Stop scheduling from memory

Every rehab facility has a scheduler who knows things.

They know that Mrs. R will not be in any group therapy ever — she had a bad experience three years ago and made it clear she will not repeat it. They know Mr. K and his roommate Mr. T cannot be in the same group session, for reasons nobody really wants to write down. They know Patient 12 only does well with the morning PT, who happens to also work with Patient 18 most mornings, who himself does best with the same OT three times a week.

This is the institutional memory of rehab scheduling. Most of it lives in one person’s head, very little of it lives in any system, and when the scheduler takes a vacation, the substitute books Mr. K and Mr. T into the same balance group within forty-eight hours.

What is actually in the scheduler’s head

If you sat down with a senior scheduler at any rehab unit and asked them to list everything they are tracking for next week, you would get something like this:

• “Patient X always asks for Linda. Whenever Linda is open, give them to her.”
• “Patient Y had a complaint last year about Linda. Keep them apart.”
• “Patient Z’s plan of care says they can only be evaluated by Dr. M. No exceptions.”
• “These three patients can never be in a group together. Don’t ask.”
• “This patient gets agitated in groups. Skip them on group day.”

Some of this is clinical (a plan-of-care requirement). Some is interpersonal (two patients who do not mix). Some is preference (a patient who connects best with one provider). All of it matters for the schedule — and almost none of it is captured in a way the next person can pick up on Monday morning.

This is the part of the job that does not show up in any training manual, and it is also the part that breaks first when the scheduler is out.

Two building blocks

Tempo handles this with two centralized data structures the whole facility shares.

Patient-provider preferences — three flavors:

• Preferred — when this provider is available, prefer them for this patient. Soft constraint; the engine uses it as a tie-breaker when multiple assignments would otherwise work.
• Avoid — keep this patient and this provider apart unless there is no other option. The engine routes around the pairing when feasible and flags the schedule when it cannot.
• Required — this patient is seen only by this provider. Hard constraint; the engine refuses to violate it and surfaces the slot as unfillable if the required provider is unavailable.

Patient group restrictions — two flavors:

• No groups — this patient is never placed in any group therapy session. Honored at group composition time, every week, automatic.
• Specific patient — this patient cannot be grouped with another specific named patient. The engine will never put them in the same therapy group, no matter what.

Each of these is entered once, on the patient’s profile, by anyone with the right role. Intake can flag a no-groups patient on day one. The DOR can record a required-provider relationship the day a complaint is logged. The scheduler can mark an avoid pairing the moment they hear about it. None of it requires a new tool or workflow — it is part of the patient record alongside disciplines, plan of care, and discharge target.

What the engine does with them

When Tempo generates a week, it walks every cell of the schedule and asks: does placing this patient with this provider, in this slot, with this group composition, violate any preference or restriction on file?

• Required preferences are hard-locked. The engine assigns the required provider or surfaces a gap.
• Avoid preferences are a strong negative weight. The engine routes around them when possible, warns when it cannot.
• Group restrictions are enforced at group composition. A no-groups patient never appears in a group session. Two patients flagged with a specific-patient restriction never share a group, ever.
• Preferred relationships are tie-breakers — small individually, but across a week of 200+ assignments they meaningfully shape who works with whom.

The whole calculation runs in well under a minute for a 25-patient unit with eleven providers. The output is a balanced week that respects every preference and restriction in the system, with warnings flagged where the engine had to compromise.

Why this matters

The reason most rehab teams have not automated their scheduling is not that the math is impossible. It is that the math depends on context that lives in one person’s head. When the scheduler takes Friday off, the substitute does not know about Mr. K and Mr. T, and on Tuesday someone files a complaint.

Once the preferences and restrictions live in the system, the scheduler stops being a single point of failure. They become someone who reviews and adjusts, not someone who reconstructs the schedule from scratch every Monday — and the substitute does not have to learn twenty years of facility lore in an afternoon to cover a vacation.

If your facility runs on someone’s memory of who can sit next to whom, join the Tempo waitlist and we will show you what it looks like when that knowledge belongs to the building, not the person.