SEND Rules

SHACL for SEND Rules

The project team considered two alternative approaches to modeling the SEND Rules:

1. Create SHACL shapes based on the FDA Validator Rules Workbook and then apply those shapes to the data. The advantage of this approach is that shapes can be constructed to provide error messages that directly correspond to the result message in the FDA documentation. However, this approach results in the creation of many overlapping and redundant SHACL shapes and does not leverage the full power of SHACL validation.

2. Create modular SHACL shapes based on the data schema that satisfy the FDA Validator Rules Workbook and provide additional, comprehensive checks as re-usable modules. This approach makes it more difficult to tie validation result messages to the original FDA Validator Messages, but the rule identifiers can be included in the messaging. Future work could process the SHACL validation report to provide more user-friendly reporting.

The second approach was chosen for the project.

Not all the rules for the selected DM and TS domains are modeled. Some rules cross multiple studies (example: identifiers that must be unique across multiple trials) and can only be evaluated within the context of the single-study in the prototype. Other rules cross multiple domains that are not included in initial development and may be reconsidered as project scope expands to include additional domains.

DM was chosen for initial development and the list of relevant rules was seledcted from the FDA Validator Rules Workbook by filtering exclusively on the DM domain for SEND 3.0 (column G, filter ‘“DM’; column K, filter ‘X’ ). Filtering results in a list of 19 rules specific to the DM domain. Of these, only 14 are independent of other domains. Additionally, Rule SD1020 is dependent on the SEND ontology and may be added at a later time.

Table 1. Rules Exclusive to DM Domain

Table 2. Rules Exclusive to TS Domain