A Delphi Process to Identify Relevant Outcomes That May Be Associated With a Predictive Analytic Tool to Detect...

Area of Science:

• Critical Care Medicine
• Artificial Intelligence in Healthcare
• Clinical Decision Support

Background:

• Growing use of AI tools for early patient event detection in ICUs.
• The hemodynamic stability index (HSI) shows promise (AUC 0.82) for predicting hemodynamic intervention needs.
• Need to align future AI tool studies with clinically relevant and achievable outcomes.

Purpose of the Study:

• To identify the most relevant and achievable outcomes for AI tools, specifically the HSI, in the ICU setting.
• To guide future research in developing and validating AI predictive analytics for critical care.

Main Methods:

• A three-round Delphi study involving 10 critical care physicians and 3 nurses in the US.
• Panelists rated potential outcomes for relevance using a 5-point scale (≥65% rating 4 or 5 for consensus).
• Kemeny-Young approach used to rank outcomes based on relevance and achievability.

Main Results:

• Consensus on relevance was achieved for 19 out of 24 potential outcomes.
• "Reduces time spent in hemodynamic instability" and "reduces times to recognition of hemodynamic instability" ranked highest.
• A matrix visualized the distribution of outcomes by relevance and achievability.

Conclusions:

• The Delphi study effectively identified relevant and achievable outcomes for ICU AI tools.
• Findings offer guidance for researchers aiming to improve critical care outcomes with AI.
• Future research should focus on AI tools targeting clinically significant outcomes like reduced hemodynamic instability duration and faster recognition.