An Open Source Benchmark for Pseudonymization Services in Translational Research

Pseudonymization is a privacy-enhancing technology for the collection and use of healthcare data. One important application is the pseudonymization of large datasets prior to secondary use, for example in research data platforms. Such applications require high scalability, but concepts and tools for systematic performance assessments for such services are lacking. We developed an open source benchmarking tool for pseudonymization services. The tool simulates real-world workloads with configurable request distributions, supporting diverse scenarios such as read- or write-heavy use cases. For example, it supports scenarios where primarily new pseudonyms are generated, such as when preparing a dataset for secondary use, or scenarios where pseudonyms are primarily resolved, e.g. when new data is assigned longitudinally. Key features include multi-threading, automated authentication, identifier handling, and detailed performance reporting. Our proposed concept provides realistic performance analyses, incorporating network-related factors and supporting continuous delivery pipelines. Our implementation includes a modular connector component to enable seamless extendibility for benchmarking new pseudonymization services or adapting to evolving requirements.