Intelligent Clinical Decision Support

Abstract

Early recognition of pathologic cardiorespiratory stress and forecasting cardiorespiratory decompensation in the critically ill is difficult even in highly monitored patients in the Intensive Care Unit (ICU). Instability can be intuitively defined as the overt manifestation of the failure of the host to adequately respond to cardiorespiratory stress. The enormous volume of patient data available in ICU environments, both of high-frequency numeric and waveform data accessible from bedside monitors, plus Electronic Health Record (EHR) data, presents a platform ripe for Artificial Intelligence (AI) approaches for the detection and forecasting of instability, and data-driven intelligent clinical decision support (CDS). Building unbiased, reliable, and usable AI-based systems across health care sites is rapidly becoming a high priority, specifically as these systems relate to diagnostics, forecasting, and bedside clinical decision support. The ICU environment is particularly well-positioned to demonstrate the value of AI in saving lives. The goal is to create AI models embedded in a real-time CDS for forecasting and mitigation of critical instability in ICU patients of sufficient readiness to be deployed at the bedside. Such a system must leverage multi-source patient data, machine learning, systems engineering, and human action expertise, the latter being key to successful CDS implementation in the clinical workflow and evaluation of bias. We present one approach to create an operationally relevant AI-based forecasting CDS system.

Keywords: database; hemodynamic monitoring; machine learning; predictive analytics.

Figure 1

Model prediction of initial tachycardic episode using external control data matched for every episode of tachycardia. Comparison of a model trained on MIMIC-II data to identify an initial episode of tachycardia (heart rate (HR) > 130/min) in an external validation cohort from that same database. Results are shown as risk score changes over time as the future tachycardic and non-future tachycardic (control) groups move toward the event. The control group’s time series data were time-matched to correspond to the future tachycardic group’s time in the ICU. Data derived from Yoon et al. [43].

Figure 2

Activity monitoring operating characteristic analysis of models developed with increasingly granular arterial pressure physiologic data, for models developed using a universal baseline (A), and models developed using a personalized baseline (B) (see text for details). Displayed as the time to detection of bleeding versus false-positive rate for arterial catheter data only for increasing granularity levels: simple metrics (SM), beat-to-beat (B2B), and waveform (WF). Data displayed with shading equal to 95% confidence range. Data derived from Pinsky et al. [41].