Early prediction of 3-month functional outcome after acute ischemic stroke: an explainable model based on routine clinical data

 

Predicting failure to recover is the sign of PURE INCOMPETENCE!

You are supposed to deliver 100% recovery protocols! If you don't know that; get the hell out of stroke!  I'd fire anyone doing this type of crapola research!

Early prediction of 3-month functional outcome after acute ischemic stroke: an explainable model based on routine clinical data

• 

  Zichen Rao ^1†

• 

  Chunyi Xu ^2†

• 

  Yanmei Yu ³

• Y

  Yong Liu ³

• C

  Chunyan Zhu ¹

• 

  Yiming Zhang ^{1‡ *}

• 1. Department of Endocrinology, Quzhou People’s Hospital, The Quzhou Affiliated Hospital, Wenzhou Medical University, Quzhou, Zhejiang, China

• 2. Department of Endocrinology and Metabolism, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China

Abstract

Background: 

Accurate early prediction of 3-month functional outcome after acute ischemic stroke is essential for prognostic communication, early management planning, and resource allocation. However, clinically applicable and interpretable prediction tools based on routinely collected data remain limited.

Methods: 

In this single-center retrospective prediction-model study, we analyzed 770 adults with acute ischemic stroke admitted within 24 h of symptom onset. Baseline clinical and early laboratory variables were collected. After an 80/20 train–test split, key predictors were selected using cross-validated least absolute shrinkage and selection operator (LASSO) in the training set. Several supervised learning models were developed and compared. Model discrimination, calibration, and clinical utility were evaluated. The final gradient boosting machine (GBM) model was interpreted using TreeSHAP, and an online Shiny-based calculator was developed.

Results: 

LASSO identified 12 predictors. Test-set discrimination was similar across models, with area under the receiver operating characteristic curve (AUC) values ranging from 0.854 to 0.867. The final GBM achieved AUCs of 0.910 in the training set and 0.872 in the test set, with corresponding areas under the precision–recall curve of 0.814 and 0.741. Calibration was satisfactory in both sets. SHAP analyses identified NIHSS, age, and C-reactive protein as dominant contributors, showing predominantly non-linear effects with limited interaction strength.

Conclusion: 

An explainable model based on routinely collected clinical data showed good discrimination and calibration in internal validation for predicting 3-month functional outcome after AIS. Independent external validation is required before the model can be considered for clinical application.