In short

What is clinical criticality scoring?

Clinical criticality scoring assigns every inbound clinical document an urgency tier — from critical to administrative — based on its content, not just its document type. The tier determines the response SLA, who it routes to, and what happens if it goes unactioned. Most document AI stops at classification; criticality scoring determines how fast a practice must respond.

The core problem

Why document type cannot set priority

Every one of these arrives labelled the same way as its harmless equivalent.

The lab result

A potassium of 4.1 and a potassium of 6.8 are both "lab result". One is filing. One is a phone call before the patient leaves work.

The referral

A routine dermatology referral and a two-week-wait suspected cancer referral are both "referral".

The imaging report

An unremarkable knee X-ray and an incidental pulmonary embolism finding are both "imaging report".

The prior auth letter

An approval is filing. A denial is a deadline with an appeal window attached. Same document type.

The discharge summary

Most are informational. A few contain a medication change nobody has actioned.

The correspondence

Buried in routine insurance correspondence, a request that expires in ten days.

Method

How scoring works

Four inputs, weighted per practice, resolved to one tier.

01

Content signals

The values, findings and language inside the document — not just its header or form type.

02

Document context

What kind of document it is, who sent it, and what it references, used as context rather than as the answer.

03

Practice configuration

Your specialty, your patient population and your thresholds. A cardiology practice scores differently from paediatrics.

04

Confidence

Where signals conflict or quality is poor, the item is escalated to human review rather than scored with false confidence.

The part other tools skip

Criticality tiers and response SLAs

A lab result and a critical lab value are the same document type. They are not the same emergency. Provenza scores on what is inside the document, then puts a clock on it.

Critical
Critical lab values, STAT imaging findings, results requiring same-day clinical action
Set per practiceminutes
Immediate notification to the on-call clinician, with supervisor visibility
Urgent
Prior authorization denials, urgent referrals, medication and coverage issues carrying a deadline
Set per practicehours
Queue owner notified, supervisor alerted if the window closes
Routine
Routine lab results, consult notes, discharge summaries, standard referrals
Set per practicebusiness days
Daily digest to the owning queue
Administrative
Insurance correspondence, records requests, vendor and marketing faxes
Set per practicebusiness days
Batched to the administrative queue

Tier definitions and response windows are configured per practice and signed off clinically before go-live. Aurus does not publish generic clinical response thresholds, because the right window depends on your specialty, staffing and patient population — not on what reads well on a website.

Questions

Common questions

What is clinical criticality scoring?

It assigns every inbound clinical document an urgency tier based on its content rather than its document type. The tier determines the response SLA, the routing and what happens if the item goes unactioned.

Why is classification not enough?

Because two documents of the same type can carry completely different urgency. A routine lab result and a critical lab value are both "lab result". Classification cannot separate them; only reading the values inside can.

Who sets the tier definitions?

You do, with clinical sign-off, before go-live. Aurus does not publish generic clinical response thresholds because the right windows depend on your specialty, staffing and patient population.

What happens when the score is uncertain?

It escalates to human review rather than being assigned a tier with false confidence. We would rather send you something to check than quietly get it wrong.

Can tiers be changed after go-live?

Yes. Tier definitions and thresholds are configuration, not code, and changes are versioned and logged so you can see what was in force when any historical item was scored.

Related

Keep reading

AI fax triage

The full pipeline this scoring sits inside.

Response SLAs

What happens once a tier is assigned.

Provenza overview

The product end to end.

How many of yesterday’s faxes were scored by a human guess?

Send a week and see how Provenza would have tiered them. Free, under a BAA, one business day.

Card required to start · BAA before PHI · Answer in one business day