Machine learning method to derive serum chlorine in acute kidney injury with traumatic brain injury

Background and purpose: Acute kidney injury (（AKI)） is a serious complication of traumatic brain injury (TBI), thatwhich affects the survival rate of patients with TBI patients. Hyperchloremia is an important factor leading to TBI-associated AKI and affects the prognosis of TBI patients with TBI; however,. but the The highest blood chlorine value is usually uselected to define it. However, the effectimpact of changes in the trajectory of early blood chlorine levels on the prognosis of patients with TBI-associated AKI remains unclear. Methods: We used the Critical Medicine Database (MIME-IV) and eICU to include TBI patients with TBI and AKI admitted to the ICU. We used the potential class model to determine the AKI class inof athe blood chlorine trial of patients with TBI. The primary endpoint was 28 day all-cause mortality, and the secondary endpoints were the development of acute kidney disease (CAD) and 7 day all-cause mortality. Multivariate regression was used performed to assess the effect of classification based on blood chlorine levels, and external validation was performed using the eICU database. Results: In 640 patients with TBI-associated AKI, we identified five categories based on blood chlorine trials with different characteristics. Compared with patients with slightly elevated blood chlorine levels late in life, those with rapidly elevated blood chlorine levels early in life had the highest risk of all-cause mortality at 28 days (HR: 2.47 (1.16, 5.65), P = 0.019), a higher risk of death at 7 days (HR: 3.18 (1.21-8.36), P = 0.019), and the highest adjusted risk of developing acute kidney disease (HR: 5.40 (1.44-20.22), P = 0.012). These associations were similar in external validation. Conclusion: The stratification of early blood chlorine trajectories can be used as a prognostic indicator for TBI-associated AKI patients.