Internal

With the new ERI 2, the occurrence of double reds has decreased significantly, but they are still present. Why?

 With the introduction of ERI2, the number of double red cases—where both simulated strategies predict ahigh risk of type 1A endoleak—has dropped by approximately 59%. This improvement is largely due to the expansion of the training and validation datasets (now 117 patients in the training set and 56 in the validation set), which has led to a better balance between sensitivity and specificity.

Despite this progress, a non-negligible proportion of double red cases remains—about 1 in 5. There are two main reasons for this:

• False Positives: Even with improved algorithm robustness, some double red results are due to false positives. In recent analyses, when ERI 2 was applied to previously archived double red cases, more than half had at least one simulation with a low ERI (unlikely or highly unlikely), suggesting that the new version reduces, but does not eliminate, false positives.
• Challenging Anatomies: Some patient anatomies are inherently unsuitable for EVAR, regardless of the device or strategy simulated. In these cases, the ERI will consistently indicate a high risk, reflecting the true clinical challenge rather than a limitation of the algorithm.

Therefore, while ERI 2 significantly reduces the frequency of double red cases, their continued presence highlights both the complexity of certain anatomies and the inherent limitations of any predictive model.

In practice, double red results should prompt careful clinical review, and we are happy to support you in this process