whether profitability, solvency and price to book value have a significant influence on stock return. Multiple linear regression analysis was used to identify the strength of the effect of Return On Assets (ROA), Return On Equity (ROE), Debt to Equity Ratio (DER) and Price to Book Value (PBV) on stock return simultaneously and partially.The result of this study indicates that simultaneously ROA, ROE, DER and PBV have a significant effect on stock return and there is a positive significant effect of return on assets on stock return. Whereas price to book value has a negative significant effect on stock return in restaurant, hotel and tourism companies listed on IDX. Furthermore, return on equity and debt to equity ratio have no significant impact on stock return. Among the predictor variables, the data analysis showed that return on assets is the most important predictor variables in this regression model.

Drivers of fish assemblage structures in a Neotropical urban watershed

Urbanization results in major changes in stream morphology, hydrology, and water quality; the consequences on fish assemblage of Neotropical urban streams remain poorly understood. Due to the importance of better understanding fish assemblage responses towards urbanization in biodiversity hotspots, we evaluated the factors structuring the fish assemblages in a Neotropical medium-sized city. Additionally, we evaluated possible differences in fish diversity and biotic homogenization between the urban and control streams. To do so, fish were sampled along with physical habitat and water quality parameters in 16 urban streams in southern Brazil. Multiple regression and distance-based linear models (DistLM) were used to assess the importance of predictor variables in determining species richness and fish assemblage composition, respectively. The fish abundance rank curves were plotted to evaluate graphic differences in fish diversity between urban and control streams, while the relationship between distance and similarity (Bray-Curtis index) was used to evaluate biotic homogenization. The results revealed the importance of variables linking habitat diversity and basin connectivity to fish assemblage richness and structure. In addition, riparian vegetation was the main driver of fish assemblages, explaining 78% of abundance variation. We identified lower diversity and higher biotic homogenization in urban streams, with higher dominance of the non-native species Poecilia reticulata. Our results highlight the importance of evaluating the physical and biotic variables of streams outside of water pollution levels alone. Finally, it is necessary to support urban planning practices that recover riparian vegetation, avoid canalization, and promote connectivity along stream networks as well as the removal of artificial structures in urban basins.

Challenges and Opportunities for Terrapene carolina carolina Under Different Climate Scenarios

An unprecedented rate of global climate change as a result of human impacts has affected both endotherms and ectotherms. This is of special concern for ectotherms, such as reptiles, as these species are suffering from large population declines and lack the dispersal ability of other taxa. There are many protected areas across the United States; however, these areas are fragmented, which hinders dispersal. We examined species distribution and dispersal capabilities for Terrapene carolina carolina, a relatively narrow range, low dispersal, and vulnerable species. We created climatic suitability models to predict changes in suitable habitat and identified important predictor variables. We modeled three time periods using MaxEnt and hypothesized that there would be an increase in northern habitat. We found that most of the suitable habitat changed at the northern end of the range and that mean temperature of driest quarter had the most influence on future predictions. Overall there were relatively moderate changes in suitable habitat, but where these changes occur affects accessibility. As an example, we examined these local scale movements within Oak Openings Region and found that individuals are capable of dispersing to new suitable habitats; however, other physical barriers will hinder movements. In conclusion, there is a critical need to protect this vulnerable reptilian species and our results suggest that T. c. carolina will expand their distribution northward. We suggest that land managers increase connectivity among protected areas to facilitate dispersal, but future studies should incorporate other dynamic ecological factors at finer spatial scale.

Risk Factors for Failure in Hammertoe Surgery.

Hammertoe correction is perhaps the most common elective surgery performed in the foot, yet rates of symptomatic recurrence and revision surgery can be high. In this study, we aimed to identify patient and provider risk factors associated with failure after hammertoe surgery. Consecutive patients with a minimum of 6 months' follow-up undergoing hammertoe surgery within a single, urban foot and ankle practice between January 1, 2011, and December 31, 2013, served as the basis of this retrospective cohort study. Cox regression analysis was used to identify important predictor variables obtained through chart and radiographic review. One hundred fifty-two patients (311 toes) with a mean age of 60.8 ± 11.2 years and mean follow-up of 29.5 ± 21.2 months were included. Statistically significant predictors of failure were having a larger preoperative transverse plane deviation of the digit (hazard ratio [HR], 1.03 for each degree; P < .001; 95% CI, 1.02, 1.04), operating on the second toe (vs third or fourth) (HR, 2.23; P = .003; 95% CI, 1.31, 3.81), use of a phalangeal osteotomy to reduce the proximal interphalangeal (PIP) joint (HR, 2.77; P = .005; 95% CI, 1.36, 5.64), and using less common/conventional operative techniques to reduce the PIP joint (HR, 2.62; P = .03; 95% CI, 1.09, 6.26). Concomitant performance of first ray surgery reduced hammertoe recurrence by 50% (HR, 0.51; P = .01; 95% CI, 0.30, 0.87). We identified risk factors that may provide guidance for surgeons during preoperative hammertoe surgery consultations. This information may better equip patients with appropriate postoperative expectations when contemplating surgery. Level III, retrospective case series.

Bayesian Function-on-Scalars Regression for High-Dimensional Data

We develop a fully Bayesian framework for function-on-scalars regression with many predictors. The functional data response is modeled nonparametrically using unknown basis functions, which produces a flexible and data-adaptive functional basis. We incorporate shrinkage priors that effectively remove unimportant scalar covariates from the model and reduce sensitivity to the number of (unknown) basis functions. For variable selection in functional regression, we propose a decision theoretical posterior summarization technique, which identifies a subset of covariates that retains nearly the predictive accuracy of the full model. Our approach is broadly applicable for Bayesian functional regression models, and unlike existing methods provides joint rather than marginal selection of important predictor variables. Computationally scalable posterior inference is achieved using a Gibbs sampler with linear time complexity in the number of predictors. The resulting algorithm is empirically faster than existing frequentist and Bayesian techniques, and provides joint estimation of model parameters, prediction and imputation of functional trajectories, and uncertainty quantification via the posterior distribution. A simulation study demonstrates improvements in estimation accuracy, uncertainty quantification, and variable selection relative to existing alternatives. The methodology is applied to actigraphy data to investigate the association between intraday physical activity and responses to a sleep questionnaire. Supplementary materials for this article are available online.

Investigating the effects of exposure to extremely low frequency electromagnetic fields on job burnout syndrome and the severity of depression; the role of oxidative stress.

ObjectivesThis study was designed to investigate the possible effect of exposure to extremely low frequency electromagnetic fields (ELF‐EMFs) on occupational burnout syndrome and the severity of depression experienced among thermal power plant workers and the role of oxidative stress.MethodsIn this cross‐sectional study, 115 power plant workers and 124 administrative personnel of a hospital were enrolled as exposed and unexposed groups, respectively, based on inclusion and exclusion criteria. Levels of oxidative stress biomarkers, including malondialdehyde (MDA), superoxide dismutase (SOD), catalase (Cat), and total antioxidant capacity were measured in serum samples. Exposure to electric and magnetic fields was measured using the IEEE Std C95.3.1 standard at each workstation. The burnout syndrome and the severity of depression were assessed using the Maslach Burnout and Beck Depression Inventory.ResultsThe levels of MDA and SOD were significantly lower in the exposed group than the unexposed group. The exposed group reported a higher prevalence of burnout syndrome and higher depression severity. Multiple linear regression showed that work experience, MDA level, and levels of exposure to magnetic fields are the most important predictor variables for burnout syndrome and severity of depression. In addition, a decrease in the level of Cat was significantly associated with increased burnout syndrome.ConclusionThe thermal power plant workers exposed to ELF‐EMFs are at risk of burnout syndrome and depression. These effects may be caused directly by exposure to magnetic fields or indirectly due to increased oxidative stress indices.

The Predictive Effect of Some Variables on Fifth and Sixth Grade Students’ Scientific Process Skills

The aim of this research was to investigate whether there was a predictive effect of the frequency of laboratory use of students, their academic success and attending various opportunities to learn science outside the classroom on the fifth and sixth grade students’ scientific process skills performance (SPSP). Also, it was investigated whether there was a significant relationship between the students’ SPSP and the grade level in which the concepts included in the test items were appropriate. The quantitative research methods were used in this study. Data were collected from 458 fifth and sixth grade students. Regression analysis was conducted to determine the effects of the frequency of laboratory use of students, their academic success and attending various opportunities to learn science outside the classroom on students’ SPSP. As a result, it was determined that students’ academic success was important predictor variables that affect the fifth and sixth grade students’ SPSP. It was also determined that the frequency of laboratory use of students had an important predictive effect on students’ causal scientific process skills performance (C-SPSP). Additionally, it was determined that various opportunities to learn science outside the classroom such as attending in science fairs and reading scientific journals had an important predictive effect on students’ SPSP. Although, it was also determined that the students’ SPSP differs according to the grade level, statistically significant relationship was not found between the students’ SPSP and the grade level in which the concepts included in the test items were appropriate.

Soils of the Brazilian Coastal Plains biome: prediction of chemical attributes via portable X-ray fluorescence (pXRF) spectrometry and robust prediction models

Portable X-ray fluorescence (pXRF) spectrometry has been successfully used for soil attribute prediction. However, recent studies have shown that accurate predictions may vary according to soil type and environmental conditions, motivating investigations in different biomes. Hence, this work attempted to accurately predict soil pH, sum of bases (SB), cation exchange capacity (CEC) at pH 7.0 and base saturation (BS) using pXRF-obtained data with high variability and robust prediction models in the Brazilian Coastal Plains biome. A total of 285 soil samples were collected to generate prediction models for A (n = 123), B (n = 162) and A+B (n = 285) horizons through stepwise multiple linear regression, support vector machine with linear kernel (SVM) and random forest. Data were divided into calibration (75%) and validation (25%) sets. Accuracy of the predictions was assessed by coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE) and residual prediction deviation (RPD). The A+B horizons dataset had optimal performance, especially for SB predictions using SVM, achieving R2 = 0.82, RMSE = 1.02 cmolc dm–3, MAE = 1.17 cmolc dm–3 and RPD = 2.33. The most important predictor variable was Ca. Predictions using pXRF data were accurate especially for SB. Limitations of the predictions caused by soil classes and environmental conditions should be further investigated in other regions.

Impacts of Occupational Cognitive Failure and Subjective Workload on Patient Safety Incidents among Intensive Care Units Nurses.

Background:Life-saving treatments and high-quality care techniques increase the opportunity for patient safety incidents in Intensive care unit.Aims:This descriptive correlation study aimed to determine the impacts of occupational cognitive failure and subjective workload on patient safety incidents among intensive care units nurses.Methods and Material:One hundred seventy-six nurses working in intensive care units were included using census sampling. The data collection tools consisted of demographic and occupational data, standard questionnaires of subjective workload (NASA-TLX) and occupational cognitive failure (OCFQ), and a question about frequency of patient safety incidents. Data analysis was performed using Mann–Whitney and Kruskal–Wallis, Spearman rank correlation coefficient, and logistic regression tests.Results:Occupational cognitive failure (OR = 1.043), subjective workload in dimension of “performance” (OR = 0.982), age (OR = 0.947), and gender (OR = 3.726) were important predictive variables of patient safety incidents.Conclusions:Nursing mangers and policymakers can consider the factors identified for staffing nurses and development of patient safety programs.

The predictive role of forgiveness and the level of repetitive thinking on mental health and marital adjustment in married individuals

Objective: A review of the literature reveals that while forgiveness, the importance attached to negative experiences to be forgiven, and repetitive thinking are associated with marital adjustment, the role of variables regarding close relationships has so far received little attention from researchers. As the marital relationship, one of the most important interpersonal relations, correlates to mental health, this study aims to examine jointly the predictive power of relevant variables for psychological symptoms and marital adjustment and additionally to investigate possible gender differences in terms of these research variables. Method: The sample of this study consisted of 157 married individuals (86 women and 71 men). Marital Adjustment Test (MAT), Brief Symptom Inventory (BSI), Forgiving Personality Scale (FP), Repetitive Thinking Questionnaire (RTQ), Importance Attached to the Incident to be Forgiven Scale (IAIFF), and a Personal Information Form developed by the researchers were administered to married persons. Correlation analysis was conducted to examine the relations between variables; subsequently, hierarchical regression analysis was applied to determine the power of the variables to predict psychological symptoms and marital adjustment. One-way MANOVA was used to reveal gender differences with regard to the research variables. Results: Regression analysis showed that gender, repetitive thinking, forgiving personality, and the importance given to negative experiences expected to be forgiven predict psychological symptoms. In addition, gender, forgiving personality, and repetitive thinking have a significant predictive value for marital adjustment. Finally, gender differences were found to be significant in marital adjustment, psychological symptoms, repetitive thinking, and the importance attached to negative experiences to be forgiven. It was determined that marital adjustment was lower in women while their psychological symptoms were higher. Conclusion: This study demonstrates that repetitive thinking and forgiveness are important predictor variables for both mental health and marital adjustment. It also suggests that marital therapists should consider relevant variables in psychotherapy when working with married couples.

Modelling Distributions of Rove Beetles in Mountainous Areas Using Remote Sensing Data

Mountain ecosystems are biodiversity hotspots that are increasingly threatened by climate and land use/land cover changes. Long-term biodiversity monitoring programs provide unique insights into resulting adverse impacts on plant and animal species distribution. Species distribution models (SDMs) in combination with satellite remote sensing (SRS) data offer the opportunity to analyze shifts of species distributions in response to these changes in a spatially explicit way. Here, we predicted the presence probability of three different rove beetles in a mountainous protected area (Gran Paradiso National Park, GPNP) using environmental variables derived from Landsat and Aster Global Digital Elevation Model data and an ensemble modelling approach based on five different model algorithms (maximum entropy, random forest, generalized boosting models, generalized additive models, and generalized linear models). The objectives of the study were (1) to evaluate the potential of SRS data for predicting the presence of species dependent on local-scale environmental parameters at two different time periods, (2) to analyze shifts in species distributions between the years, and (3) to identify the most important species-specific SRS predictor variables. All ensemble models showed area under curve (AUC) of the receiver operating characteristics values above 0.7 and true skills statistics (TSS) values above 0.4, highlighting the great potential of SRS data. While only a small proportion of the total area was predicted as highly suitable for each species, our results suggest an increase of suitable habitat over time for the species Platydracus stercorarius and Ocypus ophthalmicus, and an opposite trend for Dinothenarus fossor. Vegetation cover was the most important predictor variable in the majority of the SDMs across all three study species. To better account for intra- and inter-annual variability of population dynamics as well as environmental conditions, a continuation of the monitoring program in GPNP as well as the employment of SRS with higher spatial and temporal resolution is recommended.

Multifactorial analysis of factors influencing elite australian football match outcomes: a machine learning approach

Abstract In Australian football (AF), few studies have assessed combinations of pre- game factors and their relation to game outcomes (win/loss) in multivariable analyses. Further, previous research has mostly been confined to association-based linear approaches and post-game prediction, with limited assessment of predictive machine learning (ML) models in a pre-game setting. Therefore, our aim was to use ML techniques to predict game outcomes and produce a hierarchy of important (win/loss) variables. A total of 152 variables (79 absolute and 73 differentials) were used from the 2013–2018 Australian Football League (AFL) seasons. Various ML models were trained (cross-validation) on the 2013–2017 seasons with the–2018 season used as an independent test set. Model performance varied (66.5-73.3% test set accuracy), although the best model (glmnet – 73.3%) rivalled bookmaker predictions in the same period (70.9%). The glmnet model revealed measures of team quality (a player-based rating and a team-based) in their relative form as the most important variables for prediction. Models that contained in-built feature selection or could model non-linear relationships generally performed better. These findings show that AFL game outcomes can be predicted using ML methods and provide a hierarchy of predictors that maximize the chance of winning.

Multivariate Adaptive Generalized Poisson Regression Spline (MAGPRS) on the number of acute respiratory infection infants

Acute Respiratory Infection (ARI) is an infectious disease of the respiratory tract that affects the structure of the respiratory tract. The ARI is a health problem that should not be ignored because it causes high infant mortality. Therefore, it is important to know the factors that influence the ARI. Generalized Poisson Regression (GPR) is one of method that can handle cases of overdispersion (variance is greater than the mean) or underdispersion (variance is less than the mean). Multivariate Adaptive Regression Spline as a statistical method for fitting the relationship between a set of input variables and dependent variables. This research is the development of the MARS method and GPR namely MAGPRS. The application of the MAGPRS model was carried out in the case of the number of ARI in infants from surabaya health departement 2017. The results showed that the importance of predictor variables in MAGPRS, the variables affecting the number of ARI patients in infants are the percentage of low birth weight (X2), the percentage of unhealthy houses (X5), and the percentage given non-exclusive breastfeeding to infants (X1).

Machine Learning Classification Algorithms to Predict aGvHD following Allo-HSCT: A Systematic Review.

The acute graft-versus-host disease (aGvHD) is the most important cause of mortality in patients receiving allogeneic hematopoietic stem cell transplantation. Given that it occurs at the stage of severe tissue damage, its diagnosis is late. With the advancement of machine learning (ML), promising real-time models to predict aGvHD have emerged. This article aims to synthesize the literature on ML classification algorithms for predicting aGvHD, highlighting algorithms and important predictor variables used. A systemic review of ML classification algorithms used to predict aGvHD was performed using a search of the PubMed, Embase, Web of Science, Scopus, Springer, and IEEE Xplore databases undertaken up to April 2019 based on Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statements. The studies with a focus on using the ML classification algorithms in the process of predicting of aGvHD were considered. After applying the inclusion and exclusion criteria, 14 studies were selected for evaluation. The results of the current analysis showed that the algorithms used were Artificial Neural Network (79%), Support Vector Machine (50%), Naive Bayes (43%), k-Nearest Neighbors (29%), Regression (29%), and Decision Trees (14%), respectively. Also, many predictor variables have been used in these studies so that we have divided them into more abstract categories, including biomarkers, demographics, infections, clinical, genes, transplants, drugs, and other variables. Each of these ML algorithms has a particular characteristic and different proposed predictors. Therefore, it seems these ML algorithms have a high potential for predicting aGvHD if the process of modeling is performed correctly.

Decision tree algorithm in locally advanced rectal cancer: an example of over-interpretation and misuse of a machine learning approach.

To analyse the classification performances of a decision tree method applied to predictor variables in survival outcome in patients with locally advanced rectal cancer (LARC). The aim was to offer a critical analysis to better apply tree-based approach in clinical practice and improve its interpretation. Data concerning patients with histological proven LARC between 2007 and 2014 were reviewed. All patients were treated with trimodality approach with a curative intent. The Kaplan-Meier method was used to estimate overall survival (OS). Decision tree methods were was used to select important variables in outcome prediction. A total of 100 patients were included. The 5-year and 7-year OS rates were 76.4% and 71.3%, respectively. Age, co-morbidities, tumor size, clinical tumor classification (cT) and clinical nodes classification (cN) were the important predictor variables to the tree's construction. Overall, 13 distinct groups of patients were defined. Patients aged < 65years with cT3 disease and elderly patients with a tumor size < 5cm seemed to have highest rates of survival. But the process over-fitted the data, leading to poor algorithm performance. We proposed a decision tree algorithm to identify known and new pre-treatment clinical predictors of survival in LARC. Our analysis confirmed that tree-based machine learning method, especially classification trees, can be easily interpreted even by a non-expert in the field, but controlling cross validation errors is mandatory to capture its statistical power. However, it is necessary to carefully analyze the classification error trend to chose the important predictor variables, especially in little data. Machine learning approach should be considered the new unexplored frontier in LARC. Based on big datasets, decision trees represent an opportunity to improve decision-making process in clinical practice.

Pressure drops of fresh cemented paste backfills through coupled test loop experiments and machine learning techniques

In this paper, test loop experiments and machine learning techniques were combined to investigate pressure drops of fresh cemented paste backfill (CPB) mixes. The influence of tailings characteristics on CPB pressure drops was studied and Extensive test loop experiments were performed. The complex mapping from tailings characteristics, cement-tailings ratio, inlet velocity and solids content to pressure drop was successfully learned by decision tree regression (DTR) models. In addition, the influence of training set size and maximum tree depth on DTR performance was investigated. The relative importance of predictor variables was discussed and the visualisation for a representative DTR model was provided. Finally, the current research applied two ensemble techniques, namely random forest and gradient boosting regression tree, to increase the predictive performance of the DTR models. The study found that all ensemble techniques outperformed DTR in the pressure drop prediction of CPB.