Skip to content

Main Research Question

Main research question

How can a stack-agnostic Intermediate Representation (IR) be used to automate the translation and enforcement of healthcare compliance requirements within a CI/CD pipeline?

Problem Investigation

Healthcare software delivery faces a recurring mismatch between legal text and technical implementation. Manual compliance checks are often periodic, expensive, and weakly coupled to daily delivery workflows.

Stakeholders

  • Direct: ECC operational/compliance roles, DSTRCT engineering roles, and audit/vetting actors who decide release readiness.
  • Indirect: Patient/end user. The patient is a beneficiary of safer and more reliable delivery, but not a direct actor in certification workflow.

Why Patient Is Not a Direct Stakeholder

Shift-left compliance changes when and where controls are checked, not the release prerequisite itself. Software is still only released when certification and governance conditions are met. The most likely patient-facing effect is shorter lead time for compliant improvements, not direct participation in certification decisions.

Synthesis Framework

The main question is answered through four supporting sub-questions:

  • SQ0 establishes existing solutions and known limitations.
  • SQ1 defines the representational requirements of the IR.
  • SQ2 defines transformation from IR to technical guardrails.
  • SQ3 evaluates delivery/security impact and trade-offs.

Integration Matrix

Supporting Question Primary Output Role in Main Answer Remaining Uncertainty
SQ0 Existing Work Comparative baseline of tools and limits Justifies novelty and design choices External tools evolve quickly
SQ1 Meta-Model Required IR attributes and constraints Defines what must be encoded Completeness across all clauses
SQ2 Compiler Mapping Mapping strategy from IR to controls Defines how automation is operationalized Environment-specific edge cases
SQ3 Impact Observed or predicted delivery/security effects Validates utility and trade-offs Generalization to other contexts

Risk Management

All research risks are tracked in the centralized Research Risk Register, categorized by main question and sub-question scope.

Iteration Checkpoints

  • Checkpoint A: Stakeholder boundary validated with audit/compliance roles.
  • Checkpoint B: SQ0-SQ2 provide implementable design basis.
  • Checkpoint C: SQ3 establishes effect and limitations.

Evaluation Boundary

The thesis can validate expected effects using experiments, expert review, or case evidence, but cannot fully evaluate long-term behavior in all real-world operational contexts.

Evidence Map

  • Regulatory and scientific sources.
  • Stakeholder interview notes and decisions.
  • IR schema drafts and mapping examples.
  • Pipeline gate prototypes and validation outputs.
  • Risk register entries and compensating-control decisions.