Analyzing the serum biochemical factors that influence early neurological deterioration in ischemic Stroke patients and developing a nomogram prediction model.

Background: To investigate the risk factors associated with early neurological deterioration (END) in ischemic stroke (IS) patients and develop a predictive nomogram model.

Methods: General clinical data from 220 IS patients treated between December 2022 and November 2023 were collected for observation. The study's inclusion and exclusion criteria select patients aged 18+ with a first-time diagnosis of IS who undergo lab tests within 24 hours of admission while excluding those with multiple organ dysfunction, sensory impairments, coagulation disorders, or other serious medical conditions. Based on the National Institutes of Health Stroke Scale (NIHSS) in the United States, patients were categorized into two groups: END (n=69) and non-END (n=151). Both groups' basic demographics, medical history, and biochemical test results were compared. Influencing factors were identified using the least absolute shrinkage and selection operator (LASSO) method, and these variables were included in a multivariate logistic regression analysis to construct a nomogram for predicting END in IS patients. Model performance was evaluated using internal validation with the Bootstrap method, assessing discrimination, calibration, and clinical validity.

Results: Factors such as history of diabetes, fasting plasma glucose (FBG), triglyceride (TG), homocysteine (Hcy), and C-reactive protein (CRP) were identified as single factors for early functional deterioration in IS patients (P<0.05). A logistic regression model was established with END as the dependent variable and significant single factors (P<0.05) as independent variables. The results indicated that diabetes history (OR=1.398, P=0.301), TG (OR= 6.149, P<0.05), ASPECT score (OR=7.641, P<0.05), FBG (OR=2.172, P<0.05), CRP (OR=1.706, P<0.05), NIHSS score 7 days post-admission (OR=1.336, P<0.05), and Hcy (OR=1.425, P<0.05) were independent risk factors for END in IS patients (P<0.05). ROC analysis showed an ASPECT area under the curve of 0.910 (95% CI:0.864 to 0.944), with 84.06% sensitivity and 86.09% specificity. Hcy had an area under the curve of 0.808 (95% CI:0.750 to 0.858), with 79.71% sensitivity and 70.20% specificity. FBG had an area under the curve of 0.847 (95% CI:0.793 to 0.892), with 69.57% sensitivity and 95.36% specificity. TG had an area under the curve of 0.937 (95% CI: 0.896-0.965), with 91.30% sensitivity and 82.78% specificity. NIHSS had an area under the curve of 0.857 (95% CI: 0.803-0.900), with 89.86% sensitivity and 70.20% specificity. A nomogram model for END risk prediction was constructed based on the logistic regression analysis results, assigning preliminary scores for each of the 9 predictive factors. The total score, ranging from 0-100 points, was used to predict END risk in patients (0-100%). The constructed nomogram model showed that ASPECT was 59.2, Hcy was 84.0, FBG was 61.4, TG7.0 mmol/L was 39.4, and NIHSS was 98.1 with a total score of 345.7 which predicted the risk of END at 68.9%.

Conclusions: ASPECT, Hcy, FBG, TG, and NIHSS are independent factors influencing END after IS. On this basis, a visual predictive nomogram model is constructed to predict the risk of END in patients accurately.