Variable Importance Research Articles

A competing risks machine learning study of neutron dose, fractionation, age, and sex effects on mortality in 21,000 mice

This study explores the impact of densely-ionizing radiation on non-cancer and cancer diseases, focusing on dose, fractionation, age, and sex effects. Using historical mortality data from approximately 21,000 mice exposed to fission neutrons, we employed random survival forest (RSF), a powerful machine learning algorithm accommodating nonlinear dependencies and interactions, treating cancer and non-cancer outcomes as competing risks. Unlike traditional parametric models, RSF avoids strict assumptions and captures complex data relationships through decision tree ensembles. SHAP (SHapley Additive exPlanations) values and variable importance scores were employed for interpretation. The findings revealed clear dose–response trends, with cancer being the predominant cause of mortality. SHAP value dose–response shapes differed, showing saturation for cancer hazard at high doses (> 2 Gy) and a more linear pattern at lower doses. Non-cancer responses remained more linear throughout the entire dose range. There was a potential inverse dose rate effect for cancer, while the evidence for non-cancer was less conclusive. Sex and age effects were less pronounced. This investigation, utilizing machine learning, enhances our understanding of the patterns of non-cancer and cancer mortality induced by densely-ionizing radiations, emphasizing the importance of such approaches in radiation research, including space travel and radioprotection.

Screening for Early Biomarkers of Cisplatin-Induced Acute Kidney Injury in Rats Through Serum Metabolomics Technology.

To systematically identify early biomarkers of cisplatin-induced acute kidney injury (AKI) in rats. An experimental study. Place and Duration of the Study: Experimental Animal Laboratory of Lanzhou University, Gansu, China, and the Department of Pharmacy, The First Hospital of Lanzhou University, Gansu, China, from July 2022 to October 2023. In this study, an AKI model was established by continuously injecting cisplatin into rats at a dose of 1 mg/kg once a day for control group and for 2, 3, 4, and 5 days to other four groups, respectively. Subsequently, rat plasma samples were collected for metabolomics analysis to identify early differentiated metabolites in the plasma prior to creatinine elevation. Furthermore, accurate HPLC-MS/MS methods were developed to validate the biomarker variation in other AKI models. The occurrence of time-dependent renal cortical injury and significant alterations of creatinine (Cr) concentration were observed on day-4 and 5, which demonstrated successful model construction. Sixty-six compounds changed on Day-2 while 61 compounds changed on Day-3. Eleven compounds with variable importance in projection (VIP) >1.5 and false discover rate (FDR) <0.2 were selected and identified by HPLC-MS/MS. Among these, N-acetylglutamine and citramalic acid changed earlier than serum creatinine (sCr) in the AKI model. N-acetylglutamine and citramalic acid may serve as early biomarker of cisplatin-induced AKI. Acute kidney injury, Biomarker, Cisplatin, Metabolomics, LC-MS/MS, Rats.

Metabolic Disturbances in a Mouse Model of MPTP/Probenecid-Induced Parkinson's Disease: Evaluation Using Liquid Chromatography-Mass Spectrometry.

Parkinson's disease (PD) is a common neurodegenerative disease that severely affects patients' daily lives and places a significant burden on the global economy. There are currently no specific biomarkers for distinguishing between the different stages of PD. We divided 78 mice into six equal groups, including five model PD groups (W1-W5; based on the PD stage induced by length of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine/propofol induction time) and a control group. Then, we used metabolomics technology to detect the serum small-molecule metabolites present in each group. Ultimately, we screened for potential biomarkers using the variable importance in the projection of the orthogonal partial least squares discriminant analysis and the coefficient value of LASSO ordinal logistic regression. We identified 12 potential biomarkers, including dehydroepiandrosterone sulfate, pipecolic acid, N-acetylleucine, 2-aminoadipic acid, L-tyrosine, uric acid, and 5-hydroxyindoleacetaldehyde. Pathway analysis revealed their involvement in amino acid metabolism, caffeine metabolism, steroid hormone biosynthesis, and purine metabolism. Additionally, the receiver operating characteristic curve indicated that a biomarker panel comprising the 12 biomarkers could differentiate between the different PD stages. Different PD stages are characterized by different metabolites. The biomarkers identified in this study are helpful to understand the PD process.

Seasonality of Subsurface Shear Instabilities at Tropical Instability Wave Fronts in the Atlantic Ocean in a High‐Resolution Simulation

AbstractTropical Instability Waves (TIWs) have been shown to modulate upper ocean mixing. However, previous studies on the modulation of TIW related mixing are based on small numbers of TIWs and have not considered temporal variability, which can lead to discrepancies in the findings. In this study, using a 12‐year simulation carried out with a comprehensive, global, high‐resolution ocean model, we present evidence of seasonally modulated shear instabilities at TIW fronts in the Atlantic Ocean that reach down to the thermocline depth, potentially inducing mixing below the mixed layer depth. We find that, regardless of whether TIWs are present earlier in the year, frontal instabilities and potential mixing primarily occur in boreal summer, coinciding with a vertical shear maximum between the mean zonal currents. We argue that in the Atlantic Ocean, vertical shear at TIW fronts does usually not suffice to cause frontal instabilities below the mixed layer depth. Instead, the background shear needs to be sufficiently large in addition to TIW shear, to overcome the stability. The background shear in turn varies seasonally and is strongly driven by the variability of the northern branch of the South Equatorial Current (nSEC). As such, the variability of the nSEC strongly contributes to the generation and modulation of instabilities at TIW fronts that reach below the mixed layer depth and have the potential to induce mixing. Our results highlight the importance of seasonal variability when studying TIW impacts and their effect on mixing.

Predicting intraocular lens tilt using a machine learning concept.

To use a combination of partial least squares regression and a machine learning approach to predict intraocular lens (IOL) tilt using preoperative biometry data. Kepler University Clinic Linz, Linz, Austria. Prospective single-center study. Optical coherence tomography, autorefraction, and subjective refraction were performed at baseline and 8 weeks after cataract surgery. In analysis I, only 1 eye per patient was included and a tilt prediction model was generated. In analysis II, a pairwise comparison between right and left eyes was performed. In analysis I, 50 eyes of 50 patients were analyzed. Difference in amount, orientation, and vector from preoperative to postoperative lens tilt was -0.13 degrees, 2.14 degrees, and 1.20 degrees, respectively. A high predictive power (variable importance for projection [VIP]) for postoperative tilt prediction was found for preoperative tilt (VIP = 2.2), pupil decentration (VIP = 1.5), lens thickness (VIP = 1.1), axial eye length (VIP = 0.9), and preoperative lens decentration (VIP = 0.8). These variables were applied to a machine learning algorithm resulting in an out of bag score of 0.92 degrees. In analysis II, 76 eyes of 38 patients were included. The difference of preoperative to postoperative IOL tilt of right and left eyes of the same individual was statistically relevant. Postoperative IOL tilt showed excellent predictability using preoperative biometry data and a combination of partial least squares regression and a machine learning algorithm. Preoperative lens tilt, pupil decentration, lens thickness, axial eye length, and preoperative lens decentration were found to be the most relevant parameters for this prediction model.

Comparison of the TNM9th and 8th editions for localized and locally advanced anal squamous cell carcinoma treated nonsurgically and proposal of a new stage grouping system.

To compare the survival discrimination of the TNM9th and 8th editions for localized and locally advanced anal squamous cell carcinoma (ASCC) treated nonsurgically and suggest a simple revised staging system with data from the Surveillance, Epidemiology, and End Results (SEER) database. Overall survival (OS) was the primary endpoint. Survival comparisons between the T and N stages and the different staging systems were performed using the Kaplan-Meier method and log-rank test, followed by correlation analysis and variable importance analysis (VIA). Additionally, multivariate analysis was employed to identify significant predictors, which were further visualized using a nomogram. Finally, calibration curve, C-index, and decision curve analysis (DCA) were applied to assess the performance of the different staging systems. A total of 5384 patients with ASCC were analyzed, revealing superior discrimination OS by the TNM9th edition compared to that by the TNM8th edition. Multivariate analysis identified the T and N stages as significant OS predictors (all p < 0.001). However, ambiguity persisted in stage III subgroups within the TNM9th edition, showing OS times of 102 months for stage IIIA disease, 88 months for stage IIIB disease, and 128 months for stage IIIC disease (all p > 0.05). Correlation analysis demonstrated an increased correlation for the T stage between the TNM8th and 9th editions (ρ value from 0.7 to 0.89), while the N stage correlation decreased (ρ value from 0.84 to 0.56). VIA and the prognostic nomogram highlighted the greater importance of the T stage over the N stage. Based on these findings, a new staging system was developed, and its clinical utility was confirmed through calibration curves, C-index values (from 0.598 to 0.604), and DCAs. Our new staging system exhibited slightly better prognostic value compared to the TNM9th staging systems for nonmetastatic ASCC and warrants further validation.

Predicting Postoperative Satisfaction with Breasts: How Important is the Preoperative BREAST-Q Score?

The role that preoperative Satisfaction with Breast plays in a patient's postoperative course after postmastectomy breast reconstruction (PMBR) is not understood. The aim of this study is to understand the impact of the preoperative score on postoperative outcome as an independent variable. We examined patients who underwent PMBR between 2017 and 2021 and who completed the BREAST-Q Satisfaction with Breasts at 1year postoperatively. Two multiple linear regression models (Model 1 with the preoperative Satisfaction with Breasts score and Model 2 without the preoperative score), likelihood ratio tests, simple t-statistics, and sample patient dataset to predict the 1year score were performed. Multiple imputation was used to account for missing preoperative scores. Overall, 2324 patients were included. Model 1 showed that the preoperative score is significantly associated with the postoperative score (β = 0.09, 95% confidence interval 0.04-0.14; p < 0.001). Comparing Model 1 and Model 2 demonstrated that including preoperative Satisfaction with Breasts in a regression significantly improves model fit (test statistic = 10.04; p = 0.0021). Using the absolute value of the t-statistics as a measure of variable importance in linear regression, the importance of the preoperative score was quantified as 3.39-more important than neoadjuvant radiation, mastectomy weight, body mass index, bilateral prophylactic mastectomy, and race, but less than adjuvant radiation, reconstruction type, and psychiatric diagnoses. Preoperative Satisfaction with Breasts scores are an important independent predictor of postoperative satisfaction after PMBR. Just as vital sign and work-up are carefully documented before surgery, preoperative scores should be collected to pre-emptively gauge patients' satisfaction and optimize postoperative outcomes.

A data-driven approach to identifying PFAS water sampling priorities in Colorado, United States.

Per and polyfluoroalkyl substances (PFAS), a class of environmentally and biologically persistent chemicals, have been used across many industries since the middle of the 20th century. Some PFAS have been linked to adverse health effects. Our objective was to incorporate known and potential PFAS sources, physical characteristics of the environment, and existing PFAS water sampling results into a PFAS risk prediction map that may be used to develop a PFAS water sampling prioritization plan for the Colorado Department of Public Health and Environment (CDPHE). We used random forest classification to develop a predictive surface of potential groundwater contamination from two PFAS, perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA). The model predicted PFAS risk at locations without sampling data into one of three risk categories after being "trained" with existing PFAS water sampling data. We used prediction results, variable importance ranking, and population characteristics to develop recommendations for sampling prioritization. Sensitivity and precision ranged from 58% to 90% in the final models, depending on the risk category. The model and prioritization approach identified private wells in specific census blocks, as well as schools, mobile home parks, and public water systems that rely on groundwater as priority sampling locations. We also identified data gaps including areas of the state with limited sampling and potential source types that need further investigation. This work uses random forest classification to predict the risk of groundwater contamination from two per- and polyfluoroalkyl substances (PFAS) across the state of Colorado, United States. We developed the prediction model using data on known and potential PFAS sources and physical characteristics of the environment, and "trained" the model using existing PFAS water sampling results. This data-driven approach identifies opportunities for PFAS water sampling prioritization as well as information gaps that, if filled, could improve model predictions. This work provides decision-makers information to effectively use limited resources towards protection of populations most susceptible to the impacts of PFAS exposure.

The Absolute Age of NGC 3201 Derived from Detached Eclipsing Binaries and the Hess Diagram

We estimate the absolute age of the globular cluster NGC 3201 using 10,000 sets of theoretical isochrones constructed through Monte Carlo simulation using the Dartmouth Stellar Evolution Program. These isochrones take into consideration the uncertainty introduced by the choice of stellar evolution parameters. We fit isochrones with three detached eclipsing binaries and obtained an age independent of distance. We also fit isochrones with differential reddening corrected Hubble Space Telescope photometry data utilizing two different Hess diagram-based fitting methods. Results from three different methods analyzing two different types of data agree to within 1σ, and we find the absolute age of NGC 3201 = 11.85 ± 0.74 Gyr. We also perform a variable importance analysis to study the uncertainty contribution from individual parameters, and we find the distance is the dominant source of uncertainty in photometry-based analysis, while total metallicity, helium abundance, α-element abundance, mixing length, and treatment of helium diffusion are an important source of uncertainties for all three methods.

1H NMR Profiling of Honey Bee Brains across Varying Ages and Seasons.

Honey bees (Apis mellifera) provide a useful model for studying aging because of the differences in longevity between the relatively short-lived summer and long-lived winter bees, as well as bees lacking signs of cognitive senescence as they age. Bee brains were dissected from newly emerged, 14-day-, and 28-day- old bees in mid- and late summer, as well as brood nest bees in fall, winter, and spring, before, during, and after overwintering, respectively. Brains were examined with nuclear magnetic resonance (NMR) spectroscopy to analyze their metabolome. Nine variable importance in projection (VIP) variables were identified, primarily amino acids and choline derivatives. Differences in metabolite concentrations were found with different ages of summer bees, mostly between newly emerged and 14-day- old bees, such as a decrease in phenylalanine and an increase in β-alanine, but there were also changes in older adults, such as o-phosphocholine that declined in 28-day- old bees. Differences in brood nest bees were observed, including a decline in tryptophan and an increase in β-alanine. These may provide distinct metabolomic signatures with age and season. Such research holds promise for a better understanding of the complex interplays between bee physiology, development, and aging, which has implications for improving bee health and management.

Determinants of microbial community structure in supraglacial pool sediments of monsoonal Tibetan Plateau.

Supraglacial pools are prevalent on debris-covered mountain glaciers, yet only limited information is available on the microbial communities within these habitats. Our research questions for this preliminary study were: (1) What microbes occur in supraglacial pool sediments of monsoonal Tibet?; (2) Which abiotic and biotic habitat variables have the most influence on the microbial community structure?; and (3) Does microbial composition of supraglacial pool sediments differ from that of glacial-melt stream pool sediments? We collected microbial samples for 16S rRNA sequencing and invertebrates for enumeration and identification and measured 14 abiotic variables from 46 supraglacial pools and nine glacial-melt stream pools in 2018 and 2019. Generalized linear model analyses, small sample Akaike information criterion, and variable importance scores were used to identify the best predictor variables of microbial community structure. Multi-response permutation procedure (MRPP) was used to compare taxa composition between supraglacial pools and stream pools. The most abundant phyla in supraglacial pool sediments were Proteobacteria, Actinobacteria, Bacteroidota, Chloroflexi, and Cyanobacteria. Genera richness, indicator genera richness, and Polaromonas relative abundance were best predicted by Chironomidae larvae abundance. Angustibacter and Oryzihumus relative abundance were best predicted by pH, Acidiphilium relative abundance was best predicted by turbidity, and Sphingomonas relative abundance was best predicted by glacier zone. Taxa composition was similar between supraglacial and stream pools at the class, genus, and ASV taxonomic levels. Our results indicate that Chironomidae larvae may play a keystone species role in shaping bacterial communities of supraglacial pools on debris-covered glaciers.IMPORTANCEGlacier meltwater habitats (cryoconite holes, supraglacial pools, supraglacial ponds and lakes, glacial streams) and their biota have not been well-studied, especially on debris-covered glaciers in temperate monsoonal regions. Our study is the first to document the microbial community-habitat relationships in supraglacial pools on a debris-covered glacier in Tibet. Microbial genera richness, indicator genera richness, and Polaromonas relative abundance declined with increasing larval Chironomidae abundance, which is a novel finding that highlights the importance of larval insects in structuring microbial communities in supraglacial pools.

Chemical analysis of different parts from agarwood columns by artificially agarwood-inducing method based on GC–MS and UPLC-TOF-MS

Agarwood is resin-containing wood produced by plants that have been injured. It is widely used in herbal medicine, incense, decorative items, and so on. In this study, we conducted resin area statistical analysis, determined starch particle and reducing sugar contents, and performed multivariate statistical analysis of chemical composition by GC–MS and UPLC-Q-TOF-MS to explore the different components in sections cut from an agarwood column, designated as A1–A4. The results showed that after stimulation by Agar-Bit inducer, the internal phloem parenchyma cells of the column started to form agarwood, and then starch granules were converted into soluble reducing sugars and agarwood resin. Section A1 showed rapid loss of starch granules, resulting in higher contents of reducing sugars and resin. The resin areas of agarwood in the respective sections were different, gradually decreasing on going from A1 to A4. Total numbers of metabolites of 87 and 63 were identified by GC–MS and UPLC-Q-TOF-MS, respectively. Of these, 10 and 16 metabolites with significant differences (variable importance projection >1) were selected through multivariate statistical analysis. These metabolites included chromones, sesquiterpenes, alkanes, and fatty acids. Among them, 6-methoxy-2-(2-phenylethyl)chromone and 6,7-dimethoxy-2-(2-phenylethyl)chromone were significant markers detected by both GC–MS and UPLC-Q-TOF-MS, which may be essential substances responsible for differences in the agarwood-forming capacities of the cut sections. In conclusion, there has been limited research on the different agarwood-forming capacities of agarwood columns. Here, we explored the differences in various sections of agarwood through chemical analysis to provide a more comprehensive and in-depth understanding of its constitution.

The circadian rhythm: A key variable in aging?

The determination of age-related transcriptional changes may contribute to the understanding of health and life expectancy. The broad application of results from age cohorts may have limitations. Altering sample sizes per time point or sex, using a single mouse strain or tissue, a limited number of replicates, or omitting the middle of life can bias the surveys. To achieve higher general validity and to identify less distinctive players, bulk RNA sequencing of a mouse cohort, including seven organs of two strains from both sexes of 5 ages, was performed. Machine learning by bootstrapped variable importance and selection methodology (Boruta) was used to identify common aging features where the circadian rhythms (CiR) transcripts appear as promising age markers in an unsupervised analysis. Pathways of 11 numerically analyzed local network clusters were affected and classified into four major gene expression profiles, whereby CiR and proteostasis candidates were particularly conspicuous with partially opposing changes. In a data-based interaction association network, the CiR-proteostasis axis occupies an exposed central position, highlighting its relevance. The computation of 11,830 individual transcript associations provides potential superordinate contributors, such as hormones, to age-related changes, as in CiR. In hormone-sensitive LNCaP cells, short-term supraphysiologic levels of the sex hormones dihydrotestosterone or estradiol increase the expression of the CiR transcript Bhlhe40 and the associated senescence regulator Cdkn2b (p15). According to these findings, the bilateral dysregulation of CiR appears as a fundamental protagonist of aging, whose transcripts could serve as a biological marker and its restoration as a therapeutic opportunity.

Development and validation of early prediction models for new-onset functional impairment of patients with trauma at hospital discharge.

Early identification of individuals at risk of functional impairment after trauma is crucial for the timely clinical decision-making and intervention to improve reintegration into the society. This study aimed to develop and validate models for predicting new-onset functional impairment after trauma using predictors that are routinely collected within 2 days of hospital admission. In this multicenter retrospective cohort study of acute care hospitals in Japan, we identified adult patients with trauma with independence in carrying out activities of daily living before hospitalization, treated in the intensive or high-dependency care unit, and survived for at least 2 days between April 2008 and September 2023. The primary outcome was functional impairment defined as Barthel Index ≤60 at hospital discharge. In the internal validation data set (between April 2008 and August 2022), using the routinely collected 129 candidate predictors within 2 days of admission, we trained and tuned the four conventional and machine learning models with repeated random subsampling cross-validation. We measured the performance of these models in the temporal validation data set (between September 2022 and September 2023). We also computed the importance of each predictor variable in our model. We identified 8,529 eligible patients. Functional impairment at discharge was observed in 41% of the patients (n = 3,506/8,529). In the temporal validation data set, all four models showed moderate discrimination ability, with areas under the curve above 0.79, and extreme gradient boosting showing the best performance (0.83). In the variable importance analyses, age was the most important predictor, followed by consciousness, severity score, cervical spinal cord injury, mild dementia, and serum albumin level at admission. We successfully developed early prediction models for patients with trauma with new-onset functional impairment at discharge that achieved high predictive performance using routinely collected data within 2 days of hospital admission. Prognostic/Epidemiological; Level II.

Longitudinal metabolite and protein trajectories prior to diabetes mellitus diagnosis in Danish blood donors: a nested case–control study

Aims/hypothesisMetabolic risk factors and plasma biomarkers for diabetes have previously been shown to change prior to a clinical diabetes diagnosis. However, these markers only cover a small subset of molecular biomarkers linked to the disease. In this study, we aimed to profile a more comprehensive set of molecular biomarkers and explore their temporal association with incident diabetes.MethodsWe performed a targeted analysis of 54 proteins and 171 metabolites and lipoprotein particles measured in three sequential samples spanning up to 11 years of follow-up in 324 individuals with incident diabetes and 359 individuals without diabetes in the Danish Blood Donor Study (DBDS) matched for sex and birth year distribution. We used linear mixed-effects models to identify temporal changes before a diabetes diagnosis, either for any incident diabetes diagnosis or for type 1 and type 2 diabetes mellitus diagnoses specifically. We further performed linear and non-linear feature selection, adding 28 polygenic risk scores to the biomarker pool. We tested the time-to-event prediction gain of the biomarkers with the highest variable importance, compared with selected clinical covariates and plasma glucose.ResultsWe identified two proteins and 16 metabolites and lipoprotein particles whose levels changed temporally before diabetes diagnosis and for which the estimated marginal means were significant after FDR adjustment. Sixteen of these have not previously been described. Additionally, 75 biomarkers were consistently higher or lower in the years before a diabetes diagnosis. We identified a single temporal biomarker for type 1 diabetes, IL-17A/F, a cytokine that is associated with multiple other autoimmune diseases. Inclusion of 12 biomarkers improved the 10-year prediction of a diabetes diagnosis (i.e. the area under the receiver operating curve increased from 0.79 to 0.84), compared with clinical information and plasma glucose alone.Conclusions/interpretationSystemic molecular changes manifest in plasma several years before a diabetes diagnosis. A particular subset of biomarkers shows distinct, time-dependent patterns, offering potential as predictive markers for diabetes onset. Notably, these biomarkers show shared and distinct patterns between type 1 diabetes and type 2 diabetes. After independent replication, our findings may be used to develop new clinical prediction models.Graphical

Analysis of metabolic spectrum characteristics of naturally and cultivated Ophiocordyceps sinensis based on non-targeted metabolomics

The analysis of the differences in metabolic profiles between naturally Ophiocordyceps sinensis (NO) and cultivated Ophiocordyceps sinensis (CO) is an essential process for the medicinal value mining of Ophiocordyceps sinensis. Non-targeted metabolomics was used to compare the differences in metabolite composition and abundance between NO and CO. Total metabolite composition found that NO is rich in organic acids and derivatives, and CO is rich in lipids and lipid-like molecules. HCA found that organooxygen compounds, cinchona alkaloid, and fatty acyls had different abundances in NO and CO. The variable importance in projection value and quantitative analysis of metabolites found that NO was rich in l-iditol, malate, linoleic acid, and oleic acid; CO is rich in sucrose, perseitol, hydroquinidine, nonanoic acid, 1-hydroxy-2-naphthoic acid, hymol-β-d-glucoside, and gly-his-lys. these compounds have the potential to be biomarkers of NO and CO. KEGG enrichment analysis showed that ascorbate and aldarate metabolism, carbon metabolism, pyrimidine metabolism, and fatty acid biosynthesis were the most different metabolic pathways between NO and CO. Therefore, the analysis of the characteristics of NO and CO metabolites has reference value for finding their different medicinal functions.

The inversion of arid-coastal cultivated soil salinity using explainable machine learning and Sentinel-2

The escalating salinization of cultivated soil poses a significant threat to the ecological environment. It is imperative to establish a monitoring system and mitigate the spread of salinization in arid and coastal areas through remote sensing, incorporating high-precision cross-regional models for soil salt content inversion. This study focuses on typical saline-alkali soils in arid and coastal regions of China. Using Sentinel 2 data (including 6 bands and 27 spectral indices), along with soil texture, moisture content, temperature, precipitation, and digital elevation model (DEM) data to establish an arid-coastal salinity inversion model. Variable selection methods such as pearson correlation coefficient (PCC), variable importance in projection (VIP), gray relational analysis (GRA), and gradient boosting machine (GBM) were used, while using 9 models including adaptive boosting (Adaboost), extremely randomized trees (ERT), and light gradient boosting machine (LightGBM). The best model was further elucidated using the Shapley additive explanations method. Results indicate that the common sensitive characteristic variables of arid-coastal areas were spectral indices and soil properties in PCC, the spectral variable bands and indices in VIP, and all variables in GRA and GBM. The best inversion model GBM-ERT (R2 = 0.91, RMSE = 1.06) in arid areas exhibited higher accuracy than the best inversion model GBM-Adaboost (R2 = 0.77, RMSE = 1.74) in coastal areas. The arid-coastal inversion model PCC-LightGBM demonstrated the best inversion performance (R2 = 0.64, RMSE = 2.29) and simulation performance in arid (R2 = 0.67) and coastal areas (R2 = 0.63). Dead fuel index (DFI) had the most significant impact on model prediction (0.89) and the second ratio index (RI2) contributed the highest relative importance (18 %) to the model. Our analysis indicates that the arid-coastal model of PCC-LightGBM established using common characteristic variables, can effectively monitor large-scale soil salinity.

Biomarker panels for fruit intake assessment: a metabolomics analysis in the ELSA-Brasil study.

Food intake biomarkers are used to estimate dietary exposure; however, selecting a single biomarker to evaluate a specific dietary component is difficult due to the overlap of diverse compounds from different foods. Therefore, combining two or more biomarkers can increase the sensitivity and specificity of food intake estimates. This study aimed to evaluate the ability of metabolite panels to distinguish between self-reported fruit consumers and non-consumers among participants in the Longitudinal Study of Adult Health. A total of 93 healthy adults of both sexes were selected from the Longitudinal Study of Adult Health. A 24-h dietary recall was obtained using the computer-assisted 24-h food recall GloboDiet software, and 24-h urine samples were collected from each participant. Metabolites were identified in urine using liquid chromatography coupled with high-resolution mass spectrometry by comparing their exact mass and fragmentation patterns using free-access databases. Multivariate receiver operating characteristic curve (ROC) analysis and partial least squares discriminant analysis were used to verify the ability of the metabolite combination to classify daily and non-daily fruit consumers. Fruit intake was identified using a 24h dietary recall (24h-DR). Bananas, grapes, and oranges are included in the summary. The panel of biomarkers exhibited an area under the curve (AUC) > 0.6 (Orange AUC = 0.665; Grape AUC = 0.622; Bananas AUC = 0.602; All fruits AUC = 0.679; Citrus AUC = 0.693) and variable importance projection score > 1.0, and these were useful for assessing the sensitivity and predictability of food intake in our population. A panel of metabolites was able to classify self-reported fruit consumers with strong predictive power and high specificity and sensitivity values except for banana and total fruit intake.

Anticipating influential factors on suicide outcomes through machine learning techniques: Insights from a suicide registration program in western Iran

Suicide is a global public health concern, with increasing rates observed in various regions, including Iran. This study focuses on the province of Hamadan, Iran, where suicide rates have been on the rise. The research aims to predict factors influencing suicide outcomes by leveraging machine learning techniques on the Hamadan Suicide Registry Program data collected from 2016 to 2017. The study employs Naïve Bayes and Random Forest algorithms, comparing their performance to logistic regression. Results highlight the superiority of the Random Forest model. Based on the variable importance and multiple logistic regression analyses, the most important determinants of suicide outcomes were identified as suicide method, age, and timing of attempts, income, and motivation. The findings emphasize the cultural context's impact on suicide methods and underscore the importance of tailoring prevention programs to address specific risk factors, especially for older individuals. This study contributes valuable insights for suicide prevention efforts in the region, advocating for context-specific interventions and further research to refine predictive models and develop targeted prevention strategies.

The hotspot of consumers complaints on infant formula: Exploring the relationship among volatile components, sensory evaluation, and aroma recombination and omission experiment

Two-stage infant formula (IF) is rich in polyunsaturated fatty acids (PUFAs), However, PUFAs are susceptible to oxidation, thus affecting the flavor profile of the product. This study aimed to evaluate 4 samples with the participation of 60 consumers. Among these, three samples with the lowest consumer preference and one with the highest preference were selected as controls to explain the causes of off-flavors in two-stage infant formulas. Through qualitative and quantitative analysis of 28 flavor-active compounds in two-stage infant formulas using GC-MS, E-nose, GC-O, odor activity values (OAVs), and variable importance in projection (VIP) values, hexanal, heptanal, pentanal, octanal, and 1-penten-3-one were identified as the important compounds affecting the flavor of two-stage infant formulas. Further, recombination/deficiency experiments confirmed hexanal and heptanal as the key undesirable flavor components in two-stage infant formulas.