Random Regression Research Articles

Machine learning models for predicting interaction affinity energy between human serum proteins and hemodialysis membrane materials

Membrane incompatibility poses significant health risks, including severe complications and potential fatality. Surface modification of membranes has emerged as a pivotal technology in the membrane industry, aiming to improve the hemocompatibility and performance of dialysis membranes by mitigating undesired membrane-protein interactions, which can lead to fouling and subsequent protein adsorption. Affinity energy, defined as the strength of interaction between membranes and human serum proteins, plays a crucial role in assessing membrane-protein interactions. These interactions may trigger adverse reactions, potentially harmful to patients. Researchers often rely on trial-and-error approaches to enhance membrane hemocompatibility by reducing these interactions. This study focuses on developing machine learning algorithms that accurately and rapidly predict affinity energy between novel chemical structures of membrane materials and human serum proteins, based on a molecular docking dataset. Various membrane materials with distinct characteristics, chemistry, and orientation are considered in conjunction with different proteins. A comparative analysis of linear regression, K-nearest neighbors regression, decision tree regression, random forest regression, XGBoost regression, lasso regression, and support vector regression is conducted to predict affinity energy. The dataset, comprising 916 records for both training and test segments, incorporates 12 parameters extracted from data points and involves six different proteins. Results indicate that random forest (R² = 0.8987, MSE = 0.36, MAE = 0.45) and XGBoost (R² = 0.83, MSE = 0.49, MAE = 0.49) exhibit comparable predictive performance on the training dataset. However, random forest outperforms XGBoost on the testing dataset. Seven machine learning algorithms for predicting affinity energy are analyzed and compared, with random forest demonstrating superior predictive accuracy. The application of machine learning in predicting affinity energy holds significant promise for researchers and professionals in hemodialysis. These models, by enabling early interventions in hemodialysis membranes, could enhance patient safety and optimize the care of hemodialysis patients.

Differential Neurotoxicity Induced in Rats by Injection of PMs from 31 Major Cities in China.

The effects of particulate matter (PMs) from different cities on the nervous system remain unclear. In this study, aqueous solutions of 0.45 μm membrane-filtered PM from 31 major Chinese cities were intravenously administered to rats. Neurotoxicity and mechanisms were investigated by quantifying rat behavior, analyzing in vivo biomarkers, and examining the PM physicochemical properties. PMs from different cities had variable impacts on rat responses, as manifested by the movement speed of the right ear, particularly at 1 h and 4-6 days postexposure. Physiological mechanisms were linked to the regulation of glucocorticoids via the hypothalamic-pituitary-adrenal axis and miR-107/miR-124 expression in the blood. Additionally, PM toxicity was strongly influenced by particle morphology, size, and zeta potential, which varied greatly across cities. Using random forest and multiple linear regression, we revealed that PM particle sizes (458.67 and 531.17 nm) and PM zeta potentials (-3.78, -17.01, and -20.31 mV) had the most important impacts on rat responsiveness, which was in line with blood biomarkers levels in rats such as Glucocorticoid, IL-1β, and IFN-α. These findings indicate that PMs from 31 cities contribute to varying neurotoxicity, thus presenting a possible differential burden on Alzheimer's disease in the aging population across many different regions.

A Seasonal Fresh Tea Yield Estimation Method with Machine Learning Algorithms at Field Scale Integrating UAV RGB and Sentinel-2 Imagery

Traditional methods for estimating tea yield mainly rely on manual sampling surveys and empirical estimation, which are labor-intensive and time-consuming. Accurately estimating fresh tea production in different seasons has become a challenging task. It is possible to estimate the seasonal yield of tea at the field scale by using the spatial resolution of 10 m, 5-day revisit period and rich spectral information of Sentinel-2 imagery. This study integrated Sentinel-2 images and uncrewed aerial vehicle (UAV) RGB imagery to develop six regression models at the field scale, which were employed for the estimation of seasonal and annual fresh tea yields of the Yunlong Tea Cooperatives in Yixiang Town, Pu’er City, China. Firstly, we gathered fresh tea production data from 133 farmers in the cooperative over the past five years and obtained UAV RGB and Sentinel-2 imagery. Secondly, 23 spectral features were extracted from Sentinel-2 images. Based on the UAV images, the parcel of each farmer was positioned and three topographic features of slope, aspect, and elevation were extracted. Subsequently, these 26 features were screened using the random forest algorithm and Pearson correlation analysis. Thirdly, we applied six different regression algorithms to establish fresh tea yield models for each season and evaluated their estimation accuracy. The results showed that random forest regression models were the optimal choice for estimating spring and summer yields, with the spring model achieving an R2 value of 0.45, an RMSE of 40.38 kg/acre, and an rRMSE of 40.79%. Similarly, the summer model achieved an R2 value of 0.5, an RMSE of 78.46 kg/acre, and an rRMSE of 39.81%. For autumn and annual yield estimation, voting regression models demonstrated superior performance, with the autumn model achieving an R2 value of 0.42, an RMSE of 70.6 kg/acre, and an rRMSE of 39.77%, and the annual model attained an R2 value of 0.47, an RMSE of 168.7 kg/acre, and an rRMSE of 34.62%. This study provides a promising new method for estimating fresh tea yield in different seasons at the field scale.

Health Insurance Premium Prediction Based on Multiple Machine Learning Algorithms

The purpose of this paper is to predict the health insurance premium through a variety of machine learning algorithms, and compare and analyze the prediction effect of different algorithms. An open source dataset was selected for the study, and the experiments involved three machine learning models: linear regression, decision trees, and random forests. By testing these models, we obtain their performance in health insurance premium forecasts. The results show that the prediction performance of the random forest regression model is better than other models, and its score reaches 0.8564, which is the best algorithm among the three models. Second, the linear regression model has a score of 0.7584, and although its performance is not as good as that of random forest, it still shows some predictive power. Finally, the prediction effect of decision tree model is relatively poor, and the score is only 0.7097. To sum up, the experiments in this paper prove that the random forest model is undoubtedly the best choice in the prediction of health insurance premiums, which not only has good prediction accuracy, but also shows strong data processing ability. In the context of the growing importance of health insurance premium collection and analysis, the use of advanced machine learning algorithms such as Random Forest for forecasting will, to some extent, help insurance companies better price and assess risk. Therefore, it can be concluded that random forest regression model has the best performance for health insurance premium prediction and is an effective tool to achieve accurate prediction.

Child Maltreatment Characteristics and Adult Physical Multimorbidity in Germany

Associations between child maltreatment (CM) and health have been studied broadly, but most studies focus on multiplicity (number of experienced subtypes of CM). Studies assessing multiple CM characteristics are scarce, partly due to methodological challenges, and were mostly conducted in patient samples. To determine the importance of CM characteristics in association with physical multimorbidity in adulthood for women and men in a German representative sample. This survey study used data from a representative sample of the German population between July and October 2021. Households were randomly selected within regional areas by a German demographic consultation company, and the Kish-Grid method was applied to ensure random participation. Reasons for nonparticipation were refusal of the selected household or target person to participate and failure to contact a household after 4 visits. Analyses took place between June 2023 and July 2024. Characteristics of CM, including subtypes of CM, multiplicity, age at time of maltreatment (timing), number of years of experienced maltreatment (duration), frequency, and subjective severity (measured with the ISPCAN Child Abuse Screening Tools Retrospective version questionnaire). The main outcome was physical multimorbidity, defined as the sum score of lifetime leading morbidity and mortality causes in Western countries (obesity, diabetes, cancer, hypertension, myocardial infarction, chronic obstructive pulmonary disease, and incident stroke). Conditioned random forest regression analyses (a machine learning regression technique) were conducted to examine what characteristics of CM were most importantly associated with physical multimorbidity in adulthood while considering all other variables in the model. Of 5908 individuals invited, the study sample included 2514 participants (response rate, 42.6%), with 1297 (51.6%) women (mean [SD] age, 50.6 [17.9] years) and 1217 (48.4%) men (mean [SD] age, 49.5 [18.2] years). Duration of CM was the most important factor for physical multimorbidity in adulthood for both women (importance = 0.595; 95% CI, 0.599-0.601) and men (importance = 1.389; 95% CI, 1.386-1.394). Duration and timing variables were more importantly associated with outcomes than multiplicity in women and men. For women, severity and experiencing CM at age 4 years was significantly associated with physical multimorbidity in adulthood. For men, experiencing CM at age 11 years was most importantly associated with physical multimorbidity in adulthood. In this survey study, conditioned random forest regression analyses were applied to provide insights in the importance of duration and timing of CM for physical health in addition to the frequently studied multiplicity. These findings suggest that CM assessments should be considered in diagnostics of individuals with physical health conditions and may also inform strategies to mitigate the risk.

Palmitoylation-related gene ZDHHC22 as a potential diagnostic and immunomodulatory target in Alzheimer’s disease: insights from machine learning analyses and WGCNA

BackgroundThe mechanism of palmitoylation in the pathogenesis of Alzheimer's disease (AD) remains unclear.MethodsThis study retrieved AD data sets from the GEO database to identify palmitoylation-associated genes (PRGs). This study applied WGCNA along with three machine learning algorithms—random forest, LASSO regression, and SVM–RFE—to further select key PRGs (KPRGs). The diagnostic performance of KPRGs was evaluated using Receiver Operating Characteristic (ROC) curve analysis. Immune cell infiltration analysis was conducted to assess correlations between KPRGs and immune cell types, and a competing endogenous RNA (ceRNA) regulatory network was constructed to explore their potential regulatory mechanisms.Results17 PRGs were identified from the AD data sets, with 7 genes showing increased expression and 10 showing decreased expression. Through WGCNA and machine learning analyses, ZDHHC22 was selected as a KPRG. The ROC curve analysis demonstrated that ZDHHC22 had an area under the curve value of 0.659, indicating moderate diagnostic potential. Immune cell infiltration analysis revealed significant associations between ZDHHC22 expression and the infiltration of several immune cell types, including naïve B cells, CD8 + T cells, and M1 macrophages. In addition, 25 miRNAs and 55 lncRNAs were predicted to potentially target ZDHHC22, forming the basis for a lncRNA–miRNA–mRNA ceRNA network.ConclusionsThis study is the first to use bioinformatics methods to identify ZDHHC22 as a key KPRG in AD, highlighting its potential role in disease diagnosis and immune regulation. The regulatory network of ZDHHC22 provides new insights into the molecular mechanisms of AD and lays the foundation for future targeted therapeutic strategies.

Dengue dynamics, predictions, and future increase under changing monsoon climate in India

The global burden of dengue disease is escalating under the influence of climate change, with India contributing a third of the total. The non-linearity and regional heterogeneity inherent in the climate-dengue relationship and the lack of consistent data makes it difficult to make useful predictions for effective disease prevention. The current study investigates these non-linear climate-dengue links in Pune, a dengue hotspot region in India with a monsoonal climate and presents a model framework for predicting both the near-term and future dengue mortalities. Dengue mortality and meteorological conditions over a twelve-year period (2004–2015) are analyzed using statistical tools and machine learning methods. Our findings point to a significant influence of temperature, rainfall, and relative humidity on dengue mortality in Pune, at a time-lag of 2–5 months, providing sufficient lead time for an early warning targeted at curbing dengue outbreaks. We find that moderate rains spread over the summer monsoon season lead to an increase in dengue mortality, whereas heavy rains reduce it through the flushing effect, indicating the links between dengue and monsoon intraseasonal variability. Additionally, warm temperatures above 27°C and humidity levels between 60% and 78% elevate the risk of dengue. Based on these weather-dengue associations, we developed a machine-learning model utilizing the random forest regression algorithm. The dengue model yields a skillful forecast, achieving a statistically significant correlation coefficient of r = 0.77 and a relatively low Normalized Root Mean Squared Error score of 0.52 between actual and predicted dengue mortalities, at a lead time of two months. The model finds that the relative contributions of temperature, rainfall, and relative humidity to dengue mortality in Pune are 41%, 39%, and 20%, respectively. We use the dengue model in conjunction with the climate change simulations from the Coupled Model Intercomparison Project phase 6 for the future dengue mortality projections under a global warming scenario. In a changing climate, dengue-related mortality in Pune is projected to rise by 13% in the near future (2021–2040), 23–40% in the mid-century (2041–2060), and 30–112% in the late century (2081–2100) under low-to-high emission pathways in response to the associated increase in temperature and changes in monsoon rainfall patterns.

A parametric insurance policy for beekeepers and honey production: random forest regressions and real-world pricing

Abstract In the past years, climate change has affected honey production more and more and the reduction has become a significant risk for beekeepers. In this paper, we discuss the pricing of a parametric insurance policy drafted to cover the potential losses in terms of honey production due to unfavorable weather conditions: the payment of the insurance benefit is triggered by the breaching of predefined thresholds of a weather index, measured over specific relevant periods. The effectiveness of the coverage is verified by the means of random forests, where the honey production is forecast under different real-world weather scenarios and the beekeepers’ loss is compared with the insurance benefit reimbursed (or not) by the policy. The random forest technique is put along with more common ones, such as ordinary least squares regression and mixed linear models. A practical example is given for the Italian market, where the pricing is derived and assessed for three different zones: North, Centre, and South.

Evaluating the genetic components of growth in ostriches with multi-trait and random regression models

ABSTRACT 1. The heritability (h2) of liveweight (LW) in ostriches can be highly variable, depending on age at recording. The objective of this study was to consider random regression (RR) as an alternative to the multi-trait (MT) structure for the analysis of repeated measures of LW. 2. The data included 74 683 LW phenotypes recorded from 10 052 birds aged between 20 and 410 days (d) of age. Statistical analysis included single trait (ST), MT and RR analysis in a linear mixed model framework using the ASREML V4.2 software. 3. For ST and MT, six traits were defined to represent LW at 28, 77, 150, 230, 300 and 365 d of age. Random variance components included direct genetic and maternal permanent environment (PE) effects. A MT analysis including all six traits converged. 4. For RR, the data was transformed (LW + 10)−0.5 due to difficulty in dealing with large scale effects. The final RR model fitted direct genetic and animal PE components as third degree Legendre polynomials and heterogeneous residuals. 5. The h2 estimates was in agreement across analysis, ranging from moderate (0.16-0.20) for W28 to high (0.41-0.51) for W230 to W365. Importantly, the genetic relationship between LW recorded as a chick and juvenile was only moderate (~0.35 to 0.55). The correlations between RR and MT EBVs for the six traits were 0.85, 0.54, 0.65, 0.75, 0.83 and 0.91, showing a considerable level of re-ranking. 6. This study reaffirmed age dependent genetic variation when determining LW in ostriches. The RR structure was useful for overcoming the dimension problem of MT analysis, but was susceptible to scale effects present in the data, despite transformation. It remains unknown whether the need for cubic terms reflected scale or animal effects.

Predicting fall parameters from infant skull fractures using machine learning.

When infants are admitted to the hospital with skull fractures, providers must distinguish between cases of accidental and abusive head trauma. Limited information about the incident is available in such cases, and witness statements are not always reliable. In this study, we introduce a novel, data-driven approach to predict fall parameters that lead to skull fractures in infants in order to aid in determinations of abusive head trauma. We utilize a state-of-the-art finite element fracture simulation framework to generate a unique dataset of skull fracture patterns from simulated falls. We then extract features from the resulting fracture patterns in this dataset to be used as input into machine learning models. We compare seven machine learning models on their abilities to predict two fall parameters: impact site and fall height. The results from our best-performing models demonstrate that while predicting the exact fall height remains challenging ( 0.27 for the ridge regression model), we can effectively identify potential impact sites ( between 0.65 and 0.76 for the random forest regression model). This work not only provides a tool to enhance the ability to assess abuse in cases of pediatric head trauma, but also advocates for advancements in computational models to simulate complex skull fractures.

Childhood maltreatment and brain aging during adulthood.

Childhood maltreatment (CM) is associated with the early onset of psychiatric and medical disorders and accelerated biological aging. To identify types of maltreatment and developmental sensitive periods that are associated with accelerated adult brain aging. Participants were mothers of infants recruited from the community into a study assessing the effects of CM on maternal behavior, infant attachment, and maternal and infant neurobiology. Data were collected from July 2015 to November 2019 and were analyzed from July 2023 to October 2024. Academic medical centers. High-quality MRI scans were obtained on 92 of 150 mothers enrolled in the study. The main exclusion criteria for neuroimaging were histories of head trauma with loss of consciousness or concussion, psychotropic use before age 18, pregnancy, and customary MRI exclusions (e.g., metal implant). The primary reasons for non-completion of the neuroimaging study were unwillingness to be scanned, inability to attend the MRI study visit due to work and/or childcare, metal implants, or pregnancy. The Maltreatment and Abuse Chronology of Exposure scale was used to retrospectively assess the annual severity of exposure to ten types of CM from birth to age 18 years. Brain age was calculated from T1-weighted 3T MRI Scans using a previously published machine learning algorithm. Sensitive periods were identified using random forest regression with conditional inference trees. Forty-nine (53.3%) of the 92 mothers (mean [SD] age, 32.4 [4.3] years) reported experiencing one or more types of CM. Total CM severity was associated with accelerated brain aging (β=0.05, 95% CI, 0.02 to 0.09, p<.005). The most robust type/time risk factors for accelerated brain aging were parental physical abuse between ages 4 to 6 years, witnessing sibling violence between ages 4 to 15 years, parental verbal abuse between ages 10 to 12 years, and parental emotional neglect between ages 16 to 18 years. Several types of CM between ages 4-18 years were associated with accelerated brain aging. Understanding how these specific types and ages of exposure contribute to accelerated brain aging may provide important insights into preventing key clinical consequences of CM. Question: How does childhood adversity relate to brain aging in adulthood, and are there sensitive periods for this association? Findings: In this cohort-based study of adult women, we observed sensitive periods for the association between adult brain aging and five classes of childhood maltreatment: parental physical abuse, parental verbal abuse, parental emotional neglect, and witnessing sibling violence. Meaning: These findings suggest that maltreatment subtype and age at exposure may be important factors contributing to the impact of childhood adversity on brain aging later in life.

The association between metabolomic profiles of lifestyle and the latent phase of incident chronic kidney disease in the UK Population

Chronic kidney disease (CKD) is a global health challenge associated with lifestyle factors such as diet, alcohol, BMI, smoking, sleep, and physical activity. Metabolomics, especially nuclear magnetic resonance(NMR), offers insights into metabolic profiles’ role in diseases, but more research is needed on its connection to CKD and lifestyle factors. Therefore, we utilized the latest metabolomics data from the UK Biobank to explore the relationship between plasma metabolites and lifestyle factors, as well as to investigate the associations between various factors, including lifestyle-related metabolites, and the latent phase of CKD onset. The study enrolled approximately 500,000 participants from the UK Biobank (UKB) between 2006 and 2010, excluding 447,163 individuals with missing data for any metabolite in the NMR metabolomics, any biomarker in the blood chemistry (including eGFR, albumin, or cystatin C), any factor required for constructing the lifestyle score, or a baseline diagnosis of CKD. Lifestyle scores (LS) were calculated based on several factors, including diet, alcohol consumption, smoking, BMI, physical activity, and sleep. Each healthy lifestyle component contributed to the overall score, which ranged from 0 to 6. A total of 249 biological metabolites covering multiple categories were determined by the NMR Metabolomics Platform. Random forest algorithms and LASSO regression were employed to identify lifestyle-related metabolites. Subsequently, accelerated failure time models(AFT) were used to assess the relationship between multiple factors, including traditional CKD-related biomarkers (such as eGFR, cystatin C, and albumin) and lifestyle-related metabolites, with the latent phase of incident CKD. Finally, we performed Kaplan–Meier survival curve analysis on the significant variables identified in the AFT model. Over a mean follow-up period of 13.86 years, 2,279 incident chronic kidney disease (CKD) cases were diagnosed. Among the 249 metabolites analyzed, 15 were identified as lifestyle-related, primarily lipid metabolites. Notably, among these metabolites, each 1 mmol/L increase in triglycerides in large LDL particles accelerated the onset of CKD by 24%. Diabetes, hypertension, and smoking were associated with a 56.6%, 31.5% and 22.3% faster onset of CKD, respectively. Additionally, each unit increase in age, BMI, TDI, and cystatin C was linked to a 3.2%, 1.4%, 1.6% and 32.3% faster onset of CKD. In contrast, higher levels of albumin and eGFR slowed the onset of CKD, reducing the speed of progression by 3.0% and 3.9% per unit increase, respectively. Nuclear magnetic resonance metabolomics offers new insights into renal health, though further validation studies are needed in the future.

Phosphoric acid based geopolymer foam-activated carbon composite for methylene blue adsorption: isotherm, kinetics, thermodynamics, and machine learning studies.

In this study, a binary composite adsorbent based on activated carbon and phosphoric acid geopolymer foam (ACP) was prepared by combining phosphoric acid geopolymer (PAGP) with activated carbon (AC) and applied for the removal of methylene blue (MB). Activated carbon was thoroughly mixed with a mixture of fly ash and metakaolin in varying ratios, followed by phosphoric acid activation and thermal curing. The ACP adsorbent was characterized using scanning electron microscope (SEM), Fourier transform infrared (FTIR) spectrophotometer, X-ray diffractometer (XRD), surface area analyser (SAP), and thermogravimetric analyser (TGA). Batch analysis was performed to examine the effects of various adsorption parameters including pH (2, 4, 6, 7, 8, and 10), adsorbent dosage (0.06-0.2 g), MB concentration (50-250 mg L-1), contact duration (up to 240 minutes), and temperature (25-55 °C). The ACP with 70% PAGP and 30% AC was found to be the most suitable adsorbent as it maintained its structure and exhibited better MB adsorption. The ACP had a surface area of 47.36 m2 g-1 and a pore size of 5.6 nm and was found to be amorphous in nature. The adsorption equilibrium reached in 240 minutes at pH 7, indicating an efficient adsorption process. The adsorption increased with the initial dye concentration and decreased with the increase in temperature. The ideal parameters for adsorption of MB using ACP include 0.2 g of adsorbent, 25 °C, pH 10, and 240 minutes. The adsorption data fitted well with the Langmuir isotherm, pseudo-second-order (PSO) kinetics model, and three-step intraparticle diffusion (IPD) model. The adsorption capacity calculated using the Langmuir isotherm was 204.8 mg g-1 with an R 2 = 0.989. Thermodynamics parameters showed that the adsorption process was exothermic, energetically favourable, and associated with a decrease in entropy. According to the FTIR findings, pH effect, Langmuir isotherm, PSO kinetics, IPD model, and thermodynamics factors, chemisorption is identified as the predominant process. Different machine learning models, i.e., gaussian process regression (GPR), support vector regression (SVR and SVR-rbf), random forest regression (RFR), decision tree regression (DTR) and artificial neural network (ANN), were trained and tested using adsorption capacity and % removal data. The ANN model (random search) demonstrated better performance compared to other models, achieving an R 2 value of 0.873 for adsorption capacity and 0.799 for % removal on test data.

Deciphering Long‐Term Economic Growth: An Exploration With Leading Machine Learning Techniques

ABSTRACTExisting studies mainly focus on short‐term economic forecasts, but research on long‐term projections, particularly for periods spanning 6–10 years, remains insufficient, despite its importance. This gap may arise from the limitations of traditional linear methods in prediction tasks and pattern recognition, whereas machine learning techniques may help overcome these challenges. To address this, we employ five widely used machine learning models—artificial neural networks (ANN), random forest regression (RF), gradient boosting regression (GBR), extreme gradient boosting (XGBoost), and support vector regression (SVR)—using cross‐country data from 109 countries between 1961 and 2019. To ensure robustness, we employ two distinct sampling methods for model validation. Our findings reveal that the ANN model outperforms others, particularly in long‐term predictions (6–10 years), with an average out‐of‐sample prediction ‐squared of 0.89. Furthermore, analyses using permutation feature importance (PFI) and SHapley Additive exPlanations (SHAP) methods indicate that while current growth rates are critical for short‐term forecasts (1–3 years), two primary variables representing a country's foundational characteristics—real GDP per capita and “country‐feature,” akin to a country dummy variable—are crucial for long‐term predictions (4–10 years). This outcome demonstrates the ANN model's capacity to capture each country's unique characteristics and, through its highly non‐linear nature, successfully execute complex, long‐range forecasts. These results unveil the remarkable potential of machine learning in the realm of long‐term economic forecasting.

Hyperspectral estimation of chlorophyll density in winter wheat using fractional-order derivative combined with machine learning.

Chlorophyll density (ChD) can reflect the photosynthetic capacity of the winter wheat population, therefore achieving real-time non-destructive monitoring of ChD in winter wheat is of great significance for evaluating the growth status of winter wheat. Derivative preprocessing has a wide range of applications in the hyperspectral monitoring of winter wheat chlorophyll. In order to research the role of fractional-order derivative (FOD) in the hyperspectral monitoring model of ChD, this study based on an irrigation experiment of winter wheat to obtain ChD and canopy hyperspectral reflectance. The original spectral reflectance curves were preprocessed using 3 FOD methods: Grünwald-Letnikov (GL), Riemann-Liouville (RL), and Caputo. Hyperspectral monitoring models for winter wheat ChD were constructed using 8 machine learning algorithms, including partial least squares regression, support vector regression, multi-layer perceptron regression, random forest regression, extra-trees regression (ETsR), decision tree regression, K-nearest neighbors regression, and gaussian process regression, based on the full spectrum band and the band selected by competitive adaptive reweighted sampling (CARS). The main results were as follows: For the 3 types of FOD, GL-FOD was suitable for analyzing the change process of the original spectral curve towards the integer-order derivative spectral curve. RL-FOD was suitable for constructing the hyperspectral monitoring model of winter wheat ChD. Caputo-FOD was not suitable for hyperspectral research due to its insensitivity to changes in order. The 3 FOD calculation methods could all improve the correlation between the original spectral curve and Log(ChD) to varying degrees, but only the GL method and RL method could observe the change process of correlation with order changes, and the shorter the wavelength, the smaller the order, and the higher the correlation. The bands screened by CARS were distributed throughout the entire spectral range, but there was a relatively concentrated distribution in thevisible light region. Among all models, CARS was used to screen bands based on the 0.3-order RL-FOD spectrum, and the model constructed using ETsR reached the best accuracy and stability. Its R2c, RMSEc, R2v, RMSEv, and RPD were 1.0000, 0.0000, 0.8667, 0.1732, and 2.6660, respectively. In conclusion, based on the winter wheat ChD data set and the corresponding canopy hyperspectral data set, combined with 3 FOD calculation methods, 1 band screening method, and 8 modeling algorithms, this study constructed hyperspectral monitoring models for winter wheat ChD. The results showed that based on the 0.3-order RL-FOD, combined with the CARS screening band, ETsR modeling has the highest accuracy, and hyperspectral estimation of winter wheat ChD can be realized. The results of this study can provide some reference for the rapid and nondestructive estimation of ChD in winter wheat.

Clinical utility of tumor-infiltrating lymphocyte evaluation by two different methods in breast cancer patients treated with neoadjuvant chemotherapy.

The aim of this study was to examine the clinical utility of tumor-infiltrating lymphocytes (TILs) evaluated by "average" and "hot-spot" methods in breast cancer patients. We examined 367 breast cancer patients without neoadjuvant chemotherapy (NAC) by average and hot-spot methods to determine the consistency of TIL scores between biopsy and surgical specimens. TIL scores before NAC were also compared with the pathological complete response (pCR) rate and clinical outcomes in 144 breast cancer patients that received NAC. TIL scores evaluated by the two methods were predicted from clinicopathological data using random forest regression. Surgical specimens showed higher TIL scores than biopsy specimens using the hot-spot method (p < 0.001), while biopsy and surgical specimens showed similar TIL scores using the average method. There was a linear relationship between the pCR rate and TIL scores determined using hot-spot (p < 0.001) and average methods (p = 0.001). Patients without pCR and low TILs by the average method had significantly worse overall survival compared to other patients (p = 0.02). The root mean squared errors of the predicted TIL score for the test set were 19.662 (hot-spot) and 10.955 (average). The average method may have an advantage for breast cancer patients receiving NAC, since the TIL score using this method is more consistent between biopsy and surgical specimens, and it associates better with clinical outcomes. Our exploratory study showed that machine learning from clinicopathological data may better predict TIL scores assessed by the average, rather than hot-spot, method.

What are parents' preferences for Human Papillomavirus vaccination promotion messages and communication? Application of a discrete choice experiment to a French Caribbean setting

IntroductionHuman Papillomavirus (HPV) vaccine uptake in the French Caribbean has remained below 25% since introduction in 2007, which is well behind national and international targets. Using a discrete choice experiment (DCE), we explored parental preferences around HPV vaccination and optimized communication content in a sample of parents of middle-school pupils in Guadeloupe.MethodsWe conducted a cross-sectional survey in public and private middle age schools in Guadeloupe in June 2023 using an online questionnaire. Across a series of nine hypothetical scenarios, participants were asked to decide to vaccinate or not and how certain they were about this choice. Scenarios differed by five attributes (diseases characteristics, vaccine safety, health professionals or institutions promoting vaccination, social conformity and optimal vaccination age). We used random effect logit and linear regression models to estimate the effects of attribute levels on vaccine acceptance and vaccine eagerness.ResultsA total of 389 parents out of the 23,184 pupils’ parents completed the DCE survey. The attributes with a significant effect size on theoretical vaccine acceptance were "social conformity" and "optimal vaccination age”. Overall, the odds of scenarios stating high vaccine coverage in adolescents were at least 1.8 (95% CI: 1.2—2.6) times more likely to yield theoretical vaccine acceptance compared to a low vaccine uptake reference. The odds of providing scientific explanation along with age yielded theoretical vaccination acceptance respectively up to 3.2 times higher (95% CI: 1.7 to 6.1) in parents reporting an un vaccinated child and not intention to vaccinate. For vaccine eagerness, we observe significant positive effects of communication content overall when stating high vaccination uptake in adolescents or scientific evidence along with age or mentioning cancer prevention. Parents always refusing vaccination remained unsensitive to communication contents.Discussion and conclusionThese original DCE results highlighted the need for tailoring specific HPV vaccination promotion communication in a French Caribbean setting. Contextual features such as sexuality concerns as regard to age and peers’ adhesion to vaccination have to be thoroughly considered. The nationwide HPV vaccination campaign in middle schools should adapt communication in order to raise HPV vaccine uptake in the French Caribbean.

Genetic inbreeding load and its individual prediction for milk yield in French dairy sheep

BackgroundThe magnitude of inbreeding depression depends on the recessive burden of the individual, which can be traced back to the hidden (recessive) inbreeding load among ancestors. However, these ancestors carry different alleles at potentially deleterious loci and therefore there is individual variability of this inbreeding load. Estimation of the additive genetic value for inbreeding load is possible using a decomposition of inbreeding in partial inbreeding components due to ancestors. Both the magnitude of variation in partial inbreeding components and the additive genetic variance of inbreeding loads are largely unknown. Our study had three objectives. First, based on substitution effect under non-random matings, we showed analytically that inbreeding load of an ancestor can be expressed as an additive genetic effect. Second, we analysed the structure of individual inbreeding by examining the contributions of specific ancestors/founders using the concept of partial inbreeding coefficients in three French dairy sheep populations (Basco-Béarnaise, Manech Tête Noire and Manech Tête Rousse). Third, we included these coefficients in a mixed model as random regression covariates, to predict genetic variance and breeding values of the inbreeding load for milk yield in the same breeds.ResultsPedigrees included 190,276, 166,028 and 633,655 animals of Basco-Béarnaise, Manech Tête Noire and Manech Tête Rousse, respectively, born between 1985 and 2021. A fraction of 99.1% of the partial inbreeding coefficients were lower than 0.01 in all breeds, meaning that in practice inbreeding occurs in pedigree loops that span several generations backwards. Less than 5% ancestors generate inbreeding, because mating is essentially between unrelated individuals. Inbreeding load estimations involved 658,731, 541,180 and 2,168,454 records of yearly milk yield from 178,123, 151,863 and 596,586 females in Basco-Béarnaise, Manech Tête Noire and Manech Tête Rousse, respectively. Adding the inbreeding load effect to the model improved the fitting (values of the statistic Likelihood Ratio Test between 132 and 383) for milk yield in the three breeds. The inbreeding load variances were equal to 11,804 and 9435 L squared of milk yield for a fully inbred (100%) descendant in Manech Tête Noire and Manech Tête Rousse. In Basco-Béarnaise, the estimate of the inbreeding load variance (11,804) was not significantly different from zero. The correlations between (direct effect) additive genetic and inbreeding load effects were − 0.09, − 0.08 and − 0.12 in Basco-Béarnaise, Manech Tête Noire and Manech Tête Rousse.ConclusionsThe decomposition of inbreeding in partial coefficients in these populations shows that inbreeding is mostly due to several small contributions of ancestors (lower than 0.001) going back several generations (5 to 7 generations), which is according to the policy of avoiding close matings. There is variation of inbreeding load among animals, although its magnitude does not seem enough to warrant selection based on this criterion.

Combining machine learning and single-cell sequencing to identify key immune genes in sepsis

This research aimed to identify novel indicators for sepsis by analyzing RNA sequencing data from peripheral blood samples obtained from sepsis patients (n = 23) and healthy controls (n = 10). 5148 differentially expressed genes were identified using the DESeq2 technique and 5636 differentially expressed genes were identified by the limma method(|Log2 Fold Change|≥2, FDR < 0.05). A total of 1793 immune-related genes were identified from the ImmPort database, with 358 genes identified in both groups. Next, a Biological association network was constructed, and five key hub genes (CD4, HLA-DOB, HLA-DRB1, HLA-DRA, AHNAK) were identified using a combination of three topological analysis algorithms (MCC, Closeness, and MNC) and four machine learning algorithms (Random Forest, LASSO regression, SVM, and XGBoost). immune cell distribution showed that the key genes correlated with multiple immune cell infiltrations. Gene Set Enrichment Analysis (GSEA) revealed that the key genes involved multiple immune response and inflammation-related signaling pathways. Subsequently, diagnostic models were constructed using four machine learning algorithms (Logistic regression, AdaBoost, KNN, and XGBoost) based on the identified key genes. Models with the highest performance were then selected. Ultimately, single-cell sequencing data revealed that the identified key genes were expressed in various immune cells, while Quantitative PCR (qPCR) tests confirmed their reduced expression in the sepsis group.

Machine learning algorithms for predictive modeling of dyslipidemia-associated cardiovascular disease risk in pregnancy: a comparison of boosting, random forest, and decision tree regression

BackgroundCardiovascular diseases (CVD) are major contributors to maternal mortality and morbidity during pregnancy and increased atherogenic index of plasma levels is associated with a higher risk of CVD and obesity.MethodsIn this study, we utilized three different machine learning algorithms (boosting, random forest, and decision tree regression) to predict dyslipidemia-associated cardiovascular disease using atherogenic index and lipid profile parameters based on a cross-sectional study datasets of 112 pregnant women aged between 15 and 49 conducted at Aminu Kano Teaching Hospital.ResultsThe results showed that random forest regression outperformed both boosting and decision tree regression, recording the lowest error criteria (MSE = 0.071 and RMSE = 0.266) for evaluating the model. These findings indicated that all the three algorithms have the potential to effectively model the data from atherogenic indices and lipid profile parameters but random forest and boosting were found to outperform decision tree models with respective R2 values of 0.95 and 0.92.ConclusionsOverall, the study highlights the accuracy of machine learning models (random forest, boosting, and decision trees) in predicting dyslipidemia-associated cardiovascular diseases and the findings could contribute to the development of effective strategies for the prevention and treatment of dyslipidemia-associated cardiovascular diseases.