← Adaptive Systems
Research Program — Adaptive Interfaces for Human Flourishing

Caregiver Systems

Orchestrating support around the capacity of the person carrying the care.

Clinical visibility ends at discharge. For the next thirty days — when risk is highest — the person who carries the patient home is invisible.

Caregiver Systems studies the handoff from hospital to home as a question of capacity, not logistics — and asks how the surrounding system might notice strain early enough to help before it becomes a crisis.

A caregiver pauses at the kitchen table over a medication schedule and notebook, a mug in hand, looking away — a child's things in the background
Fig. 01 — The threshold no system can see across

Field study — recovery continues at home, in the hands of someone the chart never names.

A behavioral truth

The thirty-day blindness.

A hospital has full sight lines on a patient until the moment they leave. Then a spouse, an adult child, a friend takes over — managing medication, watching for symptoms, deciding when something is an emergency — and the system can no longer see any of it.

Caregiver strain doesn't show up in the record. It shows up in a readmission.

What stays invisible
Stress patterns that precede an escalation by days, not minutes.
Medication slips that come from cognitive overload — not unwillingness.
The decisions made alone, at three in the morning, outside any system.
The staff who carry quiet anxiety about what happens after discharge.

You cannot intervene on what you cannot see.

Narrative
Anna 52
Daughter and primary caregiver, post-discharge

“They discharged him to me. No one asked whether I could carry it.”

A binder of instructions. Eight medications on a new schedule. A list of symptoms to watch for, and no clear line between ordinary and urgent.

By day five
The schedule is harder to hold in mind. A new symptom is hard to read. Sleep is thin; patience is thinner. There is no one to ask whether this is an emergency.

When a caregiver's nervous system dysregulates, their ability to manage medication, read symptoms, and decide when to escalate collapses with it.

Friction — the caregiver paradox

A discharge is built to answer the wrong question.

The process optimizes for
Patient education — does the patient understand?
Logistical handoff — was the referral submitted?
Process compliance — were the boxes checked?
What it cannot see

Caregiver capacity is not a logistics problem. It is a nervous-system problem.

Stress escalates silently — the caregiver rarely calls until it is already a crisis.

Overload makes learning impossible — you cannot teach someone who is panicking.

The friction is invisible — it happens in the kitchen, at midnight, far from the system.

Research insight

A health system measures task completion — was the medication explained, was the follow-up booked. But the thing that predicts a return to the hospital is a state, not a task: how much load the caregiver is carrying, and whether they are still recovering between demands.

That state leaves observable traces — in behavior, in routine, and in physiology such as heart-rate variability. Read carefully, with context and explicit limits, those traces may mark a window of rising strain days before it becomes a crisis. This is where Somatag sits — the sensing surface of the loop.

This led to a reframing

Caregiver stress is not a personal, external thing. It is a clinical signal — and a signal can be noticed in time.

The approach

Behavioral systems mapping, in three layers.

The work begins with real caregivers in a real discharge — not the protocol, but what actually happens in the weeks after someone goes home.

01

Workflow Intelligence

Map the discharge journey as it is actually lived. Where does coordination break? Where does the caregiver quietly absorb the burden the system sheds?

02

Caregiver Burden Mapping

Locate the stress-point moments — medication timing, symptom confusion, escalation decisions — and read caregiver state through behavioral and physiological markers, including HRV.

03

Adaptive Intervention Design

Design low-friction support timed to the pre-escalation window — not another app, but a specific, deliverable response your team can carry. Grounding, escalation triage, a single next step.

The shift is from reactive readmission management to noticing the window where support still moves the needle.

What success could look like
Illustrative scenario — projected figures, modeled to show the shape of a result. Not yet validated in a live engagement.
The setup

A health system carrying a 25% thirty-day readmission rate in a post-acute heart-failure cohort. A solid discharge process, built around patient education. Caregiver stress entirely invisible.

What a study might surface
Cognitive overload in many caregivers by day five. Most adherence failures traced to confusion, not refusal. Pre-escalation signals five to seven days ahead of a readmission.
What we would design
A discharge that names caregiver stress, not only patient understanding. A plain triage for “is this an emergency?” usable at 3am. Early flags for high-load caregivers; just-in-time grounding.
Projected outcomes — illustrative
~18%lower 30-day readmission (25% → ~20.5%)
~40%fewer post-discharge escalation events
~30%improvement in caregiver burden scores
~2.4×modeled return on the intervention

These figures are a modeled illustration of the kind of change the approach is designed to produce. A live engagement is how they would be tested — and the proof layer is built into the work.

Evidence level
Evidence level Research-informed service concept

The approach draws on caregiver-burden research, behavioral and physiological sensing, readmission literature, and the studio's wider work on detection and timing.

The case study above is illustrative. The model has not yet been run end-to-end with a partner system, and its projected outcomes are hypotheses until a pilot tests them.

What a pilot would test
Whether caregiver signals reliably predict escalation
Whether the support window can be found without over-acting
Whether interventions reduce readmission, not just stress
Whether the workflow integrates without adding staff burden
The engagement

A ten-week sprint, from discovery to proof.

The proof layer is built in: a small caregiver cohort and a simple sensing pilot run during the engagement, so the recommendation arrives with evidence rather than theory.

Phase 01 · Weeks 1–2

Discovery

Map the discharge workflow, establish baseline metrics, and define the caregiver cohort.

Phase 02 · Weeks 2–6

Proof

A caregiver cohort uses a simple app through the critical window — burden logs, friction notes, HRV — to test whether escalation is predictable.

Phase 03 · Weeks 5–8

Design

Translate what the caregivers showed us into workflow changes, staff training, and a measurement framework.

Phase 04 · Weeks 8–14

Results

Run the pilot, measure outcomes, and deliver a financial analysis, a case study, and a roadmap for scaling.

View the engagement deck Scope and investment are shared in a first conversation, sized to your cohort.
Future hypothesis

If caregiver strain is a clinical signal, then the thirty days after discharge are not a blind spot — they are a measurable window. The first research priority is not proving that we understand a caregiver's experience.

It is noticing the break early enough that a proportionate response can keep the patient home.

What I learned

The patient is discharged. The care is not.

The most consequential part of a recovery often happens where the system has the least visibility — in a home, in the hands of someone who was never assessed, trained for, or asked about. Treating that person as a clinical asset, rather than an external variable, changes what becomes possible.

Support the people supporting your patients, and the readmission you prevent was never inevitable.

Connection to a larger body of work

Each investigates a shared question: what happens when the surrounding system carries more of the burden of adaptation?

View the Engagement Deck Explore Somatag Work With Christine →