AI in the Crash Cart: Faster Checks, Less Manual Work
Crash cart checks are some of the most repetitive work on a hospital unit. Every drawer, every item, every lot number and every expiry date has to be found, read and recorded, on every cart, on a schedule. AI computer vision does not change what a good check looks like. It takes the slowest part of it, reading and typing small print, off the nurse’s hands.

Picture the monthly drawer check. A nurse breaks the seal and works through thirty-plus items: ampules, prefilled syringes, airway adapters, IV catheters. On each one she finds a lot number and an expiry date printed in type a few millimeters high, reads it, compares it to the list, writes it down or keys it in, and moves to the next. Then the next drawer. Then the next cart.
Multiply that by every cart in the hospital, every month, plus the daily checks in between, and it adds up to a large share of skilled nursing time spent on reading and data entry.
Where the time goes in a crash cart check
Break a check into its parts and most of it is the same few steps, repeated item by item. The part that needs a nurse’s expertise, deciding what to do about what she finds, is a small share of the total.
Reading: finding and reading small-print lot numbers and expiry dates on every item.
Recording: writing results on paper, or typing them into a system, often after the check is done.
Re-work: going back when something was missed, mistyped or put in the wrong drawer.
Chasing expiries: working out which items expire soon and making sure they are replaced in time.
None of these steps is hard. Together, they are what makes a check slow, and the time spent does not always buy a complete result.
15% of carts in one study still had missing, expired or unavailable items, even though they had already been checked
Top 5 ranking for expired or missing crash cart items among the clinical problems Joint Commission surveyors found in 2022 (outside infection control and suicide prevention)
Why barcodes alone have not solved it
The obvious answer is to scan. More and more drug packaging now carries a barcode with the lot and expiry built in. But a crash cart drawer is not a warehouse shelf. It holds a mix of items from different manufacturers, often taken out of their original boxes, and their labels are anything but consistent.
Everything in one code: Some barcodes hold the item, lot and expiry together. One scan and you are done.
Only the product: Some barcodes only say what the item is. The lot and expiry are still printed text that someone has to read.
No code at all: Some items have nothing to scan, just a printed label.

A scanner that only reads barcodes speeds up the first case and leaves staff typing in the other two. The slowest step, reading small print, stays manual.
How AI speeds up the check
This is where computer vision earns its place. A phone camera reads the barcode when there is one, and reads the printed label when there is not. Staff point the camera at each item and keep moving, with nothing to type.
Read. The camera captures the item, lot and expiry from the barcode or the printed label.
Check. The system compares what it reads with the drawer’s checklist and expiry dates, and flags anything that needs attention.
Decide. The nurse confirms quantities, replaces what needs replacing and signs off.

The nurse still makes every call, and everything the camera reads is shown on screen so it can be checked in seconds. What changes is that the reading and recording happen as the check is done, in one pass.
Let the camera do the reading. Give the nurse’s time back to nursing.
What this looks like in practice
This is the approach we took in IntraPosition Crash Cart Management. Staff scan the cart to open its checklist, then point a phone camera at each item. The system reads whatever the item offers, the barcode or the printed label, and records the item, lot and expiry straight into the cart’s audit.
Two simple safeguards cut out most of the re-work. An item in the wrong drawer is flagged before the audit is closed, and the same pack cannot be counted twice by accident.
What changes for the team
Faster checks. Each drawer is done in one pass, with nothing typed in afterwards.
Less re-work. Wrong-drawer items and double counts are caught during the check, not after it.
Expiries planned ahead. Expiry dates are captured automatically, so items that are due to expire can be replaced before they become a problem.
Complete records without extra admin. Every check is saved in the cart’s history, ready when a recall or a survey comes.
More time for patient care. Time that went into reading labels and typing goes back to clinical work.
Questions to ask before you choose a tool
Not every tool that says AI will save your team time. Before you sign anything, ask four simple questions:
Does it read the printed label when there is no barcode?
Does it show staff what it captured, so they can fix a mistake on the spot?
Does it catch items in the wrong drawer, and stop the same item being counted twice?
Is every check saved in the cart’s history, so you can answer a recall or a surveyor in minutes?
If the answer to any of these is no, you are buying a faster clipboard, not a faster check.
The most useful AI on a hospital unit may be the one nobody notices. It reads the label, and the nurse gets the time back.
Crash cart readiness will always depend on the people who check the carts. The aim is simply to make every check quicker, every record complete and every minute of nursing time count. See how computer vision audits work in IntraPosition Crash Cart Management, or read what it really takes to keep a crash cart ready.
Sources: Cart-readiness findings (missing, expired or unavailable items on 15% of checked carts) from "Periodic Resuscitation Cart Checks and Nurse Situational Awareness: An Observational Study," Journal of Emergency Nursing (2022). Expired or missing crash cart items as a top-5 clinical problem from The Joint Commission’s 2022 surveyor scoring trends (standard PC.02.01.11). 2D barcode requirements for drug packages under the U.S. Drug Supply Chain Security Act (DSCSA).



