Multisource Annotation Completeness of Cardiovascular ATC-C Drugs in Biomedical and Cheminformatics Databases

Cardiovascular bioactive compounds are represented across multiple biomedical and chemical databases, but their annotations are often fragmented, heterogeneous, and unevenly complete. This limits the reuse of such data for chemoinformatics, drug classification, and translational interpretation.
This study aimed to evaluate the annotation completeness of cardiovascular drugs classified under ATC group C by developing a multisource Annotation Completeness Score.
A harmonized reference dataset of 657 cardiovascular ATC level-5 entries was analysed. The 15 ACS indicators were organized into four functional annotation layers representing semantic harmonization, chemical structure representation, clinical-regulatory annotation, and biological pathway integration. Each available annotation contributed one point to the final 15-point multisource Annotation Completeness Score.
The mean multisource Annotation Completeness Score was 8.26 out of 15, with a median of 8 and a range from 1 to 15. Overall, 214 entries were classified as richly annotated, 184 as moderately annotated, and 259 as poorly annotated. Annotation completeness differed significantly across ATC level-2 therapeutic subclasses, with the highest mean score observed for calcium channel blockers and the lowest for lipid-modifying agents. Combination-like entries showed significantly lower annotation completeness than non-combination-like entries. Component-level reconstruction was retained as a supporting analysis to interpret representational mismatches in fixed-dose and multi-component products.
The results show that database presence alone is insufficient for assessing the scientific usefulness of cardiovascular drug records. The proposed multisource Annotation Completeness Score provides a structured framework for distinguishing richly annotated records from superficially represented entries and supports more reliable downstream bioinformatics and cheminformatics analyses.