Many diagnostic error cases aren’t about one single mistake—they’re about how information moved through the system. In Imperial, that can look like:
- A patient is routed through triage quickly, but symptoms that don’t “fit” a prediction are treated as less urgent.
- Imaging or lab results are processed through automated workflows, then acknowledged late or inconsistently.
- Referral instructions are unclear, or follow-up doesn’t happen until symptoms worsen.
- Documentation is generated or organized with automation, but key clinical context gets missed.
California law looks at whether care met the standard of care at the time. That standard includes how clinicians should verify automated outputs, escalate concerns, and communicate risk. When AI is involved, the question becomes: Was the tool treated as advisory or as definitive? And did the care team respond appropriately when results conflicted with objective findings?


