Bayesian Real-Time Decision Support for Strategy Optimization in Chronic Total Occlusion Percutaneous Coronary Intervention.

BACKGROUND

Complex percutaneous coronary intervention (PCI) for chronic total occlusion (CTO) is often prolonged, underscoring the need for timely strategy selection. Although several preprocedural scoring systems exist, they offer limited guidance regarding intraprocedural dynamics. This study aimed to develop a framework that updates success probabilities to support optimal strategy transitions during the procedure.

METHODS AND

RESULTS

Two endpoints were defined: (1) successful crossing using antegrade wire escalation (AWE); and (2) time-constrained success using either AWE (<15 min) or advanced techniques (<60 min, intravascular ultrasound [IVUS] guided or retrograde). Preprocedural complexity scores (J-CTO, PROGRESS-CTO, CT-RECTOR) and procedural responses were sequentially incorporated into a Bayesian model to update real-time success probabilities. Markov chain Monte Carlo simulation compared the procedural efficiency of advanced techniques. With an initial success probability of 50%, previous failure reduced the likelihood of AWE success to 26.5% and the time-based endpoint to 40.8%. Failure with intermediate-tip-load wires (~3 g) further decreased success probabilities to 6.1% for AWE and 30.5% for the time-based endpoint. When advanced strategies were required, IVUS-guided approaches showed a trend towards higher estimated procedural efficiency (76.6%) under feasibility-constrained conditions (mean difference: 3.0%; 95% credible interval: -5.6%, 11.4%).

CONCLUSIONS

A real-time Bayesian framework enables dynamic adaptation during CTO PCI and may help identify scenarios where earlier transition to advanced techniques is warranted based on evolving procedural information.