Causal risk factor discovery for severe acute kidney injury using electronic health records

Weiqi Chen,Zilin Tang,Yong Hu,Jianqin He,Xing Song,Lijuan Wu,Kang Liu,Lemuel R Waitman,Xiangzhou Zhang,Mei Liu

doi:10.1186/s12911-018-0597-7

Abstract

BackgroundAcute kidney injury (AKI), characterized by abrupt deterioration of renal function, is a common clinical event among hospitalized patients and it is associated with high morbidity and mortality. AKI is defined in three stages with stage-3 being the most severe phase which is irreversible. It is important to effectively discover the true risk factors in order to identify high-risk AKI patients and allow better targeting of tailored interventions. However, Stage-3 AKI patients are very rare (only 0.2% of AKI patients) with a large scale of features available in EHR (1917 potential risk features), yielding a scenario unfeasible for any correlation-based feature selection or modeling method. This study aims to discover the key factors and improve the detection of Stage-3 AKI.MethodsA causal discovery method (McDSL) is adopted for causal discovery to infer true causal relationship between information buried in EHR (such as medication, diagnosis, laboratory tests, comorbidities and etc.) and Stage-3 AKI risk. The research approach comprised two major phases: data collection, and causal discovery. The first phase is propose to collect the data from HER (includes 358 encounters and 891 risk factors). Finally, McDSL is employed to discover the causal risk factors of Stage-3 AKI, and five well-known machine learning models are built for predicting Stage-3 AKI with 10-fold cross-validation (predictive accuracy were measured by AUC, precision, recall and F-score).ResultsMcDSL is useful for further research of EHR. It is able to discover four causal features, all selected features are medications that are modifiable. The latest research of machine learning is employed to compare the performance of prediction, and the experimental result has verified the selected features are pivotal.ConclusionsThe features selected by McDSL, which enable us to achieve significant dimension reduction without sacrificing prediction accuracy, suggesting potential clinical use such as helping physicians develop better prevention and treatment strategies.

Highlights

Acute kidney injury (AKI), characterized by abrupt deterioration of renal function, is a common clinical event among hospitalized patients and it is associated with high morbidity and mortality
Multiple cause Discovery combined with Structure Learning (McDSL) [6], is a causality discovery method designed to uncover the true causal relations as well as multi-causes structures by effectively removing spurious features on high-dimensional data, which in turn would improve prediction performance
Causal risk factor discovery McDSL discovered four risk factors of Stage-3 AKI from those 891 features, all of which are medications and mostly pertinent to gastrointestinal system. They are 1) Sennosides, a laxative to treat constipation and empty the large intestine before surgery; 2) 1,2,6-hexanetriol, a moisturizing agent for various creams; 3) Famotidine, a medication used in the treatment of peptic ulcer disease and gastroesophageal reflux disease, and 4) Benzimidazole, a drug class includes many anthelmintic drugs used for the treatment of a variety of parasitic worm infestations

Summary

Introduction

Acute kidney injury (AKI), characterized by abrupt deterioration of renal function, is a common clinical event among hospitalized patients and it is associated with high morbidity and mortality. This study aims to discover the key factors and improve the detection of Stage-3 AKI. In accordance with Kidney Disease Improving Global Outcomes (KDIGO) criteria, AKI is staged into three phases with ascending severity and treatment complexity. Prediction of potential AKI, especially Stage-3 AKI, can help with early identification of the high-risk patients and allow more appropriate allocation of limited clinical resources [3]. Scholars have focused on the development of machine learning methods to facilitate early detection, diagnosis and intervention, helping clinicians to provide more suitable and timely management for patients at high risk for AKI, resulting in improved clinical outcomes. It has been argued that better use of electronic health records (EHR) is the key to realize this objective [4, 5]

Objectives

Methods

Results

Discussion

Conclusion