Is cross-system clinical trial automation failing not because workflows cannot be executed, but because protocol terminology cannot be interpreted consistently across systems and clinical studies?
Several academic and industry initiatives have proposed encoding protocol elements, such as eligibility criteria, visits, milestones, consent states, into structured system logic. While technically feasible, these efforts consistently encounter a semantic problem: the protocol terms being encoded do not have a single, invariant meaning. Instead, their interpretation is context-dependent and often only becomes clear when the protocol is implemented across multiple systems such as CTMS, EDC, TMF, and finance. Automation struggles not because systems cannot execute logic, but because they cannot reliably infer what a term is supposed to mean.
A well-known example is FPI (First Patient In). On the surface, FPI appears to be a simple milestone, frequently used in project plans, dashboards, and sponsor communications. In reality, its meaning varies widely across protocols. In some studies, FPI refers to the date of informed consent; in others, it marks the first screening procedure, the first randomization, or the first dose administration. Each interpretation is valid within its own protocol context. However, when FPI is mapped into systems, problems emerge. CTMS may treat FPI as a project milestone, finance may use it to trigger budget recognition, EDC may infer it from visit data, and TMF may never record it explicitly at all. The same label points to different underlying events, none of which is universally “correct.”
| Interpretation of FPI | Typical system usage | Practical implication |
|---|---|---|
| First informed consent signed | CTMS, project timelines | FPI occurs before any clinical data exists; operational start is assumed early |
| First screening procedure performed | EDC, visit scheduling | FPI inferred indirectly from visit data rather than explicitly recorded |
| First randomization | IRT / RTSM | FPI tied to eligibility confirmation, not patient contact |
| First dose administered | EDC, safety reporting, finance | FPI becomes a safety-relevant and sometimes financial milestone |
| Protocol-specific composite definition | TMF, regulatory narrative | Meaning clarified only through documentation and justification |
From a systems perspective, this ambiguity is toxic. Automation assumes that a term like FPI corresponds to a specific event or timestamp. When it does not, logic either becomes hard-coded (and wrong for some protocols), or manual overrides and reconciliations proliferate. The issue is not a lack of interoperability, but the absence of shared semantic agreement about what the term represents in a given study. One word in the protocol can silently mean different things across systems, vendors, and stakeholders, forcing humans to continuously translate intent.
This pattern repeats across many commonly used terms: Screening, Baseline, End of Treatment, Study Completion. Each appears simple, yet each acquires meaning only through protocol context. As long as protocols rely on natural language without enforced naming conventions and explicit semantic definitions, clinical systems will require protocol-specific configuration and manual interpretation. Automation, in this sense, does not fail because it is too ambitious, but because it assumes linguistic precision where none formally exists.
A more realistic path forward is not to eliminate interpretation, but to recognize naming conventions as a first-order design problem in clinical research systems. Without stable, protocol-specific definitions of key terms, no amount of technical sophistication can deliver reliable automation. Before workflows can be executed, meanings must be agreed upon. In clinical research, that work remains largely manual and unavoidably so.
A practical solution is to treat FPI not as a single milestone, but as a small family of explicitly qualified events, such as FPI(s) for first screening or FPI(e) for eligibility confirmed. Each with a defined meaning that can be consistently implemented across systems. For example:
| Identifier | Meaning |
| FPI(s) | First patient screened |
| FPI(c) | First informed consent |
| FPI(e) | Eligibility confirmed |
| FPI(r) | First randomization |
| FPI(d) | First dose administered |
