Assessing Suturing Technique with the AI Assistant – Theator

You can ask the AI Assistant to assess the suturing technique on a case, for example by typing Assess the suturing or Rate the suturing. The assistant looks at sampled footage from the case's annotated suturing and anastomosis segments and returns a suturing report card. The card covers seven needle-handling checks based on EASE (End-to-end Assessment of Suturing Expertise), and each check shows the visual cue it was read from. This article explains how to ask, how to read the card, and what it can't tell you.

How to Ask

Open the case and open the AI Assistant.
Type Assess the suturing and send. Questions about the suturing technique over the whole case work too, such as "Rate the suturing", "How was the needle handling?" or "Evaluate the anastomosis technique".
A placeholder shows while the assistant reviews the footage, then the report card appears in the chat.

Ask about the case as a whole. If your question names a specific time in the video, the assistant describes that moment instead of assessing the suturing. If you get a description of a moment instead of a report card, ask again with Assess the suturing.

You can also ask from the assistant when you're not on the case page. Name the case so the assistant can find it first. See Finding a Specific Case with the AI Assistant.

On some cases, the assistant may suggest Assess the suturing for you. You don't need to wait for the suggestion: you can type the question on any case you can view.

What the Assistant Looks At

Annotated segments only. It uses the case's segments annotated as suturing or an anastomosis, plus closures that are all needle work, such as a vaginal vault closure, and steps marked as hand-sewn. Steps marked as fully stapled are skipped.
Up to three segments. If the case has more, it takes the longest ones. A shorter segment that sits entirely inside a longer one, such as a suturing step inside an anastomosis step, is skipped so the same footage isn't sampled twice. That means the segments assessed aren't always the three longest on the case.
At least 8 seconds long. Shorter segments are skipped because they are too brief to show a complete needle pass.
Sampled, not continuous. Inside each segment it looks at sampled footage, not every second.
You can't pick the time range. The segments come from the case's annotations. To ask about a specific moment instead, see What the AI Assistant Can and Can't Do.

Reading the Report Card

The card is titled Suturing technique — observations from sampled footage. Under the title, a line tells you how much could be judged, for example 5 of 7 sub-skills could be judged, read from 2 sampled windows. The assistant's reply sums up the coverage; the per-check detail is on the card.

The seven checks are grouped as follows:

Group	Check	Counted as ideal when	Flagged NEEDS ATTENTION when
Repositioning · read from video	Number of needle repositions	The needle is grasped at most twice before the drive begins	Three or more grasps or repositions, fumbling, or a dropped needle
Needle hold · read from frames	Needle hold depth	Grasped between halfway and three-quarters of the way from the needle tip	Grasped at the tip, at the swage, or on the suture thread
	Needle hold ratio	The needle sits in the distal quarter of the driver jaws	The needle sits deep in the jaws near the joint
	Needle hold angle	Roughly perpendicular: 90–135 degrees between needle and driver	A flat or acute grip well outside 90–135 degrees
Driving and withdrawal · read from video	Driving smoothness	One continuous supinating pass with at most one re-adjustment	Multiple pushes, stalls, tissue tearing, or repeated re-entries
	Driving wrist rotation	The needle follows its own arc through tissue, driven by rotation	The needle is pushed or dragged straight against its curve
	Withdrawal wrist rotation	A complete rotating withdrawal that follows the needle arc	A straight pull that levers or tears the exit point

Each row shows the check's name, its result, the visual cue the result was read from and, when available, Window looked at with the times.

What the Results Mean

ideal technique observed (green): the visible cue matched the ideal description for that check.
NEEDS ATTENTION (orange): the visible cue matched the needs-attention description. It's an observation about the sampled footage, not a grade.
not assessable from the sampled footage (grey): the check wasn't visible in the sample. It is not a finding about your technique.

Thin Evidence

If only one check could be judged, the card shows a Thin evidence note. In that case the reading isn't a picture of overall technique.

When None of the Checks Could Be Judged

Sometimes the footage is readable but none of the seven checks can be judged from it. Then the card is titled Suturing technique — could not be judged and shows only the sampled windows, with no check rows. You can still play the windows to see the footage yourself. This is a limit of what the recording showed, not a finding about the technique.

Sampled Windows

Under the checks, Sampled windows shows the segments that were sampled as playable clips, labeled with the annotation name and times. Play them to see the footage yourself.

The card has no overall score and no pass/fail result. That is deliberate.

What It Is Not

Not a validated skill score. The card says so: Observations from sampled footage, graded in code. Not a validated skill score, and not a substitute for a formal assessment.
Not a formal evaluation. Asking for a suturing assessment doesn't create or submit an evaluation, and doesn't add anything to a scorecard. To get a formal evaluation from your attending, see Theator for GME.
Not the whole case. Only sampled footage from up to three annotated segments is reviewed.

When You Don't Get a Report Card

What the assistant tells you	What it means	What to do
No suturing is annotated on this case	No suturing or anastomosis segment is annotated. This can also mean the AI annotations missed the suturing	Check the case timeline on the Case page
The suturing was too short to assess	Suturing is annotated, but every segment is shorter than 8 seconds	Nothing to retry; there isn't enough footage to assess
The camera was outside the patient	The sampled segments were recorded outside the patient	Asking again won't change the result
It couldn't be assessed right now	A temporary problem loading the footage, or the assistant couldn't confirm which parts of the recording are inside the patient	Try again later
The case isn't available	The case doesn't exist, or you don't have permission to view it	You can assess only cases you can view

Giving Feedback

If a reading looks wrong, rate the assistant's answer. See How to Share Feedback in Theator.

What's Next?

What the AI Assistant Can and Can't Do — Where the assistant's limits are
Rating Your Entrustability (EPA) with the AI Assistant — Self-assess against your program's EPA rubric
Theator for GME — Request a formal evaluation from your attending
Finding a Specific Case with the AI Assistant — Point the assistant at the right case
Case page Overview — Find a case's annotated steps on the timeline