Abstract: BACKGROUND: Previous studies have predominantly focused on long-term cognitive outcomes, and an instrument that can be applied immediately upon admission to quantify, based on serum biochemical markers, the risk of cognitive impairment at discharge in elderly patients with acute cerebral infarction is still lacking.
OBJECTIVE: To construct and validate a prediction model based on serum biochemical indicators at admission to evaluate the risk of cognitive impairment at discharge in elderly patients with acute cerebral infarction.
METHODS: A total of 280 elderly patients with acute cerebral infarction treated at the Department of Neurology, Yichun People’s Hospital from July 2023 to June 2024 were selected and divided into a cognitive impairment group (n=157) and a non-cognitive impairment group (n=123) based on the presence or absence of cognitive impairment at discharge. Another 120 patients treated from July 2024 to January 2025 were selected as the validation set in a 7:3 training-to-validation ratio. Clinical and serum biochemical indicators were collected. Independent risk factors were identified using univariate analysis, least absolute shrinkage and selection operator regression, and multivariate logistic regression. Spearman correlation analysis was used to assess the relationship between serum biochemical indicators at admission and clinical indicators. A restricted cubic spline model was employed to analyze the nonlinear relationship between serum biochemical indicators and cognitive impairment occurrence. Logistic models incorporating independent risk factors were constructed, and nomograms were developed. Model performance was comprehensively evaluated using the area under the area under the receiver operating characteristic curve (AUC), confusion matrix metrics (accuracy, precision, recall, and F1 score), calibration curves, and decision curves, with validation performed in the validation set.
RESULTS AND CONCLUSION: Among 400 patients, the incidence of cognitive impairment was 55.00%. (1) Eleven independent risk factors for cognitive impairment were identified: age (odds ratio [OR]=1.095, P < 0.001), National Institutes of Health Stroke Scale score (OR=1.121, P=0.013), cerebral infarction location (OR=1.785, P=0.006), cerebral infarction type (OR=1.587, P=0.017), hypertension (OR=2.200, P=0.014), diabetes (OR=2.249, P=0.011), leukoaraiosis (OR=2.031, P=0.022), homocysteine (OR=1.088, P=0.020), C-reactive protein (OR=1.142, P=0.025), high-density lipoprotein (OR=0.208, P=0.021), and 25-hydroxyvitamin D (OR=0.973, P=0.013). (2) The results of the Spearman correlation analysis showed that in elderly patients with acute cerebral infarction, the level of homocysteine was positively correlated with the location of cerebral infarction (r=0.127, P < 0.05); the level of C-reactive protein was positively correlated with hypertension and the type of cerebral infarction (r=0.259, P < 0.001; r=0.178, P < 0.001); the level of high-density lipoprotein was negatively correlated with age (r=-0.131, P < 0.05); and the level of 25-hydroxyvitamin D was negatively correlated with diabetes (r=-0.145, P < 0.05). (3) After adjusting for other independent risk factors, elevated homocysteine and C-reactive protein levels increased the risk of cognitive impairment, whereas increased levels of high-density lipoprotein and 25-hydroxyvitamin D decreased the risk of cognitive impairment. (4) Critical thresholds for homocysteine, C-reactive protein, high-density lipoprotein, and 25-hydroxyvitamin D were 21.44 μmol/L, 11.73 mg/L, 0.56 mmol/L, and 48.22 nmol/L, respectively. (5) The AUC values for the biochemical indicator model were 0.852 (0.809-0.895) in the training set and 0.836 (0.790-0.882) in the validation set, while those for the total prediction model were 0.918 (0.886-0.950) and 0.895 (0.857-0.933), respectively. Both models demonstrated good predictive performance, calibration, and clinical utility. These findings indicate that the nomogram model based on serum biochemical markers at hospital admission exhibits robust predictive power for the risk of cognitive impairment following ischemic stroke in elderly patients with acute cerebral infarction. It serves as a visual aid for the early clinical identification of patients at high risk for cognitive impairment.

Key words: serum biochemical indicators, cerebral infarction, cognitive impairment, risk factors, nomogram, prediction model