查看文章

ARF is one of the most common indications for intensive care unit (ICU) admission. The clinical decision of invasive or noninvasive ventilatory support depends on the patient's disease progression and severity. HRV is the characterization of the functional status of the autonomic nervous system (ANS) and has been studied for risk stratification for various cardiac and noncardiac diseases. There have not been any studies that examined the utility of HRV for the prediction of ARF. The objective of this retrospective study is to develop an interpretable machinelearning model for ARF prediction using features derived from continuous physiological data and Electronic Health Record (EHR) among adults in the ICU. 1873 patients admitted to Emory ICUs between December 2016 and March 2018 with mechanical ventilation (n= 991) and matched controls (n= 882) were included in this retrospective …