Artificial Intelligence for Risk Prediction of Rehospitalization with Acute Kidney Injury in Sepsis Survivors

Abstract

Sepsis survivors have a higher risk of long-term complications. Acute kidney injury (AKI) may still be common among sepsis survivors after discharge from sepsis. Therefore, our study utilized an artificial-intelligence-based machine learning approach to predict future risks of rehospitalization with AKI between 1 January 2008 and 31 December 2018. We included a total of 23,761 patients aged ≥ 20 years who were admitted due to sepsis and survived to discharge. We adopted a machine learning method by using models based on logistic regression, random forest, extra tree classifier, gradient boosting decision tree (GBDT), extreme gradient boosting, and light gradient boosting machine (LGBM). The LGBM model exhibited the highest area under the receiver operating characteristic curves (AUCs) of 0.816 to predict rehospitalization with AKI in sepsis survivors and followed by the GBDT model with AUCs of 0.813. The top five most important features in the LGBM model were C-reactive protein, white blood cell counts, use of inotropes, blood urea nitrogen and use of diuretics. We established machine learning models for the prediction of the risk of rehospitalization with AKI in sepsis survivors, and the machine learning model may set the stage for the broader use of clinical features in healthcare.

Liver Injury in Favipiravir-Treated COVID-19 Patients: Retrospective Single-Center Cohort Study

“…The liver injury of FVP was assessed based on liver function parameters: alanine aminotransaminase, total bilirubin, direct bilirubin, alkaline phosphatase, and aspartate aminotransaminase, by measuring their blood levels. Before beginning treatment with any medications and after beginning therapy with FVP or COVID-19 protocol medications, the liver function parameters were collected [ 22 ].…”

Mineral Prospectivity Mapping of Tungsten Polymetallic Deposits Using Machine Learning Algorithms and Comparison of Their Performance in the Gannan Region, China

“…The LGBM optimizes the optimal split point search strategy, leaf growth strategy, gradient sampling method, feature attribute binding method, support class feature computation, and parallel learning of the decision tree (Vaulet et al., 2022; Zhang et al., 2021a). The LGBM algorithm not only guarantees prediction accuracy, but also reduces the training time and memory consumption (Ou et al., 2022; Sun, Liu, & Sima, 2020). This algorithm is iterative and after several iterations, each weak classifier is weighted to become a stronger classifier (Figure 12).…”

“…The CNN is an efficient deep learning method, which can process complex data through its high capacity for feature learning and convolutional filtering (LeCun et al., 1998; Li et al., 2021). The LGBM is a new machine learning method that has a fast running speed, low memory consumption, and is widely used in biology, medicine, chemistry, social science (Ou et al., 2022; Sun, Liu, & Sima, 2020; Vaulet et al., 2022; Zafari et al., 2020), and lithofacies classification (Zhang et al., 2021a). However, it is rarely used to integrate multisource geospatial data for MPM.…”

Role of C-Reactive Protein in Kidney Diseases

“…Indeed, the CRP/albumin ratio is an independent risk factor for postoperative AKI occurred in elderly cystectomy patients and CRP/albumin ratio ≥0.1 has been shown to be associated with the increased incidence of AKI [ 15 ]. Emerging data have also suggested that CRP is an important predictive indicator of sepsis-induced AKI [ 16 ]. Moreover, the serum level of CRP is found to be associated with the mortality in older AKI patients and oncology patients with AKI [ 17 , 18 ].…”