Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1650
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-17T01:10:15.770971Z
status: published
lens: g_space
sector: health
headline: An Explainable Machine Learning Model Integrating Clinical Parameters for Identifying Left Atrial Appendage Thrombus in Non-valvular Atrial Fibrillation
dek: Objective This study aimed to develop a predictive model for left atrial appendage thrombus (LAAT) in patients with non-valvular atrial fibrillation by integrating clinical and echocardiographic features using machine learning algorithms. The SHapley Additive exPlanations (SHAP) framework was applied to interpret the model and quantify the contribution of each predictor, thereby enhancing the accuracy and clinical interpretability of LAAT risk prediction. Methods In this retrospective study, 325 patients with no…
gain_title: A machine learning model with SHAP explanations uses readily available clinical and echocardiographic variables to predict thrombus in the left atrial appendage, which may support individualized anticoagulation decisions.
problem_title: (none)
trace_subject: (none)
gain_reading: A machine learning model with SHAP explanations uses readily available clinical and echocardiographic variables to predict thrombus in the left atrial appendage, which may support individualized anticoagulation decisions.
gain_evidence: providing a practical tool to assist in individualized anticoagulation management and clinical decision-making
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built a prediction model for left atrial appendage thrombus in people with non-valvular atrial fibrillation. They combined clinical and echocardiographic features and compared six machine learning algorithms, then applied SHAP to show how much each predictor contributed.

The work matters because it aims to give clinicians an interpretable aid for judging clot risk and tailoring anticoagulation. The model was developed retrospectively at one institution, so how it performs in other hospitals and patient groups remains uncertain.
limitation: The model was developed in a retrospective study of 325 patients at a single institution, so its generalizability and prospective clinical performance are not established in the supplied text.
tag: Evidence-backed gain
key_points: The study included 325 patients with non-valvular atrial fibrillation who underwent transesophageal echocardiography between September 2022 and October 2024. | Six algorithms were trained and evaluated: logistic regression, support vector classifier, random forest, GradientBoosting, Extreme Gradient Boosting and Light Gradient Boosting Machine. | Multivariable logistic regression linked lower LAA blood flow velocity to thrombus (odds ratio, 0.805). | SHAP analysis visualized each feature's effect on model predictions to improve interpretability.
rundown: The retrospective cohort comprised 325 patients with non-valvular atrial fibrillation who had transesophageal echocardiography at a single institution. Feature selection used LASSO regression with fivefold cross-validation performed only within the training cohort, followed by multivariable logistic regression to quantify independent associations.

Lower LAA blood flow velocity was identified as an independent factor linked to thrombus, with an odds ratio of 0.805. The supplied text is truncated after this point, so other predictors and the performance metrics are not available.

The authors conclude the model integrates readily available variables and could support individualized anticoagulation management. SHAP was used to interpret predictions and quantify each predictor's contribution, aiming to make the model clinically understandable.
sources:
- peer_reviewed | Ultrasound in Medicine & Biology | https://doi.org/10.1016/j.ultrasmedbio.2026.09.012 | 2026-10-09
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

82b635599ec482a632082cc2ff04eaa2333ceef8800b1ea26f7fa6ee153e1bdd

previous

0000000000000000000000000000000000000000000000000000000000000000