Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-0993
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-09-06T06:04:39.882658Z
status: published
lens: g_space
sector: health
headline: Prognostic Value of Admission D-dimer Levels and Total Bleeding Volume in Aneurysmal Subarachnoid Hemorrhage: A Retrospective Cohort Study with Machine Learning-Based Modeling
dek: Background Plasma D-dimer levels are independently associated with poor prognosis following aneurysmal subarachnoid hemorrhage (aSAH). However, the underlying mechanisms contributing to early D-dimer elevation remain unclear. This study aimed to evaluate the association between admission D-dimer levels and total bleeding volume (TBV) and to further explore their combined predictive power for functional outcomes using interpretable machine learning approaches. Methods We analyzed data from 473 patients with aSAH…
gain_title: Interpretable machine learning combining admission D-dimer and total bleeding volume improved prediction of 12-month functional outcome after aneurysmal subarachnoid hemorrhage, with XGBoost achieving AUC 0.904 and combined D-dimer+TBV+Hunt-Hess outperforming single markers, to inform early risk stratification.
problem_title: (none)
trace_subject: (none)
gain_reading: Interpretable machine learning combining admission D-dimer and total bleeding volume improved prediction of 12-month functional outcome after aneurysmal subarachnoid hemorrhage, with XGBoost achieving AUC 0.904 and combined D-dimer+TBV+Hunt-Hess outperforming single markers, to inform early risk stratification.
gain_evidence: XGBoost achieved the highest discriminative performance (AUC = 0.904) | Combined D-dimer + TBV + Hunt-Hess yielded the highest area under the curve (AUC; 0.867; 95% CI 0.764-0.856), outperforming D-dimer (0.735), TBV (0.783), and Hunt-Hess (0.833) | Integrating machine learning and SHAP interpretability enhances our understanding of these relationships and may inform early risk stratification in clinical practice.
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers analyzed 473 patients with aneurysmal subarachnoid hemorrhage from the retrospective PROSAH-MPC cohort to test whether admission D-dimer levels and total bleeding volume together predict long-term function. They stratified patients by D-dimer quartiles, ran multivariable logistic regression for 12-month modified Rankin Scale outcomes, selected features with Boruta, and built seven machine learning models interpreted with SHAP.

The work matters because early risk stratification after aSAH remains difficult, and the models showed that combining a blood biomarker with imaging volume improves discrimination over single markers alone, with XGBoost reaching AUC 0.904. Uncertainty remains about generalizability beyond this retrospective cohort and about the biological mechanism linking early D-dimer elevation to bleeding volume and outcome.
limitation: Retrospective single-cohort design with 473 patients limits generalizability and causal inference about D-dimer mechanisms, and performance was evaluated within the same PROSAH-MPC cohort without external validation reported.
tag: Evidence-backed gain
key_points: Retrospective analysis of 473 patients with aSAH from the PROSAH-MPC cohort stratified by D-dimer quartiles. | Unfavorable 12-month mRS outcomes occurred in 125 patients (26.4%) and elevated D-dimer was associated with unfavorable outcomes with adjusted OR 1.08. | Boruta feature selection identified D-dimer, TBV, age, Hunt-Hess grade, and modified Fisher score as top predictors, with significant TBV and D-dimer interaction P = 0.032. | Seven machine learning models were developed and SHAP analysis confirmed D-dimer and TBV as major contributors to prediction.
rundown: The study analyzed 473 aSAH patients from the retrospective PROSAH-MPC cohort, including clinical, radiological, and laboratory parameters, stratified by D-dimer quartiles and evaluated with multivariable logistic regression for 12-month mRS.

Feature selection via Boruta identified five top predictors, and seven machine learning models were trained; XGBoost had the highest AUC at 0.904, while the combined D-dimer + TBV + Hunt-Hess model reached AUC 0.867, with SHAP used to interpret feature contributions and a significant TBV-D-dimer interaction noted.
sources:
- peer_reviewed | Neurocritical Care | https://doi.org/10.1007/s12028-026-02558-4 | 2026-09-04
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

26a928fcdd475c5b88bebb53dc1bb4f1fd21383a29293fc2ca36ef27fe4bc6a5

previous

0000000000000000000000000000000000000000000000000000000000000000