Global Hypoperfusion Index (GHI) Algorithm

The Global Hypoperfusion Index (GHI) is a software algorithm designed to predict the likelihood of an impending global hypoperfusion event in adult critical care patients by analyzing real-time data from a Swan-Ganz catheter. Utilizing a machine-learning model, it provides an adjunctive, predictive cardiovascular indicator approximately 10-15 minutes before mixed venous oxygen saturation (SvO2) is expected to drop to critical levels, aiding clinicians in anticipating hemodynamic instability.

The Global Hypoperfusion Index (GHI) algorithm is a software feature for compatible monitoring platforms, intended for use by qualified clinicians in critical care hospital settings. Its primary function is to offer physiological insight into an adult patient's likelihood of experiencing a future global hypoperfusion event, defined as mixed venous oxygen saturation (SvO2) dropping to 60% or below for at least one minute. Designed for both surgical and non-surgical adult patients undergoing advanced hemodynamic monitoring with a Swan-Ganz catheter, the GHI algorithm aims to identify patients at risk for clinical deterioration, as well as those at low risk. The predictions provided are adjunctive, serving as a reference and not as the sole basis for therapeutic decisions. The GHI algorithm employs a machine-learning model that analyzes real-time patient health data to compute the risk of an impending global hypoperfusion event. It is designed to alert clinicians on average 10-15 minutes before the SvO2 is predicted to reach the critical threshold of 60%. The system outputs this risk as an index from 0 to 100, where higher values signify an increased likelihood of the event. The algorithm's performance was validated using clinical data, including prospective analyses of retrospective data from a diverse set of adult surgical patients with invasive monitoring, confirming it met its design and performance specifications.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What specific input parameters from the Swan-Ganz catheter and other real-time patient data are most heavily weighted by the GHI machine-learning model in predicting hypoperfusion?; How does the GHI algorithm's performance (sensitivity, specificity, positive predictive value, negative predictive value) vary across different subpopulations of adult patients (e.g., cardiac surgery vs. sepsis vs. trauma)?; Given the 10-15 minute predictive window, what are the recommended clinical actions or protocols when the GHI indicates an increased likelihood of a hypoperfusion event, and how effective are these interventions in preventing the event or mitigating its severity?; The definition of a global hypoperfusion event is SvO2 ≤ 60% for at least 1 minute. What is the rationale for this specific threshold and duration, and how does the algorithm's predictive accuracy change if these parameters are modified?; How does the GHI algorithm account for or adapt to common interventions in the critical care setting (e.g., vasopressor administration, fluid boluses) that might acutely alter hemodynamic parameters and potentially influence the SvO2 reading?