A hospital group we worked with was spending weeks credentialing each clinician before they could see patients. Every file — licences, training certificates, references, prior privileges — had to be read in full by a qualified reviewer to confirm nothing was missing, expired, or inconsistent. The reading itself was the bottleneck; a single application could run to hundreds of pages, and the reviewers were senior people whose time was scarce. New hires waited, and the medical staff office carried a permanent backlog.
What mattered most
- A clinician — not software — had to remain accountable for every privileging decision. That was non-negotiable, for patient safety and for the regulator.
- The files contained sensitive personal data that could not leave the group’s control or be used to train anything.
- A missed expiry or an unverified credential is a serious failure, so a tool that was confidently wrong would be worse than no tool at all.
- Every step needed to be auditable, so the committee could show how a decision was reached.
How we approached it
We framed the tool narrowly: it does not approve anyone. It reads the file and prepares a structured summary for the reviewer — which documents are present, which are missing, which licences are near or past expiry, where two parts of the file disagree. It points to the page each finding came from, so the reviewer checks the source rather than trusting a summary. The clinician still reads what matters and signs every decision; the tool just stops them hunting for the three problems buried in three hundred pages.
Everything runs inside the group’s own environment, with no data leaving it and nothing retained for training. Each action is logged, so the committee can reconstruct exactly what was flagged and what the reviewer did with it. We measured against the time from a complete file arriving to a signed decision, and against a target the medical staff office set for clearing the backlog — not against how often the tool’s summary happened to agree with the reviewer, which would have been the wrong thing to optimise.
Where it stands
Reviewers now open a file already knowing where to look, and spend their time on judgement rather than page-turning. The decision still belongs to a clinician, which is the point. The backlog has come down to something the office can manage, and the audit trail has made the committee’s job in front of the regulator easier rather than harder.