Prognostic value of glycaemic variability for mortality in critically ill atrial fibrillation patients and mortality prediction model using machine learning

Cardiovascular Diabetology, Journal Year: 2024, Volume and Issue: 23(1)

Published: Nov. 26, 2024

Abstract Background The burden of atrial fibrillation (AF) in the intensive care unit (ICU) remains heavy. Glycaemic control is important AF management. variability (GV), an emerging marker glycaemic control, associated with unfavourable prognosis, and abnormal GV prevalent ICUs. However, impact on prognosis patients ICU uncertain. This study aimed to evaluate relationship between all-cause mortality after admission at short-, medium-, long-term intervals patients. Methods Data was obtained from Medical Information Mart for Intensive Care IV 3.0 database, admissions (2008–2019) as primary analysis cohort (2020–2022) external validation cohort. Multivariate Cox proportional hazards models, restricted cubic spline analyses were used assess associations outcomes. Subsequently, other clinical features construct machine learning (ML) prediction models 30-day admission. Results included 8989 (age 76.5 [67.7–84.3] years; 57.8% male), while 837 72.9 [65.3–80.2] 67.4% male). revealed that higher quartiles risk (Q3: HR 1.19, 95%CI 1.04–1.37; Q4: 1.33, 1.16–1.52), 90-day 1.25, 1.11–1.40; 1.34, 1.29–1.50), 360-day 1.21, 1.09–1.33; 1.20–1.47) mortality, compared lowest quartile. Moreover, our data suggests needs be contained within 20.0%. Among all ML light gradient boosting had best performance (internal validation: AUC [0.780], G-mean [0.551], F1-score [0.533]; [0.788], [0.578], [0.568]). Conclusion a significant predictor short-term, mid-term, (the potential stratification threshold 20.0%). incorporating demonstrated high efficiency predicting short-term ranked anterior importance. These findings underscore valuable biomarker guiding decisions improving patient outcomes this high-risk population.

Language: Английский

Association and predictive ability between significant perioperative cardiovascular adverse events and stress glucose rise in patients undergoing non-cardiac surgery

Cardiovascular Diabetology, Journal Year: 2024, Volume and Issue: 23(1)

Published: Dec. 18, 2024

The predictive importance of the stress hyperglycemia ratio (SHR), which is composed admission blood glucose (ABG) and glycated hemoglobin (HbA1c), has not been fully established in noncardiac surgery. This study aims to evaluate association capability SHR for major perioperative adverse cardiovascular events (MACEs) surgery patients. Individuals who underwent surgical procedures between 2011 2020, including both diabetic non-diabetic patients, were identified medicine database (INSPIRE 1.1) classified into tertiles based on their SHR. connection risk MACEs was studied using Cox proportional hazards regression analysis, then restricted cubic spline (RCS) employed assess association's form. Additionally, SHR's incremental utility assessed by C-statistic, continuous net reclassification improvement (NRI), integrated discrimination (IDI), thereby quantifying enhancement accuracy brought incorporating existing models. Feature models generated utilizing Boruta algorithm machine learning approaches. A total 5609 patients enrolled. With an upwards shift vertices, rate cardiac death event steadily rose. RCS analysis indicated J-shaped associations. Inflection points occurred at = 0.81 0.97 death. model's fit improved significantly, with a NRI 0.067 (95% CI: 0.025–0.137, P < 0.001) IDI 0.305 0.155–0.430, 0.001). When added as categorical variable (> 0.81), C-statistic increased 0.785 0.756–0.814) ΔC-statistic 0.035 (P 0.009), 0.007 0.000-0.021, 0.016), 0.076 CI -0.024-0.142, 0.092). In algorithm, variables important features green area incorporated development. related following surgery, highlighting its potential useful reliable tool assessing MACEs.

Language: Английский