JMIR Publications

ABSTRACT

Background:

Atrial fibrillation (AF) ablation is an effective treatment for reducing episodes and improving quality of life in patients with AF. However, long-term AF-free rates after AF ablation are inconsistent across the population. Thus, there is a need to address the limited benefits some patients experience by developing predictive algorithms to identify long-term AF ablation outcomes. Yet, current patient selection relies on individual clinical assessment and highlights a critical gap in population-level predictive analytics. Leveraging large administrative claims databases represents an opportunity to optimize procedural outcomes and resource allocation at a national level

Objective:

This study utilizes machine learning models on claims data to explore if integrating International Classification of Diseases (ICD) billing codes outperforms traditional stroke and AF risk scores in predicting AF ablation outcomes.

Methods:

The Merative MarketScan® Research Medicare data was used to examine claims for AF ablation. To predict 1-year AF-free outcomes after AF ablation, logistic regression and XGBoost models were used. Model predictions were compared with established risk scores CHADS2, CHA2DS2-VASC, and CAAP-AF. The machine learning models were also assessed on subgroups of patients with paroxysmal AF, persistent AF, and both AF and atrial flutter from October 2015 onwards.

Results:

Among 14,521 patients with claims for AF ablation, XGBoost achieved an area under the receiver operating characteristic curve of 0.528, 0.521, and 0.529 for the entire AF ablation population, female, and male, respectively. Machine learning models outperformed risk scores in all AF ablation patient groups (whole population, female, and male).

Conclusions:

Machine learning models including ICD codes demonstrated better performance to conventional risk scores in predicting AF ablation outcomes. This approach can enhance population health management by optimizing patient selection, reducing unnecessary procedures, and improving cost-effectiveness.