Abstract: Objective: To construct a risk prediction model for Exposure Keratopathy in neurosurgery ICU patients and to verify the prediction effect of the model. Methods: A total of 102 patients in neurosurgical ICU of a tertiary grade A hospital in Hubei province from January to April 2021 were selected and were divided into the exposed keratitis group (39 cases) and the non-exposed keratitis group (63 cases). Logistic regression was used to establish a risk prediction model for Exposure Keratopathy in neurosurgery ICU patients. The area under Receiver Operating Characteristic (ROC) curve was used to test the prediction effect of the model, and 41 patients were included to verify the model. Results: The incidence of Exposure Keratopathy was 38.24%, 64.86% and 73.91% in neurosurgery ICU patients within a day, after 3 days and 7 days in ICU respectively. The influencing factors were: eyelid closure, blink per minute, other eye symptoms, consciousness state. For Hosmer-Lemeshow test, P=0.379, the area under ROC curve of the subjects was 0.925, the Youden index was 0.773, the optimal critical value was 0.541, the sensitivity was 0.821, and the specificity was 0.952. The correct rate of actual application was 97.56%. Conclusion: The model is effective in predicting the risk of Exposure Keratopathy in ICU patients in neurosurgery, and can provide reference for clinical medical staff to identify high-risk patients and take preventive measures in time.

Key words: Exposure Keratopathy, Intensive Care Unit, neurosurgery department, risk prediction model, nursing