Acute liver failure (ALF) is a serious disease that progresses from acute liver injury (ALI) and that often leads to multiorgan failure and ultimately death. Currently, effective treatment strategies for ALF, aside from transplantation, remain elusive, partly because ALI is highly heterogeneous. Furthermore, clinicians lack a quantitative indicator that they can use to predict which patients hospitalized with ALI will progress to ALF and the need for liver transplantation. In our study, we retrospectively analyzed data from 319 patients admitted to the hospital with ALI. By applying a machine-learning approach and by using the SHapley Additive exPlanations (SHAP) algorithm to analyze time-course blood test data, we identified prothrombin time activity percentage (PT%) as a biomarker reflecting individual ALI status. Unlike previous studies predicting the need for liver transplantation in patients with ALF, our study focused on PT% dynamics. Use of this variable allowed us to stratify the patients with highly heterogeneous ALI into six groups with distinct clinical courses and prognoses, i.e. self-limited, intensive care–responsive, or intensive care–refractory patterns. Notably, these groups were well predicted by clinical data collected at the time of admission. Additionally, utilizing mathematical modeling and machine learning, we assessed the predictability of individual PT% dynamics during the early phase of ALI. Our findings may allow for optimizing medical resource allocation and early introduction of tailored individualized treatment, which may result in improving ALF prognosis.
Background/Objectives: To develop and validate a simple prediction model for identifying individuals with low fat-free mass index (FFMI) using routinely available clinical variables. Methods: A total of 8736 individuals were included (mean age = 54.2 years). Among men, 1674 (40.1%) had a low FFMI (<18 kg/m2), while among women, 2160 (47.3%) had a low FFMI (<15 kg/m2). Candidate predictors were preprocessed to reduce redundancy and multicollinearity. Repeated LASSO logistic regression and XGBoost-based SHAP analysis were used to identify robust predictors, and sex-specific logistic regression models were developed using the final selected variables. Model performance was evaluated using 500 repeated 8:2 train-test splits, calibration analysis, risk stratification, and decision curve analysis (DCA). Results: BMI and estimated GFR (eGFR) were consistently selected as key predictors and were used to construct the final sex-specific models. The BMI-eGFR model showed excellent discrimination. Calibration plots demonstrated good agreement between predicted probabilities and observed event rates. Mean held-out sensitivities and specificities were 0.861/0.875 in men and 0.869/0.805 in women, respectively. DCA showed that the model provided a higher net benefit than treat-all or treat-none strategies over most of the evaluated threshold range of 0.05–0.50. Conclusions: Our model for predicting low FFMI may be helpful when direct muscle mass measurement is not readily available.