Advancing risk factor identification for pediatric lobar pneumonia: the promise of machine learning technologies

Frontiers in Pediatrics, Journal Year: 2025, Volume and Issue: 13

Published: March 7, 2025

Background Community-acquired pneumonia (CAP) is a prevalent pediatric condition, and lobar (LP) considered severe subtype. Early identification of LP crucial for appropriate management. This study aimed to develop compare machine learning models predict in children with CAP. Methods A total 25 clinical laboratory variables were collected. Missing data (<2%) imputed, the dataset was split into training (60%) validation (40%) sets. Univariable logistic regression Boruta feature selection used identify significant predictors. Four algorithms-Logistic Regression (LR), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Decision Tree (DT)-were compared using area under curve (AUC), balanced accuracy, sensitivity, specificity, F1 score. SHAP analysis performed interpret best-performing model. Results 278 patients CAP included this study, whom 65 diagnosed LP. The XGBoost model demonstrated best performance an AUC 0.880 (95% CI: 0.807–0.934) set 0.746 0.664–0.843) set. identified age, CRP, CD64 index, lymphocyte percentage, ALB as top five predictive factors. Conclusion showed superior predicting enabled early diagnosis risk assessment LP, thereby facilitating decision-making.

Language: Английский

---

OWL: an optimized and independently validated machine learning prediction model for lung cancer screening based on the UK Biobank, PLCO, and NLST populations

EBioMedicine, Journal Year: 2023, Volume and Issue: 88, P. 104443 - 104443

Published: Jan. 25, 2023

A reliable risk prediction model is critically important for identifying individuals with high of developing lung cancer as candidates low-dose chest computed tomography (LDCT) screening. Leveraging a cutting-edge machine learning technique that accommodates wide list questionnaire-based predictors, we sought to optimize and validate model.

Language: Английский

---

Glutamine and amino acid metabolism as a prognostic signature and therapeutic target in endometrial cancer

Cancer Medicine, Journal Year: 2023, Volume and Issue: 12(15), P. 16337 - 16358

Published: June 30, 2023

Abstract Introduction Endometrial cancer (EC) is the most common female reproductive system in developed countries with growing incidence and associated mortality, which may be due to prevalence of obesity. Metabolism reprogramming including glucose, amino acid, lipid remodeling a hallmark tumors. Glutamine metabolism has been reported participate tumor proliferation development. This study aimed develop glutamine metabolism‐related prognostic model for EC explore potential targets treatment. Method Transcriptomic data survival outcome were retrieved from The Cancer Genome Atlas (TCGA). Differentially expressed genes related recognized utilized build by univariate multivariate Cox regressions. was confirmed training, testing, entire cohort. A nomogram combing clinicopathologic features established tested. Moreover, we explored effect key metabolic enzyme, PHGDH, on biological behavior cell lines xenograft model. Results Five genes, OTC, ASRGL1, ASNS, NR1H4, involved construction. Kaplan–Meier curve suggested that patients as high risk underwent inferior outcomes. receiver operating characteristic (ROC) showed sufficient predict survival. Enrichment analysis DNA replication repair dysfunction high‐risk whereas immune relevance revealed low scores group. Finally, integrating clinical factors created verified. Further, knockdown PHGDH growth inhibition, increasing apoptosis, reduced migration. Promisingly, NCT‐503, inhibitor, significantly repressed vivo ( p = 0.0002). Conclusion Our work validated favorably evaluates prognosis patients. crucial point linked metabolism, acid progression. High‐risk stratified not therapy. might target links serine well

Language: Английский

---

Multidimensional analysis of immune cells from COVID-19 patients identified cell subsets associated with the severity at hospital admission

PLoS Pathogens, Journal Year: 2023, Volume and Issue: 19(6), P. e1011432 - e1011432

Published: June 13, 2023

SARS-CoV-2 emerged as a new coronavirus causing COVID-19, and it has been responsible for more than 760 million cases 6.8 deaths worldwide until March 2023. Although infected individuals could be asymptomatic, other patients presented heterogeneity wide range of symptoms. Therefore, identifying those being able to classify them according their expected severity help target health efforts effectively.Therefore, we wanted develop machine learning model predict who will severe disease at the moment hospital admission. We recruited 75 analysed innate adaptive immune system subsets by flow cytometry. Also, collected clinical biochemical information. The objective study was leverage techniques identify features associated with progression. Additionally, sought elucidate specific cellular involved in following onset Among several models tested, found that Elastic Net better score modified WHO classification. This 72 out individuals. Besides, all revealed CD38+ Treg CD16+ CD56neg HLA-DR+ NK cells were highly correlated severity.The stratify uninfected COVID-19 from asymptomatic patients. On hand, these here understand induction progression symptoms

Language: Английский

---

Evaluating Machine Learning Stability in Predicting Depression and Anxiety Amidst Subjective Response Errors

Healthcare, Journal Year: 2024, Volume and Issue: 12(6), P. 625 - 625

Published: March 10, 2024

Major Depressive Disorder (MDD) and Generalized Anxiety (GAD) pose significant burdens on individuals society, necessitating accurate prediction methods. Machine learning (ML) algorithms utilizing electronic health records survey data offer promising tools for forecasting these conditions. However, potential bias inaccuracies inherent in subjective responses can undermine the precision of such predictions. This research investigates reliability five prominent ML algorithms—a Convolutional Neural Network (CNN), Random Forest, XGBoost, Logistic Regression, Naive Bayes—in predicting MDD GAD. A dataset rich biomedical, demographic, self-reported information is used to assess algorithms’ performance under different levels response inaccuracies. These simulate scenarios with memory recall interpretations. While all demonstrate commendable accuracy high-quality data, their diverges significantly when encountering erroneous or biased responses. Notably, CNN exhibits superior resilience this context, maintaining even achieving enhanced accuracy, Cohen’s kappa score, positive both highlights CNN’s ability handle unreliability, making it a potentially advantageous choice mental conditions based data. findings underscore critical importance algorithmic prediction, particularly relying They emphasize need careful algorithm selection contexts, emerging as candidate due its robustness improved uncertainties.

Language: Английский

---

Comparison of machine learning and nomogram to predict 30-day in-hospital mortality in patients with acute myocardial infarction combined with cardiogenic shock: a retrospective study based on the eICU-CRD and MIMIC-IV databases

BMC Cardiovascular Disorders, Journal Year: 2025, Volume and Issue: 25(1)

Published: March 19, 2025

To evaluate the predictive utility of machine learning and nomogram in predicting in-hospital mortality patients with acute myocardial infarction complicated by cardiogenic shock (AMI-CS), to visualize model results order analyze impact these predictors on patients' prognosis. A retrospective analysis was conducted 332 adult who were diagnosed AMI-CS admitted ICU for first time within eICU Collaborative Research Database (eICU-CRD). AdaBoost, XGBoost, LightGBM, Random Forest logistic regression developed utilizing random forest recursive elimination (RF-RFE) least absolute shrinkage selection operator (LASSO) algorithms feature selection. Compared models, demonstrated superior accuracy AMI-CS, an AUC value 0.869 (95% CI: 0.803, 0.883) F1 score 0.897 internal test set nomogram, 0.770 0.702, 0.801) 0.832 external validation set. Nomogram enhance interpretability transparency leading more reliable prognostic predictions patients. This facilitates clinicians making precise decisions, thereby enhancing patient

Language: Английский

---

A multicenter study on developing a prognostic model for severe fever with thrombocytopenia syndrome using machine learning

Frontiers in Microbiology, Journal Year: 2025, Volume and Issue: 16

Published: March 19, 2025

Background Severe Fever with Thrombocytopenia Syndrome (SFTS) is a disease caused by infection the virus (SFTSV), novel Bunyavirus. Accurate prognostic assessment crucial for developing individualized prevention and treatment strategies. However, machine learning models SFTS are rare need further improvement clinical validation. Objective This study aims to develop validate an interpretable model based on (ML) methods enhance understanding of progression. Methods multicenter retrospective analyzed patient data from two provinces in China. The derivation cohort included 292 patients treated at Second Hospital Nanjing January 2022 December 2023, 7:3 split training internal external validation consisted 104 First Affiliated Wannan Medical College during same period. Twenty-four commonly available features were selected, Boruta algorithm identified 12 candidate predictors, ranked Z-scores, which progressively incorporated into 10 models. Model performance was assessed using area under receiver-operating-characteristic curve (AUC), accuracy, recall, F1 score. utility best-performing evaluated through decision analysis (DCA) net benefit. Robustness tested 10-fold cross-validation, feature importance explained SHapley Additive exPlanation (SHAP) both globally locally. Results Among models, XGBoost demonstrated best overall discriminatory ability. Considering AUC index simplicity, final 7 key constructed. showed high predictive accuracy outcomes (AUC = 0.911, 95% CI: 0.842–0.967) validations 0.891, 0.786–0.977). A tool this has been developed implemented Streamlit framework. Conclusion XGBoost-based shows translated tool. model's serve as valuable indicators early prognosis SFTS, warranting close attention healthcare professionals practice.

Language: Английский

---

Identifying the Most Critical Predictors of Workplace Violence Experienced by Junior Nurses: An Interpretable Machine Learning Perspective

Journal of Nursing Management, Journal Year: 2025, Volume and Issue: 2025(1)

Published: Jan. 1, 2025

Background: Workplace violence, defined as any disruptive behavior or threat to employees, seriously threatens junior nurses. Compared with senior nurses, nurses are more vulnerable workplace violence due inexperience, low professional recognition, and limited mental resilience. However, there is an absence of research discussing the risk in particular, lack analysis critical factors within multiple influences targeted prediction models. Objective: Considering influencing faced by this study aims predict using interpretable machine learning models identify their nonlinear effects. Design: An observational, cross-sectional design. Participants: A total 5663 registered 90 tertiary hospitals Sichuan Province, China. Methods: Data all obtained through a questionnaire survey. framework, including Light Gradient Boosting Machine (LightGBM) model two post hoc methods, Accumulate Local Effect SHapely Additive exPlanations (SHAP), conjoined. Results: The LightGBM accurate than other achieving area under receiver operating characteristic curve 0.761 Brier score 0.198 on task. Among dozens potential input into predictive model, seeing medical complaints, psychological demands, identity, etc., most predictors violence. Conclusions: proposed LightGBM-SHAP-ALE approach dynamically effectively identifies at high providing foundation for timely detection intervention.

Language: Английский

---

XGBoost-based machine learning test improves the accuracy of hemorrhage prediction among geriatric patients with long-term administration of rivaroxaban

BMC Geriatrics, Journal Year: 2023, Volume and Issue: 23(1)

Published: July 10, 2023

Abstract Background Hemorrhage is a potential and serious adverse drug reaction, especially for geriatric patients with long-term administration of rivaroxaban. It essential to establish an effective model predicting bleeding events, which could improve the safety rivaroxaban use in clinical practice. Methods The hemorrhage information 798 (over age 70 years) who needed anticoagulation therapy was constantly tracked recorded through well-established follow-up system. Relying on 27 collected indicators these patients, conventional logistic regression analysis, random forest XGBoost-based machine learning approaches were applied analyze hemorrhagic risk factors corresponding prediction models. Furthermore, performance models tested compared by area under curve (AUC) receiver operating characteristic (ROC) curve. Results A total 112 (14.0%) had events after treatment more than 3 months. Among them, 96 gastrointestinal intracranial during treatment, accounted 83.18% events. regression, XGBoost established AUCs 0.679, 0.672 0.776, respectively. showed best predictive terms discrimination, accuracy calibration among all Conclusion An good discrimination built predict rivaroxaban, will facilitate individualized patients.

Language: Английский

---

Predicting Co-Occurring Mental Health and Substance Use Disorders in Women: An Automated Machine Learning Approach

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(4), P. 1630 - 1630

Published: Feb. 18, 2024

Significant clinical overlap exists between mental health and substance use disorders, especially among women. The purpose of this research is to leverage an AutoML (Automated Machine Learning) interface predict distinguish co-occurring (MH) disorders (SUD) By employing various modeling algorithms for binary classification, including Random Forest, Gradient Boosted Trees, XGBoost, Extra SGD, Deep Neural Network, Single-Layer Perceptron, K Nearest Neighbors (grid), a super learning model (constructed by combining the predictions Forest XGBoost model), aims provide healthcare practitioners with powerful tool earlier identification, intervention, personalised support women at risk. present presents machine (ML) methodology more accurately predicting co-occurrence in women, utilising Treatment Episode Data Set Admissions (TEDS-A) from year 2020 (n = 497,175). A was constructed model. demonstrated promising predictive performance MH SUD AUC 0.817, Accuracy 0.751, Precision 0.743, Recall 0.926 F1 Score 0.825. accurate prediction models can substantially facilitate prompt identification implementation intervention strategies.

Language: Английский

---