PASTEC: an open clinical infrastructure for data aggregation and prospective validation of AI in cardiac remote monitoring.

Abstract

Remote monitoring (RM) of cardiac implantable electronic devices generates large volumes of data that must be reviewed using vendor-specific platforms, creating a growing clinical workload. Although artificial intelligence (AI) could alleviate this burden, the absence of infrastructures enabling prospective data aggregation and evaluation within routine workflows limits clinical translation. We developed PASTEC, an open platform integrated into existing RM systems via a browser extension, enabling client-side pseudonymization, annotation, and centralized academic data storage. Platform usability was prospectively evaluated during one week of routine activity in a high-volume RM center and compared with a week of standard RM review. A total of 1276 recordings were reviewed, including 697 with structured annotation. Median processing time per recording was 5.31 s with annotation versus 2.88 s without annotation (p < 0.01), corresponding to a limited additional workload compatible with routine clinical activity. No identifiable patient information was detected in reviewed pseudonymized data. PASTEC provides a scalable infrastructure addressing a key bottleneck in the prospective clinical validation of AI for cardiac remote monitoring.