Objective: The mortality rate of critically ill patients in ICUs is relatively high. In order to evaluate patients’ mortality risk, different scoring systems are used to help clinicians assess prognosis in ICUs, such as the Acute Physiology and Chronic Health Evaluation III (APACHE III) and the Logistic Organ Dysfunction Score (LODS). In this research, we aimed to establish and compare multiple machine learning models with physiology subscores of APACHE III—namely, the Acute Physiology Score III (APS III)—and LODS scoring systems in order to obtain better performance for ICU mortality prediction. Methods: A total number of 67,748 patients from the Medical Information Database for Intensive Care (MIMIC-IV) were enrolled, including 7055 deceased patients, and the same number of surviving patients were selected by the random downsampling technique, for a total of 14,110 patients included in the study. The enrolled patients were randomly divided into a training dataset (n = 9877) and a validation dataset (n = 4233). Fivefold cross-validation and grid search procedures were used to find and evaluate the best hyperparameters in different machine learning models. Taking the subscores of LODS and the physiology subscores that are part of the APACHE III scoring systems as input variables, four machine learning methods of XGBoost, logistic regression, support vector machine, and decision tree were used to establish ICU mortality prediction models, with AUCs as metrics. AUCs, specificity, sensitivity, positive predictive value, negative predictive value, and calibration curves were used to find the best model. Results: For the prediction of mortality risk in ICU patients, the AUC of the XGBoost model was 0.918 (95%CI, 0.915–0.922), and the AUCs of logistic regression, SVM, and decision tree were 0.872 (95%CI, 0.867–0.877), 0.872 (95%CI, 0.867–0.877), and 0.852 (95%CI, 0.847–0.857), respectively. The calibration curves of logistic regression and support vector machine performed better than the other two models in the ranges 0–40% and 70%–100%, respectively, while XGBoost performed better in the range of 40–70%. Conclusions: The mortality risk of ICU patients can be better predicted by the characteristics of the Acute Physiology Score III and the Logistic Organ Dysfunction Score with XGBoost in terms of ROC curve, sensitivity, and specificity. The XGBoost model could assist clinicians in judging in-hospital outcome of critically ill patients, especially in patients with a more uncertain survival outcome.

Background: The brain contains the highest level of cholesterol in the body, and the total amount of serum cholesterol in the blood has a huge impact on brain aging and cognitive performance. However, the association of total serum cholesterol with cognitive function remains uncertain. This study determines whether there is an association between the total amount of cholesterol in the blood and cognitive performance in elderly females without a history of stroke. Methods: This population-based cross-sectional study was conducted on elderly (over 60 years old) females and males without a history of stroke from 2011 to 2014 in the US National Health and Nutrition Examination Survey (NHANES). The primary exposure was total blood cholesterol, and the main outcome was cognitive performance; this association was assessed with logistic regression analysis and restricted cubic splines. Results: 1309 female and 1272 male participants were included. In females, higher total cholesterol was significantly associated with higher cognitive scores, particularly in the digit symbol substitution test (OR 0.51, 95% CI (0.36–0.72)) and the animal fluency test (OR 0.64, 95% CI (0.45–0.91)). This association remained significant in models adjusted for age, race, smoking status, education level, and chronic conditions (OR 0.40, 95% CI (0.25–0.63)). This association was not significant in males, however. Conclusions: A higher concentration of total cholesterol measured in later life may be a protective factor for cognitive performance among females over 60 years old without a history of stroke. Further, this association was more pronounced among women with higher levels of education than women with lower or no education.

Objective: Globally, the burden of drug use– and high‐BMI–associated hepatocellular carcinoma is increasing, yet systematic epidemiological assessments and risk factor analyses remain inadequate. This study aims to utilize Global Burden of Disease (GBD) data and genetic epidemiological methods to evaluate trends in the disease burden of drug use– and high‐BMI–associated hepatocellular carcinoma, identify potential risk factors, and predict future burden through machine learning models, thereby providing evidence for clinical prevention and precision interventions. Methods: Drawing from the GBD 2021 database, we extracted data on liver cancer deaths, disability‐adjusted life years (DALYs), years lived with disability (YLDs), and years of life lost (YLLs) from 204 countries and regions between 1990 and 2021, conducting stratified analyses by sex, age, and SDI region. Two‐sample Mendelian randomization (MR) was employed to assess potential causal relationships between 14 lifestyle factors and liver cancer. Furthermore, an XGBoost predictive model was constructed, combined with SHAP analysis to identify key influencing factors, and disease burden projections were made for 2022–2050. Results: In 2021, drug use–related liver cancer caused 65,499 deaths globally, while high‐BMI–related liver cancer caused 46,204 deaths, with annual average growth rates of 2.39% and 3.02%, respectively, between 1990 and 2021. Regions with high sociodemographic index (SDI) bore the greatest absolute burden, though regions with low SDI exhibited higher growth rates. MR analysis suggested a potential genetic association between alcohol consumption and hepatocellular carcinoma. The XGBoost model projected that without interventions, the burden of drug us– and high‐BMI–related hepatocellular carcinoma would continue to rise through 2050. SHAP analysis revealed age, sex, and time as key factors influencing disease burden. Conclusion: The global burden of drug use– and high‐BMI–related hepatocellular carcinoma continues to rise, particularly in middle‐ and high‐SDI regions. Integrating genetic epidemiology with machine learning methods provides data support for identifying high‐risk populations, enabling early intervention and precision prevention. This approach facilitates clinical translation and public health policy formulation.