Analysis of the Impact of Medical Features and Risk Prediction of Acute Kidney Injury for Critical Patients Using Temporal Electronic Health Record Data With Attention-Based Neural Network.

Zhimeng Chen,Jinting Xu,Xiahong Lin,Yu Liu,Ming Chen,Yuren Zhang,Qiuna Li,Lianwei Wu,Yinqiong Huang,Xuri Sun,Xieli Guo,Yuming Fang

doi:10.3389/fmed.2021.658665

Abstract

Acute kidney injury (AKI) is one of the most severe consequences of kidney injury, and it will also cause or aggravate the complications by the fast decline of kidney excretory function. Accurate AKI prediction, including the AKI case, AKI stage, and AKI onset time interval, can provide adequate support for effective interventions. Besides, discovering how the medical features affect the AKI result may also provide supporting information for disease treatment. An attention-based temporal neural network approach was employed in this study for AKI prediction and for the analysis of the impact of medical features from temporal electronic health record (EHR) data of patients before AKI diagnosis. We used the publicly available dataset provided by the Medical Information Mart for Intensive Care (MIMIC) for model training, validation, and testing, and then the model was applied in clinical practice. The improvement of AKI case prediction is around 5% AUC (area under the receiver operating characteristic curve), and the AUC value of AKI stage prediction on AKI stage 3 is over 82%. We also analyzed the data by two steps: the associations between the medical features and the AKI case (positive or inverse) and the extent of the impact of medical features on AKI prediction result. It shows that features, such as lactate, glucose, creatinine, blood urea nitrogen (BUN), prothrombin time (PT), and partial thromboplastin time (PTT), are positively associated with the AKI case, while there are inverse associations between the AKI case and features such as platelet, hemoglobin, hematocrit, urine, and international normalized ratio (INR). The laboratory test features such as urine, glucose, creatinine, sodium, and blood urea nitrogen and the medication features such as nonsteroidal anti-inflammatory drugs, agents acting on the renin–angiotensin system, and lipid-lowering medication were detected to have higher weights than other features in the proposed model, which may imply that these features have a great impact on the AKI case.

Highlights

Acute kidney injury (AKI) refers to a sudden or sustained decline in renal function, clinically manifested as azotemia, water electrolyte and acid–base balance disorders, and systemic symptoms, accompanied by oliguria or anuria (1)
In each of the three experimental cases, we firstly applied our models together with the baseline models mentioned in Baseline methods to get the performance of AKI case prediction by specificity, sensitivity, and AUC, and the proposed model was implemented to get the performances of AKI stage prediction and AKI onset time interval prediction
We observe that: 1) The performances of the deep learning models were better than the machine learning algorithms on the AKI case prediction, and this result may be explained by the fact that the deep learning model can better obtain the feature dependencies among the electronic health record (EHR) data, which could be beneficial to the AKI case prediction

Summary

Introduction

Acute kidney injury (AKI) refers to a sudden or sustained decline in renal function, clinically manifested as azotemia, water electrolyte and acid–base balance disorders, and systemic symptoms, accompanied by oliguria or anuria (1). Once AKI occurs, the length of hospital stay, medical burden, incidence of chronic kidney disease, and mortality increase significantly (3). Since the factors that lead to AKI are complex, statistical, or machine learning algorithms are used to analyze the important pathogenic factors and build risk assessment models based on various electronic health record (EHR) data, which is currently an important approach for the early detection and prognosis analysis of AKI (4). Rough et al (5) used the long short-term memory (LSTM) model to predict inpatient medication orders from EHRs. Yang et al (6) predicted discharge medications at admission time based on the convolutional neural network (CNN). Darabi et al (8) proposed a time-aware patient representation method from EHR data based on the feedforward neural network (FNN). Choi et al (9) extracted clinical diagnosis codes as base data and used recurrent neural network models for early detection of heart failure onset. Nguyen et al (10) constructed a convolutional net to represent patient features from medical records

Methods

Results

Conclusion