Development and Validation of a Machine Learning Model Integrating Stress Hyperglycemia for Predicting Early Neurological Deterioration in Acute Mild-to-Moderate Ischemic Stroke
Objective: To investigate the predictive value of stress hyperglycemia and various machine learning models for early neurological deterioration (END) in patients with acute mild-to-moderate ischemic stroke, and to provide a quantitative reference for early clinical identification of high-risk patients and formulation of individualized intervention strategies. Methods: In this retrospective study, a total of 1,169 patients with acute mild-to-moderate ischemic stroke who met the inclusion criteria were enrolled from Guangdong Second Provincial General Hospital between January 2017 and June 2023. According to the sequential exclusion, 933 eligible subjects were finally included. After applying the predefined exclusion criteria, 933 eligible subjects were finally included. Results: Among the 933 enrolled patients, the END incidence was 18.97%. Baseline data showed that the END group had higher body mass index (BMI), admission blood pressure, baseline National Institutes of Health Stroke Scale (NIHSS) score, admission blood glucose, glycated hemoglobin, and stress hyperglycemia ratio (SHR), while Alberta Stroke Program Early CT Score (ASPECTS) and Activities of Daily Living (ADL) scores were lower. Stress hyperglycemia was also more frequent in the END group. A history of hypertension was significantly more common in the END group, whereas the difference in diabetes mellitus history was not statistically significant (P = 0.064). Four machine learning prediction models were constructed based on the dataset divided at a 7:3 ratio. The XGBoost model demonstrated a wider applicable clinical threshold range and superior clinical decision-making value for screening high-risk populations. Conclusion: Four machine learning models integrating stress hyperglycemia can effectively predict the risk of early neurological deterioration in patients with acute mild-to-moderate ischemic stroke. Among them, the XGBoost model demonstrates superior clinical decision-making value and may serve as a reference for stratified clinical intervention.
[1] Lin L, Jiang LJ, Wang XG, et al., 2023, Effect of Leukoaraiosis on Intravenous Thrombolysis Efficacy in Acute Mild-to-moderate Cerebral Infarction. Cardiovascular and Cerebrovascular Disease Prevention and Treatment, 23(7): 32–36.
[2] Yang W, Kim KY, Lee JK, et al., 2026, Abstract DP329: Impact of White Matter Hyperintensity Burden on Response to Induced Hypertension Therapy in Small Vessel Occlusion Stroke with Early Neurological Deterioration. Stroke, 2026(57): Suppl_1. https://doi.org/10.1161/str.57.suppl_1.DP329
[3] Khasiyev F, Pandit M, Jumani S, et al., 2026, Abstract WP118: Exploring Intracranial Arterial Variants and Early Neurological Deterioration in Lacunar Stroke. Stroke, 2026(57): Suppl_1. https://doi.org/10.1161/str.57.suppl_1.WP118.
[4] Wang N, Lin H, Deng B, et al., 2026, Comparative the Predictive Value of Neutrophil-to-HDL Ratio and Monocyte-to-HDL Ratio for Early Neurological Deterioration in Acute Ischemic Stroke Patients Following Intravenous Thrombolysis. Neurological Sciences, 47(5): 431. https://doi.org/10.1007/s10072-025-08782-8
[5] Zhong X, Yang J, Xu R, et al., 2026, Anticoagulant or Antithrombotic Therapy after Early Neurological Deterioration in Patients with Branch Atheromatous Disease Receiving Dual Antiplatelet Therapy. International Journal of Clinical Practice, 2026(8636214): 7. https://doi.org/10.1155/ijcp/8636214
[6] Zhong X, Chen Z, Cao T, et al., 2026, Development and Validation of a Nomogram to Predict Early Neurological Deterioration in Patients with Acute Ischemic Stroke and Type 2 Diabetes Mellitus: The Pivotal Role of Glycemic Variability and Thrombo-Inflammation. Neurological Sciences, 47(6): 544. https://doi.org/10.1007/s10072-026-09160-8
[7] Font MN, García-Sánchez SM, 2026, Catalan Stroke Code and Reperfusion Consortium. Impact of Arterial Occlusion Location on Early Neurological Deterioration in Patients with Minor Stroke Treated with Intravenous Thrombolysis. The MINORCAT-END study. Neurological Sciences, 47(1): 12. https://doi.org/10.1007/s10072-025-08721-7
[8] Xu S, Jiang Z, Zhou Y, et al., 2026, The effect of Indobufen and Aspirin on Platelet Function in Acute Ischemic Stroke Patients with Early Neurological Deterioration. European Journal of Pharmacology, 2026(1011): 178439. https://doi.org/10.1016/j.ejphar.2025.178439
[9] Neurology Branch and Cerebrovascular Diseases Group of the Chinese Medical Association, 2026, Chinese Guidelines for the Diagnosis and Treatment of Acute Ischemic Stroke 2023: Stroke Recognition, Management, and Transfer. Practical Journal of Cardiac Cerebral Pneumal and Vascular Disease, 34(4): 24.
[10] Writing Group of the Clinical Guidelines for the Prevention and Treatment of Type 2 Diabetes Mellitus in the Elderly in China, Geriatric Endocrinology and Metabolism Branch of the Chinese Geriatrics Society, Geriatric Endocrinology and Metabolism Branch of the Chinese Association of Geriatric Research, et al., 2026, Clinical Guidelines for the Prevention and Treatment of Type 2 Diabetes Mellitus in the Elderly in China (2026 edition) Released in Beijing. Diabetes Friend, 2026(5): 7.
[11] Huang S, Zou DJ, 2025, Treatment of Type 2 Diabetes Mellitus: Should Metformin Step Down to Second-line? Reflections on the Glucose-lowering Pathway of the Chinese Guidelines for Diabetes Prevention and Treatment (2024 edition). Chinese Journal of Diabetes Mellitus, 17(10): 1261–1266.
[12] Zhou XY, 2025, Dietary Habits Survey of Patients with Type 2 Diabetes Mellitus Complicated with Cardiovascular Disease: Based on the Chinese Guidelines for Prevention and Treatment of Type 2 Diabetes (2020 edition), thesis, Xi’an Medical University. https://doi.org/10.27909/d.cnki.gxaxy.2025.000105
[13] Khasiyev F, Pandit M, Jumani S, et al., 2026, Abstract DP320: Elevated Neutrophil-to-lymphocyte Ratio Does not Predict Early Neurological Deterioration in Patients with Acute Lacunar Infarct. Stroke, 2026(57): Suppl_1. https://doi.org/10.1161/str.57.suppl_1.DP320
[14] Stutz A, Borei K, Pan A, et al., 2026, Abstract DP382: Increased Blood Pressure Variability over the first 24 Hours of Admission is Independently Associated with Neurological Deterioration among Patients with Acute Ischemic Stroke. Stroke, 2026(57): Suppl_1. https://doi.org/10.1161/str.57.suppl_1.DP382
