Publications by authors named "I D KUZNETSOV"
Ultrasonographic assessment of the diameters of various veins and their indices are among the most applied diagnostic tools for evaluating fluid responsiveness in clinical practice. Despite their widespread use, there is no definitive answer on which is preferable. Our study aimed to investigate the diagnostic accuracy of different venous diameters and their indices to assess fluid responsiveness.
View Article and Find Full Text PDFThe course of COVID-19 is highly variable, with genetics playing a significant role. Through large-scale genetic association studies, a link between single nucleotide polymorphisms and disease susceptibility and severity was established. However, individual single nucleotide polymorphisms identified thus far have shown modest effects, indicating a polygenic nature of this trait, and individually have limited predictive performance.
View Article and Find Full Text PDFArticle Synopsis
- The study aims to evaluate if machine learning can effectively predict early outcomes of laparoscopic nephron-sparing surgery (NSS) for kidney tumors, taking the surgeons' experience into account.
- Analyzing data from 320 surgery cases by four surgeons, the research utilizes eXtreme Gradient Boosting and the SHAP method to identify key factors influencing surgical success, including patient demographics, tumor characteristics, and the surgical learning curve.
- Results show that the SHAP method offers valuable insights from the machine learning model, highlighting the importance of certain new features in predicting surgical outcomes such as procedure duration and postoperative kidney function.
Objective: To systematically evaluate the impact of the Trendelenburg position on hemodynamic parameters in adult patients.
Design: Systematic literature review and meta-analysis using PubMed and Medline.
Setting: All prospective interventional studies comparing the hemodynamic characteristics of patients in the horizontal supine position and Trendelenburg position.
Article Synopsis
- Machine learning (ML) is increasingly important for predicting sepsis, a critical condition where timely intervention is crucial; this study explores the effectiveness of ML models in clinical settings.
- Researchers analyzed 73 studies involving nearly 458,000 patients and found that ML models, particularly Neural Networks and Decision Trees, performed better than traditional scoring systems, achieving a pooled AUC of 0.825.
- The study emphasizes the need for standardized practices in reporting and validating ML models to enhance their clinical application and effectiveness in predicting sepsis across diverse patient populations.