A Multifactor Prediction Model of Recovery of Consciousness in Patients With Unresponsive Wakefulness Syndrome: A Multicenter, Retrospective Study.

AIMS: This multicenter retrospective cohort study aimed to develop 3 predictive models to estimate consciousness status 3 months after admission. These models were designed to serve as complementary prognostic tools for patients diagnosed with unresponsive wakefulness syndrome (UWS) using the Coma Recovery Scale-Revised (CRS-R). METHODS: We retrospectively collected data from 154 patients with UWS across 2 clinical centers, encompassing demographic, clinical, and laboratory biomarkers. Variable selection was performed using adaptive least absolute shrinkage and selection operator (LASSO) and ridge regression. The final predictive models were constructed using binary logistic regression, while additional machine-learning algorithms (Random Forest, SVM, and XGBoost) were applied for comparative evaluations of predictive performance. RESULTS: Of the patients studied, 88 (57%) regained responsiveness within 3 months, while 66 (43%) remained in UWS. Model UWS-base, incorporating clinical factors such as traumatic brain injury (TBI), hypoxic encephalopathy, hydrocephalus, and diffuse injury, demonstrated utility for outpatient preliminary screening. Model UWS-plus achieved superior accuracy (AUC = 0.84) by integrating key biomarkers, including fT3, albumin score, lymphocyte count, and alkaline phosphatase. Model UWS-lite, retaining only lymphocyte count alongside clinical variables, maintained robustness in resource-limited settings (AUC = 0.83). Notably, these biomarkers emerged as factors potentially associated with recovery, generating hypotheses for future interventional studies. CONCLUSION: We propose a biomarker-integrated model that complements the prognosis of UWS. Our findings also encourage a more holistic approach to clinical practice, wherein hydrocephalus management and systemic biomarkers may be considered alongside traditional scales to inform prognosis and guide therapeutic decisions.