Objective To construct a predictive model of 1-year stroke recurrence for stroke patients in Xi'an. Methods Data of acute stroke patients from Xi'an Stroke Database from January 1 to December 31, 2016 were collected. Clinical data included in-hospital data and the follow-up results at 1 year after discharge. After analyzing the risk factors affecting recurrent stroke, a predictive model of 1-year stroke recurrence was constructed using the Cox proportional hazard regression model. Results A

more »

... l of 2775 patients with acute stroke were enrolled in this study. Recurrent stroke ocurred in 144 patients (accumulated risk 5.34%) within 1 year. A predictive model for the 1-year stroke recurrence as follows: h(t) = h0exp (0.02900×age + 0.83649×prior stroke + 0.26683×peripheral vascular disease history + 0.12887×NIHSS score at admission). The area under the ROC curve was 0.82, the best cut-off point was 0.197, the sensitivity and specificity were 0.69 and 0.78, respectively. The total accuracy of the predictive model is 83%. Conclusions The 1-year stroke recurrence predictive model for stroke patients in Xi'an is targeted, simple, practical and easy to operate. The high-risk stroke patients in Xi'an can be identified by the predictive model and are given individualized treatment timely, which will reduce the recurrence rate of stroke and improve the prognosis of stroke patients.