Hui-Min Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hui-Min Wang has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
Jian-Ting Zhao
Department of Neurology, Xinxiang Central Hospital, Xinxiang, China
1 shared publicationLu-Lu Chang
Department of Neurology, Xinxiang Central Hospital, Xinxiang, China
1 shared publicationMiao-Miao Li
Department of Neurology, The Fourth Clinical College of Henan Medical University, Xinxiang, China
1 shared publicationShuang-Mo Zhao
Department of Neurology, The Fourth Clinical College of Henan Medical University, Xinxiang, China
1 shared publication- Wen-Song Yang
Department of Neurology, The First Affiliated Hospital of Chongqing Medical University, Chongqing, China
1 shared publication Xiao-Yu Jia
Department of Neurology, The Fourth Clinical College of Henan Medical University, Xinxiang, China
1 shared publication
Latest Publications
- Open Access
Inflammatory Biomarkers and Interpretable Machine Learning Model for Stroke-Associated Pneumonia Risk Stratification in Patients Undergoing Bridging Therapy for Acute Ischemic Stroke
Stroke-associated pneumonia (SAP) is a major complication affecting the prognosis of acute ischemic stroke (AIS) patients who undergo intravenous thrombolysis followed by bridging mechanical thrombectomy (MT). Existing predictive tools are mostly subjective, and the value of combined biomarkers and interpretable machine learning (ML) models in this specific population remains unclear. This study aimed to construct an interpretable ML model to predict SAP risk in AIS patients undergoing bridging therapy, and to identify key predictive factors using Shapley Additive Explanations (SHAP). A single-center retrospective observational study was conducted on 135 AIS patients who received intravenous thrombolysis followed by bridging MT at Xinxiang Central Hospital from January 2019 to December 2023. Clinical data, laboratory indicators (including neutrophil-to-lymphocyte ratio [NLR], platelet-to-lymphocyte ratio [PLR], systemic immune-inflammation index [SII], and systemic inflammatory response index [SIRI]) were collected. LASSO regression was used for feature selection, and ten ML models were constructed to predict SAP. The optimal model was selected based on area under the receiver operating characteristic curve (AUC-ROC) and decision curve analysis (DCA), and SHAP analysis was applied to interpret the model. Among 135 included patients, 70 (51.9%) developed SAP. LASSO regression selected 11 key variables associated with SAP. The CatBoost model showed the best performance, with an AUC of 0.952 in the training set and 0.932 in the test set. SHAP analysis revealed that the 7-day National Institutes of Health Stroke Scale (NIHSS_7d), 24-h SIRI (SIRI_24h), and 24-h white blood cell count (WBC_24h) were the top three factors contributing to SAP prediction. Dynamic detection showed that 24-h and 48-h NLR, 24-h SII, and 24-h and 48-h SIRI were significantly higher in the SAP group than in the non-SAP group (all P < 0.05), while no significant difference in PLR was observed between the two groups. Inflammatory biomarkers (24-h NLR, 24-h SII, 24-h SIRI, 48-h NLR, 48-h SIRI) are closely associated with SAP in AIS patients undergoing bridging therapy. The interpretable CatBoost model constructed with LASSO-selected variables exhibits high predictive value for SAP, which can help clinicians identify high-risk patients early and guide personalized treatment.
Open AccessInflammatory Biomarkers and Interpretable Machine Learning Model for Stroke-Associated Pneumonia Risk Stratification in Patients Undergoing Bridging Therapy for Acute Ischemic Stroke
Stroke-associated pneumonia (SAP) is a major complication affecting the prognosis of acute ischemic stroke (AIS) patients who undergo intravenous thrombolysis followed by bridging mechanical thrombectomy (MT). Existing predictive tools are mostly subjective, and the value of combined biomarkers and interpretable machine learning (ML) models in this specific population remains unclear. This study aimed to construct an interpretable ML model to predict SAP risk in AIS patients undergoing bridging therapy, and to identify key predictive factors using Shapley Additive Explanations (SHAP). A single-center retrospective observational study was conducted on 135 AIS patients who received intravenous thrombolysis followed by bridging MT at Xinxiang Central Hospital from January 2019 to December 2023. Clinical data, laboratory indicators (including neutrophil-to-lymphocyte ratio [NLR], platelet-to-lymphocyte ratio [PLR], systemic immune-inflammation index [SII], and systemic inflammatory response index [SIRI]) were collected. LASSO regression was used for feature selection, and ten ML models were constructed to predict SAP. The optimal model was selected based on area under the receiver operating characteristic curve (AUC-ROC) and decision curve analysis (DCA), and SHAP analysis was applied to interpret the model. Among 135 included patients, 70 (51.9%) developed SAP. LASSO regression selected 11 key variables associated with SAP. The CatBoost model showed the best performance, with an AUC of 0.952 in the training set and 0.932 in the test set. SHAP analysis revealed that the 7-day National Institutes of Health Stroke Scale (NIHSS_7d), 24-h SIRI (SIRI_24h), and 24-h white blood cell count (WBC_24h) were the top three factors contributing to SAP prediction. Dynamic detection showed that 24-h and 48-h NLR, 24-h SII, and 24-h and 48-h SIRI were significantly higher in the SAP group than in the non-SAP group (all P < 0.05), while no significant difference in PLR was observed between the two groups. Inflammatory biomarkers (24-h NLR, 24-h SII, 24-h SIRI, 48-h NLR, 48-h SIRI) are closely associated with SAP in AIS patients undergoing bridging therapy. The interpretable CatBoost model constructed with LASSO-selected variables exhibits high predictive value for SAP, which can help clinicians identify high-risk patients early and guide personalized treatment.