An AI-driven EuroTR risk score trained on a large real-world registry (N=1,826; derivation n=1,225, validation n=601) stratified 1-year mortality after tricuspid transcatheter edge-to-edge repair (T-TEER), separating low- vs high-risk patients (HR 4.26; 95% CI 2.71–6.67). Overall 1-year survival was 82.1% (95% CI 80.1%–84.2%), with similar performance across derivation and validation cohorts.

Why It Matters To Your Practice

• T-TEER candidates often have advanced right-sided disease and competing comorbidities, making pre-procedure risk conversations and selection decisions high-stakes.

• An externally validated, registry-based AI score may offer more individualized 1-year mortality estimates than relying on clinician gestalt or non–procedure-specific models.

Clinical Implications

• Use EuroTR-style risk stratification to support shared decision-making (expected benefit vs risk) and to frame goals-of-care discussions before T-TEER.

• In multidisciplinary valve conferences, a standardized score can help align thresholds for referral, intervention timing, and post-procedure surveillance intensity.

• Consider risk tiering when planning peri-procedural optimization (volume status, renal function, anticoagulation strategy, and hemodynamic assessment).

Insights

• The model was built on 18 clinical, lab, echo, and hemodynamic features using extreme gradient boosting—an example of “AI as a calculator” embedded in routine variables rather than novel sensors.

• External validation (separate validation cohort) is a key differentiator from many single-center ML tools and improves the odds of transportability.

• The output is best viewed as risk stratification (low vs high risk) to inform decisions—not as a stand-alone directive to treat or defer.

The Bottom Line

• In EuroTR registry patients undergoing T-TEER, an AI-derived EuroTR score stratified 1-year mortality (HR 4.26) with overall 1-year survival ~82%.

• Near-term impact for clinicians: better calibrated pre-procedure counseling and more consistent heart-team decision-making using routine data.