eCAPRI: a novel tool combining clinical and imaging data for post-TAVI mortality prediction

Comprehensive Summary

This study aims to create an enhanced and automated version of the CAPRI program in order to more accurately and efficiently predict 1 year mortality after a transcatheter aortic valve implantation (TAVI). The CAPRI system was developed in a previous study completed by Lantelme and colleagues. The CAPRI system utilizes biological data along with thoracic aortic calcium and volume extracted from CT to predict the 1-year mortality; however, due to it requiring a lot of time and effort, it is not suited for clinical use. The authors used a deep learning model to automate the thoracic aortic calcium results based on the CT scan given. Additionally biomarkers were assessed using a deep learning model. Both of these two characteristics make up the eCAPRI score. The program’s accuracy (eCAPRI) was analyzed by testing it on the training data set and comparing it to the results of the CAPRI score and EuroSCORE based on their results from that same set. The eCAPRI score was seen to be the most accurate out of the three. However before being used in clinical practice, the program should be tested on a more diverse population.

Outcomes and Implications

The research completed by this study allowed for the generation of a more accurate and efficient CAPRI model to be created. The use of AI in the eCAPRI system allows for the thoracic aortic calcium to be calculated much quicker than when done by the original CAPRI formula. It also was seen to have more accurate results. The eCAPRI program allows for patients with higher risk for adverse outcomes to be identified earlier allowing ample time for treatments.