A Machine Learning-derived Risk Score Improves Prediction of Outcomes After LVAD Implantation: An Analysis of the...

Area of Science:

• Cardiology
• Medical Technology
• Machine Learning in Healthcare

Background:

• Left ventricular assist device (LVAD) implantation is a critical intervention for advanced heart failure (HF).
• Significant variability in post-LVAD outcomes necessitates precise pre-operative risk stratification.
• Existing risk assessment tools may not fully capture patient-specific mortality risks.

Purpose of the Study:

• To evaluate the Machine Learning Assessment of Risk and Early Mortality in Heart Failure (MARKER-HF) model.
• To assess the predictive accuracy of MARKER-HF for post-LVAD implantation mortality.
• To determine if MARKER-HF enhances risk prediction compared to the Interagency Registry of Mechanically Assisted Circulatory Support (INTERMACS) profile.

Main Methods:

• Analysis of 25,365 patients from the INTERMACS database.
• Categorization of patients into risk groups (low, moderate, high, very high) using the MARKER-HF score.
• Comparison of in-hospital and 1-year post-discharge mortality between MARKER-HF risk strata and INTERMACS profiles.

Main Results:

• MARKER-HF demonstrated a progressive increase in in-hospital mortality from 4.4% (low-risk) to 15.2% (very high-risk).
• The combined use of MARKER-HF and INTERMACS profile identified extreme mortality risks (3.5% lowest, 19.8% highest).
• MARKER-HF score was independently associated with 1-year post-discharge mortality (HR 1.27; P < .001), outperforming the INTERMACS profile.

Conclusions:

• The MARKER-HF score is a valuable tool for mortality risk assessment in LVAD recipients.
• MARKER-HF provides prognostic information that complements the INTERMACS profile.
• Utilizing MARKER-HF can aid shared decision-making for LVAD implantation.