Establishment and analysis of an early prognosis model of patients with acute kidney injury in intensive care unit
10.3760/cma.j.cn121430-20231106-00944
- VernacularTitle:重症监护病房急性肾损伤患者早期预后模型的建立与分析
- Author:
Yu'an GENG
1
;
Congmei WANG
;
Zhijing XU
;
Lu QI
;
Yangang SHI
;
Shiqiong SU
;
Kai WANG
;
Ruifang LIU
Author Information
1. 河南省直第三人民医院重症医学科，郑州 450006
- Keywords:
Mathematical modeling;
Severe case;
Acute kidney injury
- From:
Chinese Critical Care Medicine
2024;36(2):178-182
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish a predictive model for the progression of acute kidney injury (AKI) to stage 3 AKI (renal failure) in the intensive care unit (ICU), so as to assist physicians to make early and timely decisions on whether to intervene in advance.Methods:A retrospective analysis was conducted. Thirty-eight patients with AKI admitted to the intensive care medicine of the Third People's Hospital of Henan Province from January 2018 to May 2023 were enrolled. Patient data including acute physiology and chronic health evaluation Ⅱ (APACHEⅡ) upon admission, serum creatinine (SCr), blood urea nitrogen (BUN), daily urine output during hospitalization, and the timing of continuous renal replacement therapy (CRRT) intervention were recorded. Based on clinically collected pathological data, standardized creatinine value ratio mean polynomial fitting models were established as the first criterion for judging the progression to stage 3 AKI after data cleansing, screening, and normalization. Additionally, standardized creatinine value ratio index fitting models were established as the second criterion for predicting progression to stage 3 AKI.Results:A total of 38 AKI patients were included, including 25 males and 13 females. The average age was (58.45±12.94) years old. The APACHEⅡ score was 24.13±4.17 at admission. The intervention node was (4.42±0.95) days. Using a dual regression model approach, statistical modeling was performed with a relatively small sample size of statistical data samples, yielding a scatter index non-linear regression model for standardized creatinine value ratio data relative to day " n", with y = 1.246?2 x1.164?9 and an R2 of 0.860?1, indicating reasonable statistical fitting. Additionally, a quadratic non-linear regression model was obtained for the mean standardized creatinine value ratio relative to day " n", with y = -0.260?6 x2+3.010?7 x-1.612 and an R2 of 0.998?9, indicating an excellent statistical fit. For example, using a baseline SCr value of 66 μmol/L for a healthy individual, the dual regression model predicted that the patient would progress to stage 3 AKI within 3-5 days. This prediction was consistent when applied to other early intervention renal injury patients. Conclusion:The established model effectively predicts the time interval of the progression of AKI to stage 3 AKI (renal failure), which assist intensive care physicians to intervene AKI as early as possible to prevent disease progression.
