XGBoost Machine-learning Model Research Articles

Spinach genomes reveal migration history and candidate genes for important crop traits.

Spinach (Spinacia oleracea) is an important leafy crop possessing notable economic value and health benefits. Current genomic resources include reference genomes and genome-wide association studies. However, the worldwide genetic relationships and the migration history of the crop remained uncertain, and genome-wide association studies have produced extensive gene lists related to agronomic traits. Here, we re-analysed the sequenced genomes of 305 cultivated and wild spinach accessions to unveil the phylogeny and history of cultivated spinach and to explore genetic variation in relation to phenotypes. In contrast to previous studies, we employed machine learning methods (based on Extreme Gradient Boosting, XGBoost) to detect variants that are collectively associated with agronomic traits. Variant-based cluster analyses revealed three primary spinach groups in the Middle East, Asiaand Europe/US. Combining admixture analysis and allele-sharing statistics, migration routes of spinach from the Middle East to Europe and Asia are presented. Using XGBoost machine learning models we predict genomic variants influencing bolting time, flowering time, petiole color, and leaf surface texture and propose candidate genes for each trait. This study enhances our understanding of the history and phylogeny of domesticated spinach and provides valuable information on candidate genes for future genetic improvement of the crop.

Performance Optimization of Bloch Surface Wave based Devices using XGBoost Machine Learning Model

Performance Optimization of Bloch Surface Wave based Devices using XGBoost Machine Learning Model

A machine learning approach to predict daptomycin exposure from two concentrations based on Monte Carlo simulations.

Daptomycin is a concentration-dependent lipopeptide antibiotic for which exposure/effect relationships have been shown. Machine learning (ML) algorithms, developed to predict the individual exposure to drugs, have shown very good performances in comparison to maximum a posteriori Bayesian estimation (MAP-BE). The aim of this work was to predict the area under the blood concentration curve (AUC) of daptomycin from two samples and a few covariates using XGBoost ML algorithm trained on Monte Carlo simulations. Five thousand one hundred fifty patients were simulated from two literature population pharmacokinetics models. Data from the first model were split into a training set (75%) and a testing set (25%). Four ML algorithms were built to learn AUC based on daptomycin blood concentration samples at pre-dose and 1 h post-dose. The XGBoost model (best ML algorithm) with the lowest root mean square error (RMSE) in a 10-fold cross-validation experiment was evaluated in both the test set and the simulations from the second population pharmacokinetic model (validation). The ML model based on the two concentrations, the differences between these concentrations, and five other covariates (sex, weight, daptomycin dose, creatinine clearance, and body temperature) yielded very good AUC estimation in the test (relative bias/RMSE = 0.43/7.69%) and validation sets (relative bias/RMSE = 4.61/6.63%). The XGBoost ML model developed allowed accurate estimation of daptomycin AUC using C0, C1h, and a few covariates and could be used for exposure estimation and dose adjustment. This ML approach can facilitate the conduct of future therapeutic drug monitoring (TDM) studies.

Next-generation non-linear and collapse prediction models for short- to long-period systems via machine learning methods

Since the 1960 s, a cornerstone of earthquake engineering has been estimating the non-linear response of structures based just on lateral strength, modal properties, and the anticipated seismic demand. Over the years, several studies have quantified this empirical relationship and integrated it within seismic design and assessment methodologies. These have been widely accepted for practical application and adopted in building codes worldwide. While these models work reasonably well, there are still areas in which improvements can be made, especially concerning their robust quantification of uncertainty. This is mainly due to the amount of data used to quantify these empirical relationships, the choice of functional forms during the fitting, the model fitting and testing process, and how the ground motion shaking intensity is characterised. This study tackles these issues via the non-linear analysis of single-degree-of-freedom oscillators to train several machine learning (ML) models. This was to examine the accuracy and applicability of such models within a seismic engineering context and explore potential gains in quantifying the non-linear response of structures via next-generation intensity measures, namely average spectral acceleration, Saavg. The results show that the Decision Tree and XGBoost models worked well across a broad range of periods, accurately predicting collapse and non-collapse responses. Appraising these with existing models showed a notable improvement all around. It indicates that the models based on data-driven ML approaches represent a positive step and can be seamlessly integrated with seismic analysis methodologies utilised worldwide. Subsequently, a Python-based library of these XGBoost ML models was developed and made available online to foster practical application.

Evaluation of bone marrow invasion on the machine learning of 18 F-FDG PET texture analysis in lower gingival squamous cell carcinoma.

Lower gingival squamous cell carcinoma (LGSCC) has the potential to invade the alveolar bone. Traditionally, the diagnosis of LGSCC relied on morphological imaging, but inconsistencies between these assessments and surgical findings have been observed. This study aimed to assess the correlation between LGSCC bone marrow invasion and PET texture features and to enhance diagnostic accuracy by using machine learning. A retrospective analysis of 159 LGSCC patients with pretreatment 18 F-fluorodeoxyglucose (FDG) PET/computed tomography (CT) examination from 2009 to 2017 was performed. We extracted radiomic features from the PET images, focusing on pathologic bone marrow invasion detection. Extracted features underwent the least absolute shrinkage and selection operator algorithm-based selection and were then used for machine learning via the XGBoost package to distinguish bone marrow invasion presence. Receiver operating characteristic curve analysis was performed. From the 159 patients, 88 qualified for further analysis (59 men; average age, 69.2 years), and pathologic bone marrow invasion was identified in 69 (78%) of these patients. Three significant radiological features were identified: Gray level co-occurrence matrix_Correlation, INTENSITY-BASED_IntensityInterquartileRange, and MORPHOLOGICAL_SurfaceToVolumeRatio. An XGBoost machine-learning model, using PET radiomic features to detect bone marrow invasion, yielded an area under the curve value of 0.83. Our findings highlighted the potential of 18 F-FDG PET radiomic features, combined with machine learning, as a promising avenue for improving LGSCC diagnosis and treatment. Using 18 F-FDG PET texture features may provide a robust and accurate method for determining the presence or absence of bone marrow invasion in LGSCC patients.

DNA Methylation Profiling Enables Accurate Classification of Nonductal Primary Pancreatic Neoplasms

Background and aimsCytological and histopathological diagnosis of non-ductal pancreatic neoplasms can be challenging in daily clinical practice while it is crucial for therapy and prognosis. The cancer methylome is successfully used as a diagnostic tool in other cancer entities. Here, we investigate if methylation profiling can improve the diagnostic work-up of pancreatic neoplasms. MethodsDNA methylation data were obtained for 301 primary tumors spanning six primary pancreatic neoplasms and 20 normal pancreas controls. Neural Network, Random Forest, and XGBoost machine learning models were trained to distinguish between tumor types. Methylation data of 29 non-pancreatic neoplasms (n = 3708) were used to develop an algorithm capable of detecting neoplasms of non-pancreatic origin. ResultsAfter benchmarking three state-of-the-art machine learning models, the Random Forest model emerged as the best classifier with 96.9% accuracy. All classifications received a probability score reflecting the confidence of the prediction. Increasing the score threshold, improved the Random Forest classifier performance up to 100% with 87% of samples with scores surpassing the cutoff. Using a logistic regression model, detection of non-pancreatic neoplasms achieved an area under the curve (AUC) of > 0.99. Analysis of biopsy specimens showed concordant classification with their paired resection sample. ConclusionPancreatic neoplasms can be classified with high accuracy based on DNA methylation signatures. Additionally, non-pancreatic neoplasms are identified with near perfect precision. In summary, methylation profiling can serve as a valuable adjunct in the diagnosis of pancreatic neoplasms with minimal risk for misdiagnosis, even in the pre-operative setting.

Utilizing Electronic Dental Records to Predict Neuro-Degenerative Diseases in a Dental Setting: A Pilot Study.

Limited research demonstrates the possible correlations between dental diseases and neurodegenerative diseases like Alzheimer's disease (AD) and Parkinson's disease (PD). Nevertheless, dental diseases are often overlooked while assessing the risk of AD and PD in clinical settings. It is unknown whether AD/PD risk can be predicted using electronic dental record (EDR) data collected in a routine dental setting. This pilot study determined the feasibility of predicting AD/PD using 84 features routinely captured in the EDR. We utilized the Temple University School of Dentistry clinic data of 27,138 patients. Using a natural language processing (NLP) approach (accuracy=97%), we identified patients with AD/PD and their matched controls (matched by age and gender). XGBoost machine learning model with 10-fold cross-validation was applied for prediction. With 77% accuracy, we found 53 features significantly associated with AD/PD that could be utilized to predict the risk of AD/PD. Further studies are warned to confirm these findings.

Utilizing a novel high-resolution malaria dataset for climate-informed predictions with a deep learning transformer model

Climatic factors influence malaria transmission via the effect on the Anopheles vector and Plasmodium parasite. Modelling and understanding the complex effects that climate has on malaria incidence can enable important early warning capabilities. Deep learning applications across fields are proving valuable, however the field of epidemiological forecasting is still in its infancy with a lack of applied deep learning studies for malaria in southern Africa which leverage quality datasets. Using a novel high resolution malaria incidence dataset containing 23 years of daily data from 1998 to 2021, a statistical model and XGBOOST machine learning model were compared to a deep learning Transformer model by assessing the accuracy of their numerical predictions. A novel loss function, used to account for the variable nature of the data yielded performance around + 20% compared to the standard MSE loss. When numerical predictions were converted to alert thresholds to mimic use in a real-world setting, the Transformer’s performance of 80% according to AUROC was 20–40% higher than the statistical and XGBOOST models and it had the highest overall accuracy of 98%. The Transformer performed consistently with increased accuracy as more climate variables were used, indicating further potential for this prediction framework to predict malaria incidence at a daily level using climate data for southern Africa.

Enhanced formation of methane hydrates via graphene oxide: Machine learning insights from molecular dynamics simulations

Gas hydrates, with their significant applications in energy and environment sectors, have emerged as promising technologies for gas separation, natural gas and energy storage. The formation of methane hydrates occurs in substrate-contact conditions, where the properties of the substrate surfaces play a crucial role in determining hydrate formation pathways. Herein, the influence of the hydroxyl group (-OH) content on the graphene surface on methane hydrate formation in confined systems using molecular dynamics (MD) simulations with machine learning (ML) technique. The MD results show that increasing the oxidation degree effectively enhances the formation of methane hydrate within the oxidation degree of 10 %–40 %. This enhancement is attributed to the hydrophilicity of the (GO) surfaces. When the oxidation degree reaches 50 %, however, the formation rate of methane hydrates slows down compared to the GO-40 system. This deceleration can be attributed to the formation of dense water layers that adhere to the GO surfaces, resulting in impeding the hydrate formation process. Markov State Model (MSM) is employed to analyze the induction time of GO on methane hydrate formation, highlighting the important role of GO in the formation of methane hydrates. Additionally, an eXtreme Gradient Boosting (XGboost) ML model is developed to predict the formation behaviors of methane hydrates. Ternary water-ring aggregations (TWRAs) are adopted as indicators to understand to the formation behaviors, and the XGboost ML model facilitate the identification of key TWRAs associated with the formation process. This work provides molecular-level insights into the influence of –OH group on formation of methane hydrates, and offers a predictive framework for understanding and characterizing the formation clathrate hydrates in confined systems.

Classification of tumor types using XGBoost machine learning model: a vector space transformation of genomic alterations

BackgroundMachine learning (ML) represents a powerful tool to capture relationships between molecular alterations and cancer types and to extract biological information. Here, we developed a plain ML model aimed at distinguishing cancer types based on genetic lesions, providing an additional tool to improve cancer diagnosis, particularly for tumors of unknown origin.MethodsTCGA data from 9,927 samples spanning 32 different cancer types were downloaded from cBioportal. A vector space model type data transformation technique was designed to build consistently homogeneous new datasets containing, as predictive features, calls for somatic point mutations and copy number variations at chromosome arm-level, thus allowing the use of the XGBoost classifier models. Considering the imbalance in the dataset, due to large difference in the number of cases for each tumor, two preprocessing strategies were considered: i) setting a percentage cut-off threshold to remove less represented cancer types, ii) dividing cancer types into different groups based on biological criteria and training a specific XGBoost model for each of them. The performance of all trained models was mainly assessed by the out-of-sample balanced accuracy (BACC) and the AUC scores.ResultsThe XGBoost classifier achieved the best performance (BACC 77%; AUC 97%) on a dataset containing the 10 most represented tumor types. Moreover, dividing the 18 most represented cancers into three different groups (endocrine-related carcinomas, other carcinomas and other cancers),such analysis models achieved 78%, 71% and 86% BACC, respectively, with AUC scores greater than 96%. In addition, the model capable of linking each group to a specific cancer type reached 81% BACC and 94% AUC. Overall, the diagnostic potential of our model was comparable/higher with respect to others already described in literature and based on similar molecular data and ML approaches.ConclusionsA boosted ML approach able to accurately discriminate different cancer types was developed. The methodology builds datasets simpler and more interpretable than the original data, while keeping enough information to accurately train standard ML models without resorting to sophisticated Deep Learning architectures. In combination with histopathological examinations, this approach could improve cancer diagnosis by using specific DNA alterations, processed by a replicable and easy-to-use automated technology. The study encourages new investigations which could further increase the classifier’s performance, for example by considering more features and dividing tumors into their main molecular subtypes.

Machine learning and decision making in aortic arch repair

BackgroundDecision making during aortic arch surgery regarding cannulation strategy and nadir temperature are important in reducing risk, and there is a need to determine the best individualized strategy in a data-driven fashion. Using machine learning (ML), we modeled the risk of death or stroke in elective aortic arch surgery based on patient characteristics and intraoperative decisions. MethodsThe study cohort comprised 1323 patients from 9 institutions who underwent an elective aortic arch procedure between 2002 and 2021. A total of 69 variables were used in developing a logistic regression and XGBoost ML model trained for binary classification of mortality and stroke. Shapely additive explanations (SHAP) values were studied to determine the importance of intraoperative decisions. ResultsDuring the study period, 3.9% of patients died and 5.4% experienced stroke. XGBoost (area under the curve [AUC], 0.77 for death, 0.87 for stroke) demonstrated better discrimination than logistic regression (AUC, 0.65 for death, 0.75 for stroke). From SHAP analysis, intraoperative decisions are 3 of the top 20 predictors of death and 6 of the top 20 predictors of stroke. Predictor weights are patient-specific and reflect the patient's preoperative characteristics and other intraoperative decisions. Patient-level simulation also demonstrates the variable contribution of each decision in the context of the other choices that are made. ConclusionsUsing ML, we can more accurately identify patients at risk of death and stroke, as well as the strategy that better reduces the risk of adverse events compared to traditional prediction models. Operative decisions made may be tailored based on a patient's specific characteristics, allowing for maximized, personalized benefit.

P071 Simple Vestibular-Occular Motor Assessment as a Predictor of Driving Performance Vulnerability following extended Wakefulness

Abstract Introduction Driver fatigue is a significant contributor to road crashes, but identifying individuals at driving risk is challenging. We examined the potential of simple baseline vestibular ocular motor system (VOMS) assessment via virtual reality goggles to predict subsequent vulnerability to driving simulator impairment following extended wakefulness. Methods 49 individuals (Mean±SD Age 32.6±12.9, 45% Males) underwent 9hr baseline sleep opportunity followed by approximately ~29hrs extended wakefulness with five 60min driving assessments. Cluster analysis classified drivers into vulnerable (n=17) or resistant (n=32) groups based on their worst steering deviation and number of crashes from driving tests. Baseline VOMS were performed ~10mins prior to the first three drives (1, 7 and 13hrs of wakefulness). XGBoost machine learning model was trained using baseline VOMs features to predict vulnerable vs resistant groups from driving tests 4 and 5 (19 and 25hrs of wakefulness) Model performance was evaluated using 5-fold cross-validation approach using ROC analysis. Results XGBoost machine learning ranked all 70 VOMS metrics on their importance in predicting vulnerable vs resistant groups. Top 10 VOMs metrics assessed during baseline non-sleep deprived tests demonstrated a strong ability to predict the driver's performance following extended wakefulness, differentiating between the vulnerable vs resistant groups (AUC 0.73 [95%CI 0.61-0.83, p<0.001]). Conclusion VOMs tests conducted at baseline holds promise for predicting future driving impairment. This approach has the potential to be highly valuable in determining an individual's fitness to drive. Future validation in independent samples, sleep disordered population and in-field on-road testing are needed to confirm these promising findings.

O072 Simple Vestibular-Occular Motor Assessment as a Predictor of Alertness State and Driving Impairment during Extended Wakefulness

Abstract Introduction Driver fatigue contributes to 2-16% of road crashes, highlighting the need for improved detection of at-risk drivers. We used a novel and brief test of vestibular ocular motor system (VOMS) assessed via virtual reality goggles to predict alertness state and driving simulator performance during 29hr extended wakefulness. Methods 49 individuals (Mean±SD Age, 32.6±12.9, 45% Males) undergone 9hr baseline sleep opportunity followed by ~29hrs extended wakefulness with five 60min driving assessments. Cluster analysis, combining steering deviation and number of crashes were used to split participants into groups of either poor vs good driving performance. VOMS assessment was conducted using virtual reality goggles approximately 10mins before and after each drive. Predictive importance of VOMs metrics were ranked using XGBoost machine learning model, which was then utilized to distinguish between poor vs good driving groups. Model performance was evaluated using a 5-fold cross-validation approach using ROC analysis. Results XGBoost machine learning ranked all 70 VOMS metrics on their importance in predicting driving performance group for each drive. Top 10 metrics from pre-drive VOMS test predicted both daytime driving (tests 1-3, AUC 0.8 [95%CI 0.64-0.93], p<0.001) and night-time driving (tests 4-5, AUC 0.78 [95%CI 0.6-0.95, p<0.001]). Post-driving VOMS assessments also predicted daytime (AUC 0.74 [95%CI 0.53-0.9, p<0.001] and night-time driving (AUC 0.76 [95%CI 0.52-0.94, p<0.001]). Conclusion VOMS assessment show promise as a short and effective assessment of sleepiness to predict driving failure. Future validation in independent samples, sleep disordered population and in-field on-road testing are needed to confirm these promising findings.

Ecosystem Evapotranspiration Partitioning and Its Spatial–Temporal Variation Based on Eddy Covariance Observation and Machine Learning Method

Partitioning evapotranspiration (ET) into vegetation transpiration (T) and soil evaporation (E) is challenging, but it is key to improving the understanding of plant water use and changes in terrestrial ecosystems. Considering that the transpiration of vegetation at night is minimal and can be negligible, we established a machine learning model (i.e., extreme gradient boosting algorithm (XGBoost)) for soil evaporation estimation based on night-time evapotranspiration observation data from eddy covariance towers, remote sensing data, and meteorological reanalysis data. Daytime T was consequently calculated as the difference between the total evapotranspiration and predicted daytime soil evaporation. The soil evaporation estimation model was validated based on the remaining night-time ET data (i.e., model test dataset), the non-growing season ET data of the natural ecosystem, and ET data during the fallow periods of croplands. The validation results showed that XGBoost had a better performance in E estimation, with the average overall accuracy of NSE 0.657, R 0.806, and RMSE 11.344 W/m2. The average annual T/ET of the examined ten ecosystems was 0.50 ± 0.08, with the highest value in deciduous broadleaf forests (0.68 ± 0.11), followed by mixed forests (0.61 ± 0.04), and the lowest in croplands (0.40 ± 0.08). We further examined the impact of the leaf area index (LAI) and vapor pressure deficit (VPD) on the variation in T/ET. Overall, at the interannual scale, LAI contributed 28% to the T/ET variation, while VPD had a small (5%) influence. On a seasonal scale, LAI also exerted a stronger impact (1~90%) on T/ET compared to VPD (1~77%). Our study suggests that the XGBoost machine learning model has good performance in ET partitioning, and this method is mainly data-driven without prior knowledge, which may provide a simple and valuable method in global ET partitioning and T/ET estimation.

Ensemble machine learning framework for daylight modelling of various building layouts

AbstractThe application of machine learning (ML) modelling in daylight prediction has been a promising approach for reliable and effective visual comfort assessment. Although many advancements have been made, no standardized ML modelling framework exists in daylight assessment. In this study, 625 different building layouts were generated to model useful daylight illuminance (UDI). Two state-of-the-art ML algorithms, eXtreme Gradient Boosting (XGBoost) and random forest (RF), were employed to analyze UDI in four categories: UDI-f (fell short), UDI-s (supplementary), UDI-a (autonomous), and UDI-e (exceeded). A feature (internal finish) was introduced to the framework to better reflect real-world representation. The results show that XGBoost models predict UDI with a maximum accuracy of R2 = 0.992. Compared to RF, the XGBoost ML models can significantly reduce prediction errors. Future research directions have been specified to advance the proposed framework by introducing new features and exploring new ML architectures to standardize ML applications in daylight prediction.

Machine Learning Predicts Patients With New-onset Diabetes at Risk of Pancreatic Cancer.

New-onset diabetes represent a high-risk cohort to screen for pancreatic cancer. Develop a machine model to predict pancreatic cancer among patients with new-onset diabetes. A retrospective cohort of patients with new-onset diabetes was assembled from multiple health care networks in the United States. An XGBoost machine learning model was designed from a portion of this cohort (the training set) and tested on the remaining part of the cohort (the test set). Shapley values were used to explain the XGBoost's model features. Model performance was compared with 2 contemporary models designed to predict pancreatic cancer among patients with new-onset diabetes. In the test set, the XGBoost model had an area under the curve of 0.80 (0.76 to 0.85) compared with 0.63 and 0.68 for other models. Using cutoffs based on the Youden index, the sensitivity of the XGBoost model was 75%, the specificity was 70%, the accuracy was 70%, the positive predictive value was 1.2%, and the negative predictive value was >99%. The XGBoost model obtained a positive predictive value of at least 2.5% with a sensitivity of 38%. The XGBoost model was the only model that detected at least 50% of patients with cancer one year after the onset of diabetes. All 3 models had similar features that predicted pancreatic cancer, including older age, weight loss, and the rapid destabilization of glucose homeostasis. Machine learning models isolate a high-risk cohort from those with new-onset diabetes at risk for pancreatic cancer.

What would other emergency stroke teams do? Using explainable machine learning to understand variation in thrombolysis practice.

The aim of this work was to understand between-hospital variation in thrombolysis use among emergency stroke admissions in England and Wales. A total of 88,928 patients who arrived at all 132 emergency stroke hospitals in England Wales within 4 h of stroke onset, from 2016 to 2018. Machine learning was applied to the Sentinel Stroke National Audit Programme (SSNAP) data set, to learn which patients in each hospital would likely receive thrombolysis. We used XGBoost machine learning models, coupled with a SHAP model for explainability; Shapley (SHAP) values, providing estimates of how patient features, and hospital identity, influence the odds of receiving thrombolysis. Thrombolysis use in patients arriving within 4 h of known or estimated stroke onset ranged 7% -49% between hospitals. The odds of receiving thrombolysis reduced 9-fold over the first 120 min of arrival-to-scan time, varied 30-fold with stroke severity, reduced 3-fold with estimated rather than precise stroke onset time, fell 6-fold with increasing pre-stroke disability, fell 4-fold with onset during sleep, fell 5-fold with use of anticoagulants, fell 2-fold between 80 and 110 years of age, reduced 3-fold between 120 and 240 min of onset-to-arrival time and varied 13-fold between hospitals. The majority of between-hospital variance was explained by the hospital, rather than the differences in local patient populations. Using explainable machine learning, we identified that the majority of the between-hospital variation in thrombolysis use in England and Wales may be explained by differences in in-hospital processes and differences in attitudes to judging suitability for thrombolysis.

Classifying interpersonal synchronization states using a data-driven approach: implications for social interaction understanding

This study presents a data-driven approach to identifying interpersonal motor synchrony states by analyzing hand movements captured from a 3D depth camera. Utilizing a single frame from the experiment, an XGBoost machine learning model was employed to differentiate between spontaneous and intentional synchrony modes with nearly 90% accuracy. Our findings demonstrate a consistent pattern across subjects, revealing that movement velocity tends to be slower in synchrony modes. These insights support the notion that the relationship between velocity and synchrony is influenced by the cognitive load required for the task, with slower movements leading to higher synchrony in tasks demanding higher cognitive load. This work not only contributes to the limited literature on algorithms for identifying interpersonal synchrony but also has potential implications for developing new metrics to assess real-time human social interactions, understanding social interaction, and diagnosing and developing treatment strategies for social deficits associated with conditions such as Autism Spectrum Disorder.

Sales Prediction of Walmart Sales Based on OLS, Random Forest, and XGBoost Models

The technique of estimating future sales levels for a good or service is known as sales forecasting. The corresponding forecasting methods range from initially qualitative analysis to later time series methods, regression analysis and econometric models, as well as machine learning methods that have emerged in recent decades. This paper compares the different performances of OLS, Random Forest and XGBoost machine learning models in predicting the sales of Walmart stores. According to the analysis, XGBoost model has the best sales forecasting ability. In the case of logarithmic sales, R2 of the XGBoost model is as high as 0.984, while MSE and MAE are only 0.065 and 0.124, respectively. The XGBoost model is therefore an option when making sales forecasts. These results compare different types of models, find out the best prediction model, and provide suggestions for future prediction model selection.

Identification and Validation of Anoikis-Related Signatures for Predicting Prognosis in Lung Adenocarcinoma with Machine Learning.

Lung adenocarcinoma (LUAD) is an aggressive cancer that has an extremely poor prognosis. As well as facilitating the detachment of cancer cells from the primary tumor site, anoikis plays an important role in cancer metastasis. Few studies to date, however, have examined the role of anoikis in LUAD, in patient prognosis. A total of 316 anoikis-related genes (ANRGs) integrated from Genecards and Harmonizome portals. LUAD transcriptome data were retrieved from the Genotype-Tissue Expression Project (GEO) and The Cancer Genome Atlas (TCGA). Anoikis-related prognostic genes (ANRGs) were primarily screened by univariate Cox regression. All ANRGs were included in the Least Absolute Shrinkage and Selection Operator (LASSO) Cox regression model to construct the powerful prognostic signature. This signature was validated and assessed using the Kaplan-Meier method as well as univariate and multivariate Cox regression analyses. Anoikis-related regulators of risk score were identified using a XG-boost machine learning model. The expression of ITGB4 protein was examined in a ZhengZhou University (ZZU) tissue cohort by immunohistochemistry, and the potential mechanisms of action of ITGB4 in LUAD were explored by GO, KEGG, and ingenuity pathway analyses and by GSEA. A risk score signature was constructed based on eight ANRGs, with high risk scores found to closely correlate with unfavorable clinical features. ITGB4 expression may be associated with 5-year over survival, with immunohistochemistry showed that the expression of ITGB4 was higher in LUAD than in nontumor tissues. Enrichment analysis suggested that ITGB4 may promote LUAD development by targeting E2F, MYC, and oxidative phosphorylation signaling pathways. Our anoikis-related signature from RNA-seq data may be a novel prognostic biomarker in patients with LUAD. It may help physicians develop personalized LUAD treatments in clinical practice. Moreover, ITGB4 may affect the development of LUAD through the oxidative phosphorylation pathway.