From referral to report, with a clinician at every decision.
HealthAgency is a governed workflow platform for private MRI, ultrasound and specialist clinics. It takes the inbox-and-spreadsheet work out of referrals, vetting, booking, patient messages and report delivery, and leaves an evidence trail for every step. The software suggests. Your radiologists decide.
Talk about a pilotSee how it worksThe problem it solves
A referral arrives as a form, an email or an attachment. Someone re-keys it, checks it is complete, asks a radiologist whether the scan is right, finds a slot, chases the patient, and later chases the referrer about the report. Each hand-off is a place for delay, error and missing evidence.
Slow vetting
Radiologist time spent reading incomplete or routine referrals that could be pre-checked and queued.
Lost patients
Invitations, reminders and replies handled by hand, so slots go empty and patients drop out.
Report chasing
Staff chase referrers to confirm a report arrived, with no proof it was opened.
Weak audit trail
Hard to show who decided what, when, and on which version of the information.
How it works
One governed path from first contact to a delivered report.
Product tour
The staff console and the patient page, as they look today. Select a picture to enlarge it.










Prototype screens on synthetic data only: every name, number and message here is invented. Not in clinical use.
Governed, not just automated
Anything that leaves the system, such as a message, a booking confirmation or a report, follows the same chain. Nothing is sent on a model's say-so.
Clinicians keep the decisions
The software never clears MRI safety, never downgrades a red flag and never writes patient-facing text with a model. Low confidence or an error always routes to a person.
Unknown means unknown
If a provider does not confirm a send, the platform checks before it tries again, so patients are not messaged twice.
Evidence by default
A tamper-evident audit chain, a record of every decision and its inputs, and signed receipts for external actions.
Privacy built in
Tenant isolation enforced in the database, encrypted contact data, subject access export, restriction, retention and legal-hold controls, and logs that carry identifiers rather than patient detail.
Pick the pieces you need
A small fixed core, with everything else plugged in. A clinic switches on the modules, channels and rules that match its location and specialties.
Feature modules
Referral intake, vetting assist, booking, patient communications, report distribution.
Swappable decision engine
Starts with Jev. Replace it with Microsoft's Decision-1, Laya or another model by configuration. The engine only suggests: policy decides what may happen on its own.
Jurisdiction packs
Wales and England rules as data: residency, retention, language and standards. Add a region without changing the core.
Specialty packs
MRI, ultrasound and musculoskeletal referral templates, protocols and preparation text.
Three deployment profiles: hosted in the UK, inside the clinic's own cloud tenant, or hybrid with references only.
Where it stands, honestly
The platform is built and tested on synthetic data. It is not yet in clinical use.
| Area | State |
|---|---|
| Core, referral intake, vetting, booking | Built |
| Patient communications (email, guarded SMS) | Built SMS provider not chosen |
| Report distribution | Built Production malware scanner not chosen |
| Privacy, governance and observability | Mechanisms built Policies need sign-off |
| Staff console and public booking and report pages | Built Welsh translation pending |
| Production deployment and calibration | Next |
| Claims and insurance | Deferred |
What a pilot looks like
Start with one clinic, one modality and the referral-to-booking path. Run in shadow first: the platform makes its suggestions alongside your current process, and your team compares them with what actually happened. Switch on automation only where the evidence supports it, with your radiologists setting the confidence thresholds.
Contact ScaleRight