Development of a model to predict the risk of cerebral infarction in acute vestibular syndrome.

Objectives: This study aimed to develop a model to predict the risk of cerebral infarction in acute vestibular syndrome and assist emergency physicians in quickly identifying patients with cerebral infarction. Materials and methods: We looked at 262 patients who were split into cerebral infarction and peripheral vertigo groups. Stepwise regression and Lasso's approach were used to screen for variables, and Boothstrap's method was used to evaluate the model's discrimination and calibration. The model's performance was compared against TriAGe+, ABCD2, and PCI scores using the area under the receiver operator characteristic curve. Clinical decision-making was aided by the use of clinical impact and decision curves. Results: In the end, nine risk factors were chosen for model 2, and ten risk factors were chosen for model 1. Model 2 was adopted as the final model. The areas under the receiver operator curve value of the model2 were 0.910(P = 0.000), much higher than the areas under the receiver operator curve value of the TriAGe + scores system and that of the PCI scores system. The clinical decision curve shows that when the threshold probability is 0.05, using the nomogram to predict cerebral infarction has more benefits than either the treat-all-patients scheme or the treat-none scheme. The clinical impact curve shows that when the threshold probability is 0.6 the model predicts disease occurrence in general agreement with the occurrence of the real disease. Conclusion: This study model can help emergency room physicians quickly triage and treat patients by accurately identifying cerebral infarction patients.