New England CareFlow

The research

Measuring what a visual note changes for the nurse reading it.

The pilot study

Forty-one registered critical-care nurses reviewed the same patient admission twice — first as conventional written SOAP notes, then as a structured visual note — and scored their perceived workload on the raw NASA-TLX after each.

Workload · written notes
54.95
Workload · written notes
Workload · visual note
30.68
Workload · visual note
Change (P < .001)
−44.2%
Change (P < .001)

Improvements appeared on all six subscales. As a fixed-order pilot on a single case, the finding is preliminary: order and familiarity effects cannot be ruled out.

Read the preprint

Two guarantees

The body systems are fixed

Fourteen body systems, identical for every patient, taken from the headers nurses actually write. A system can never be dropped because a model forgot it.

Every fact shows its source

Each statement on the chart carries the note it came from and its exact words. Anything that can't be found in the source never reaches the chart.

How it works

Five steps take a nurse's narrative notes to a verified chart.

  1. 01

    Segment

    Notes are split on the section headers nurses already write — NEURO, RESP, CV. No AI involved.

  2. 02

    Extract

    Each section is mapped onto a fixed set of fourteen body systems, identical for every patient.

  3. 03

    Ground

    Every fact keeps a link to the exact words it came from, and is checked against the source.

  4. 04

    Cover

    Sentences left unused get a second pass, so nothing is quietly dropped.

  5. 05

    Render

    The verified result is drawn the same way every time — same notes, same chart.

See it for yourself

Visual Notes runs on a synthetic sample case — no real patient data.

Open Visual Notes