StudyWeave Loom

Loom does the first pass. You make the calls.

Loom is StudyWeave’s AI. It reads your protocol, tables and footnotes included, and drafts the study map and then the eCRFs. Everything it takes from the protocol points back to its page, and what it can’t support comes to you as a question.

  1. 1 · Reads
  2. 2 · Cites
  3. 3 · Asks
  4. 4 · Hands over
EndpointVisitAssessmentForm

Loom reads the protocol, schedule tables and their footnotes included, and tells you about any page it couldn’t read cleanly.

ProtocolS1505 · ClinicalTrials.gov NCT02562716
Primary objectives

To assess 2-year overall survival in each treatment arm

Detailed description · Outline

Treatment repeats every 28 days for 3 courses in the absence of disease progression or unacceptable toxicity. Patients achieving stable disease or better undergo pancreatectomy 4-8 weeks after completion of first 3 courses of treatment. … patients are followed up every 3 months for 2 years and then every 6 months for 2 years.

Secondary objectives

frequency and severity of adverse events associated with chemotherapy in the perioperative setting. … overall response rate following preoperative chemotherapy … per Response Evaluation Criteria in Solid Tumors (RECIST) 1.1.

Outcome measures

within 4 to 8 weeks after the last dose of Cycle 3 preoperative chemotherapy … R0 resection classification is defined as macroscopically complete tumor removal with negative microscopic surgical margins

Study mapDraft · Example
Endpoints
Overall survivalPrimary · 2 years
Visits
ScreeningBefore registration
Preoperative cycles 1–3Every 28 days
Surgery4–8 weeks from when?
Needs a decisionThe outline counts from the end of the first 3 courses; the outcome measures from the last dose of Cycle 3.
Follow-upEvery 3, then 6 months
Assessments
ResponseRECIST 1.1
Resection classificationR0, R1
Adverse eventsPerioperative
eCRFsDrafted by Loom · Example
Response assessment
Overall response
Date of restaging scan
Surgery
Resection classification
Date of pancreatectomy Waiting on your decision
Adverse events
Event term
Grade
Nothing is applied until a person reviews it. Review and sign

While you do other things

It works in the background.

Start Loom and get on with something else. A notice tells you when it’s done, and you can open the job at any point to see how far it has got, or stop it.

Studies / S1505 / Protocol

Protocol

Version 1 · uploaded

Loom · Drafting the study mapStop
Loom · The study map draft is ready for review.
Illustrative. The loader is the app’s working indicator.

What Loom can read and build.

Reading

  • Every page’s text, keeping the exact place of each passage it uses: page, section and words.
  • Schedule-of-assessments tables, rebuilt as grids with their footnote marks, then checked.
  • A report of any page it couldn’t read faithfully, instead of skipping it.

Building

  • Forms drafted from the approved study map, grouped and placed at their visits.
  • Eligibility checklists, lab panels and standard scales from our forms library.
  • Checks and cross-form references, written in plain language.
  • A specification, OpenClinica 4 forms and CDISC ODM from any approved version.

Checked by code, not taken on trust.

Every draft Loom writes goes through a code check before anyone sees it. The hard rules are final. Loom gets a round or two to fix what the check found, and anything still wrong reaches you as a review item, marked as such.

Studies / S1505 / Forms / Checks

Draft checks

Before your team sees the draft

✓Each citation matches the protocol’s words
✓Each form fits the export format
✓Each cross-form reference has a value to read
11 for review
Illustrative checks, never the rule catalogue.

Boundaries

What Loom won’t do.

Browse the web.

Loom has no web access. It works from your protocol and your study.

Learn from your protocols.

We don’t train AI models on your protocols.

Approve anything.

It drafts and suggests. People decide, and sign.

Guess.

Where the evidence isn’t in the protocol, it asks.

Every run is recorded: the model, the prompt version, a fingerprint of what it read, and the draft it produced.

One flag, up close

When the protocol is unclear, Loom asks.

A flag says what’s unclear and shows the passages behind it. Your team decides, and the answer is recorded with your name, the time and, if you give one, your reason.

S1505 · Detailed description · Outline

Patients achieving stable disease or better undergo pancreatectomy 4-8 weeks after completion of first 3 courses of treatment.

S1505 · Outcome measures

Up to 8 months post registration (and within 4 to 8 weeks after the last dose of Cycle 3 preoperative chemotherapy).

Loom Example

Does the 4–8 week window before surgery start at the end of Cycle 3, or at its last dose?

Your choice is recorded with your name, the time and, if you give one, your reason.

[A real flag from the S1505 run replaces this example, quoted with its section and page.] Public protocol · S1505 · Example

Where you’ll meet Loom.

LoomDrafting the study map

Working in the background

While Loom works, its job shows each step; open it to see where it is.

Days since first dose
Loom suggested · Protocol

Marked where it helped

Anything Loom proposes is marked as Loom’s, with its source where it took it from the protocol.

Ask Loom to draft a check…

In the form designer

Ask for a check or a cross-form reference in plain language; you review the draft.

See Loom on a real protocol.