Taking ACTION: A Prognostic Tool for Pediatric Ventricular Assist Device Mortality

Area of Science:

• Pediatric cardiology
• Medical device technology
• Machine learning in healthcare

Background:

• Pediatric ventricular assist device (VAD) therapy is complex.
• Accurate preimplantation risk stratification is crucial for patient outcomes.
• Existing risk models may not reflect contemporary VAD use in children.

Purpose of the Study:

• To develop and validate a novel risk assessment tool for pediatric VAD candidates.
• To estimate the risk of mortality on the device using preimplantation clinical data.
• To create a model that aids in patient selection and team expectation management.

Main Methods:

• Utilized the Advanced Cardiac Therapies Improving Outcomes Network (ACTION) registry data (2012-2021).
• Developed a risk prediction model using a random forest machine learning algorithm.
• Assessed model performance using the area under the receiver operating characteristic curve (AUC).

Main Results:

• Identified nine significant preimplantation risk factors for mortality.
• The final predictive model demonstrated excellent discrimination with an AUC of 0.95.
• The model provides a framework for pediatric-specific risk profiling.

Conclusions:

• The developed risk assessment tool accurately predicts mortality in pediatric VAD candidates.
• This tool can inform clinical decision-making, optimize patient selection, and improve outcomes.
• Contemporary clinical data and machine learning enhance risk stratification for pediatric VAD support.