Prediction and Clinically Important Factors of Acute Kidney Injury Non-recovery

Abstract

ObjectiveThis study aimed to identify phenotypic clinical features associated with acute kidney injury (AKI) to predict non-recovery from AKI at hospital discharge using electronic health record data.MethodsData for hospitalized patients in the AKI Recovery Evaluation Study were derived from a large healthcare delivery system in Taiwan between January 2011 and December 2017. Living patients with AKI non-recovery were used to derive and validate multiple predictive models. In total, 64 candidates variables, such as demographic characteristics, comorbidities, healthcare services utilization, laboratory values, and nephrotoxic medication use, were measured within 1 year before the index admission and during hospitalization for AKI.ResultsAmong the top 20 important features in the predictive model, 8 features had a positive effect on AKI non-recovery prediction: AKI during hospitalization, serum creatinine (SCr) level at admission, receipt of dialysis during hospitalization, baseline comorbidity of cancer, AKI at admission, baseline lymphocyte count, baseline potassium, and low-density lipoprotein cholesterol levels. The predicted AKI non-recovery risk model using the eXtreme Gradient Boosting (XGBoost) algorithm achieved an area under the receiver operating characteristic (AUROC) curve statistic of 0.807, discrimination with a sensitivity of 0.724, and a specificity of 0.738 in the temporal validation cohort.ConclusionThe machine learning model approach can accurately predict AKI non-recovery using routinely collected health data in clinical practice. These results suggest that multifactorial risk factors are involved in AKI non-recovery, requiring patient-centered risk assessments and promotion of post-discharge AKI care to prevent AKI complications.

Kidney Function Trajectory within Six Months after Acute Kidney Injury Inpatient Care and Subsequent Adverse Kidney Outcomes: A Retrospective Cohort Study

“…The findings of this study provide valuable insight into the continuity of post-AKI care and complement previously published short-term (≤90 days) kidney recovery trajectory data. Regarding baseline kidney function and multi-comorbidities populations, such as diabetes, glomerular kidney disease, surgeries, or intensive care [ 4 ], continuous clinical monitoring for at least 6 months post-AKI discharge could be beneficial to develop predictive models with multi-factorial risk factors to identify patients’ degree of risk during post-AKI follow-up [ 21 , 28 , 29 , 30 ]. It is also important for health professionals and patients who experienced AKI to understand the ongoing impacts of AKI on readmission and mortality [ 31 , 32 , 33 , 34 ].…”

“…To assess patient kidney function recovery from AKI, hospitalized patients who had data of at least one SCr value both at admission and ≤ 90 days prior to the index hospitalization, survived to 6 months following the index hospital discharge, and had at least one SCr within 3- and 6 months after the discharge were included in this retrospective cohort of Acute Kidney Injury Recovery Evaluation Study ( Figure 1 ). AKI was defined using the modified 2012 Kidney Disease Improving Global Outcomes criteria [ 6 ] based on an increase in SCr concentration of at least 50% or 0.3 mg/dL (or more within 2 days) above baseline SCr within 2 and up to 90 days before the index hospital admission [ 20 , 21 ]. Patients with AKI at the index hospital admission were classified as CA-AKI, and the highest SCr concentration during the hospitalization was compared with the index SCr at the admission date to define HA-AKI (i.e., peak SCr > 1.5× SCr at admission).…”

Using tumor habitat-derived radiomic analysis during pretreatment 18F-FDG PET for predicting KRAS/NRAS/BRAF mutations in colorectal cancer

“…We used SHAP analysis to interpret the “black-box model” at both the global and the local levels [ 43 ]. The SHAP method enhances the interpretability of a machine learning model, and can estimate the positive and negative contributions of each feature to its prediction [ 44 ]. The SHAP results showed that the ranking of the features according to importance was peritumoral_wavelet-LLH_glrlm_LowGrayLevelRunEmphasis, peritumoral_wavelet-LLH_glcm_Correlation, peritumoral_wavelet-LHH_firstorder_Skewness, high_metabolism_wavelet-LLH_glszm_SizeZoneNonUniformity, and high_metabolism_wavelet-HLH_firstorder_Skewness.…”

A machine learning-based approach for predicting renal function recovery in general ward patients with acute kidney injury

“…Consequently, this research area experiences a significant dearth of research, necessitating the need for new investigations. Prior efforts to predict renal function recovery have been hindered by limitations, such as small sample sizes [8][9][10], exclusive focus on ICU patients, and the absence of an all-encompassing definition for renal function recovery, which impedes the application of these research outcomes in clinical settings [11][12][13][14][15]. Therefore, this study aimed to fill this gap by developing a machine learning-based approach that includes validation in external settings, aimed to predict renal function recovery in patients with AKI, with a particular focus on patients in general wards.…”