A Universal Solution to Automated Compliance Verification for Clinical Trials
Abstract
Purpose
Clinical trials require the extraction of specific metrics from DICOM plan datasets for protocol-driven data analysis. Because these metrics vary widely across trials, it has been difficult to develop an automated system capable of supporting diverse study requirements. This study proposes a methodology and system architecture that enable universal, automated compliance verification across a broad range of clinical trials.
Methods
We designed a compact but comprehensive library of forty-six commonly used metrics, most parameterized by a variable x. To improve interpretability and flexibility, internal metric symbols are formed using two prefixes and eight suffixes, which define how data is extracted and what elements are evaluated. Users can create custom metric labels for exported records in standardized formats. Each metric can be paired with one of seven comparison operators to evaluate extracted values against protocol-defined goals and tolerance limits. To extend functionality, the system supports user-defined variables and arithmetic expressions, enabling the calculation of derived metrics from primary measurements.
Results
The system has been successfully deployed in multiple internal and multi-institutional clinical trials. Data analysis follows a largely automated workflow: (1) construct a protocol template using the built-in protocol builder; (2) load a dataset and evaluate plan compliance using the template; and (3) generate printable and exportable review reports. To date, the system has accommodated all clinical trials encountered by the authors.
Conclusion
A universal automated solution for compliance verification in clinical trials is feasible. A concise, well-structured metric set combined with variables and arithmetic operations provides an effective and versatile foundation for broad clinical trial support.