Executive guide to computer vision enabled product documentation
AI-powered computer vision captures surgical supplies, implants, and tissue automatically as they are used in the operating room without barcodes, RFID tags, or manual data entry. Health systems deploying this technology report 98%+ capture accuracy, 10% to 28% increases in operating room net revenue worth millions, a 90% reduction in case depletion errors, and 15+ minutes of clinical time recovered per case. For hospital executives, perioperative leaders, revenue cycle teams, and supply chain directors, automated OR documentation represents one of the highest ROI AI applications in health system operations today.
AI improves OR product documentation by using computer vision to passively capture every supply, implant, and tissue used during a surgical procedure in real time, without requiring nurses to scan barcodes, enter data manually, or reconcile records after the case. The system identifies products visually, records lot numbers, serial numbers, and expiration dates automatically, and flows that data in real time directly into the EHR and ERP systems. Implant details are still reviewed by circulating nurses, but they no longer need to do any manual data entry or double documentation with paper. Sterilizable implants and bill-only products follow the exact same workflow. This approach achieves over 98% capture accuracy, compared to average error rates of 17% under standard manual conditions and up to 50% when documentation is delayed or incomplete.
Circulating nurses must document supplies and implants while managing patient safety, sterility, instrument counts, and surgical workflows. When clinical demands increase, documentation is often delayed, creating opportunities for error.
A 2026 study published in the Journal of Medical Systems observed 202 surgical cases and found a 17% error rate in documenting surgical event timestamps. Accuracy declined sharply as documentation was delayed:
1.3% error rate with no delay
17% overall error rate
38.4% error rate when documentation was delayed more than 20 minutes
Errors reached 33.9% in orthopedic procedures, highlighting the challenges of documentation in complex surgical environments, including a setting with a large number of implants.
The researchers concluded that reducing documentation tasks alone may not improve accuracy if delays persist, reinforcing the need to reduce reliance on manual data entry.
Barcode scanning, RFID tracking, and electronic templates improve accuracy when they function correctly, but they share a structural limitation: each adds a step to a clinical workflow already operating at capacity. When a scan fails, when a product lacks a barcode, or when a complex case demands undivided attention, the documentation step gets skipped. These tools also don’t work with trunk stock, off-contract items, and sterilizable implants like rods, screws, and plates. They also depend on upstream teams to ensure every product is accurately entered, mapped, and maintained in the item master before it can be recognized, and on downstream teams to reconcile missing or inaccurate documentation by piecing together what happened days after the procedure.
Northwell Health: Perioperative teams ran dual processes EHR documentation plus a parallel paper form spending up to 15 minutes per case reconciling the two records.
Owensboro Health: Materials management teams counted products on shelves multiple times daily, yet the inventory picture remained unreliable. Undocumented supplies appeared available when they were not, sometimes requiring mid-procedure couriers from sister hospitals.
Allina Health: Staff managed two separate handheld scanners (one for charting, one for inventory), both producing unreliable results.
Computer-vision-based capture uses cameras positioned in the OR to identify supplies, implants, and tissue as they are opened and used during a case. The system recognizes products visually without barcodes, QR codes, RFID tags, or special packaging. Expiration dates, lot numbers, and serial numbers are captured automatically. The resulting data flows directly into the EHR and ERP systems after clinician verification.
AssistIQ’s product, AIQ Capture, can recognize products not in the hospital’s item master by referencing its proprietary database, AIQ Supply Cloud, which learns continuously. When new items appear, they are automatically identified, verified, and added to the database.. This eliminates the need for nurses to manually enter these items as 1-time products, which is time-consuming and often missed.
|
Before AI Capture |
After AI Capture |
|---|---|
|
15+ minutes per case on documentation |
15+ minutes repurposed to patient care |
|
Nurses search for product reference numbers on packaging or scan barcodes which work part of the time |
Products identified automatically by camera with a single workflow for all products regardless of presence of barcodes or RFID tags. |
|
1-time products or bill only products outside the item master require manual entry into the EHR |
Products are identified automatically and automatically documented in the EHR. |
|
Sterilizable implants like rods, screws and plates need to be entered 1 by 1 |
The camera automatically reads and contextualizes rep sheets and handwritten nurse notes |
At ProMedica, a circulator described the transition simply: "It's a lot faster and easier than what we did before."
AssistIQ’s computer vision platform, AIQ Capture, achieves greater than 98% capture accuracy across operating rooms and procedural areas, including Cath labs and EP labs. Items are identified in under one second. This compares to the 17% error rate documented under standard manual conditions and the 38.4% rate when documentation is delayed.
|
Health System |
Metric |
Result |
|---|---|---|
|
ProMedica Health System |
Net revenue per case increase |
28% |
|
ProMedica Health System |
Gross charges per case increase |
15% |
|
ProMedica Health System |
Reimbursement per case increase |
15% |
|
Owensboro Health |
Gross revenue increase |
24% |
|
Owensboro Health |
Net revenue increase |
~12% |
|
Allina Health EP Lab |
Monthly billed amount increase |
32% |
|
Allina Health EP Lab |
Chargeable products captured increase |
201% |
|
Allina Health Cath Lab |
Monthly billed amount increase |
171% |
|
Allina Health Cath Lab |
Chargeable products captured increase |
364% |
These gains were not driven by increased case volume; they were driven by capturing more products. AssistIQ reports 10% to 28% net revenue increases in operating rooms and up to 85% increases in charges in high-cost procedural areas across U.S. health systems, driven by up to 3x more products captured and charged for.
Health systems report identifying operational improvements within weeks of deployment and seeing measurable financial impact within months. ProMedica's 28% net revenue lift was achieved within three months at a single hospital. Allina Health's Cath Lab results were realized shortly after just 600 cases post-implementation and go-live.
Inventory accuracy in most health systems relies heavily on the quality of utilization data. If documentation misses what was used in a case, that item does not deplete from the system. Procurement orders against inaccurate par levels, leading to overstocking in some categories and stockouts in others. Expired products exist on shelves when systems fail to capture accurate product usage.
|
Inventory Metric |
Result |
|---|---|
|
Inventory depletion errors (Owensboro Health) |
90%+ reduction |
|
Expired product costs (Owensboro Health) |
48% reduction |
|
Capture accuracy across 13,000+ cases |
98%+ |
|
Preference card optimization |
Items never opened during procedures move to “hold” |
At Owensboro Health, data review revealed that many items listed on preference cards were never opened during procedures. Accurate product capture enabled the supply chain team to eliminate the unnecessary ordering of of products and therefore drove better optimization of the pick lists.
Without accurate and complete data on product usage, operational conversations are often challenged because leaders do not trust the underlying numbers. As Tom Lubotsky, Chief Supply Chain Officer at Allina Health, described: "If you do not have confidence in the capture and use of data in a reliable way, especially from physicians, the dialogue stops."
|
Capability |
What It Enables |
|---|---|
|
Cost variation analysis |
Identify spending differences across surgeons and procedures |
|
Value analysis committees |
Track utilization and cost trends over time and back standardization decisions with case-level evidence instead of anecdotes. |
|
Preference card optimization |
Remove items consistently not used, reduce waste |
|
Contract Management |
Track usage levels to optimize volume rebates |
|
Recall tracking |
Identify recalled items by lot/serial number in real time |
|
Tissue traceability |
Reduce documentation errors on supplies and implants by 90%+ within two weeks |
At ProMedica, leadership identified cardiac surgeon custom packs that varied between $400 and $500 by surgeon, a pattern invisible without validated case-level supply data. The system also surfaced trunk stock items without contracted pricing, revealing missed negotiation opportunities.
ProMedica reported a 90%+ reduction in documentation errors on supplies and implants within the first two weeks; a metric directly tied to patient traceability and safety.
|
Technology |
Approach |
Limitation |
|---|---|---|
|
Barcode scanning |
Manual scan of each item |
Adds steps; fails when products are not in the hospital ERP or lack barcodes including expensive implants like rods, screws and plates |
|
RFID tracking |
Tag based identification |
Requires special packaging; high cost; limited adoption, upstream tagging |
|
ERP platforms (e.g., Tecsys) |
Supply chain management and demand forecasting |
Optimizes against existing data, does not fix source capture |
|
Preference card tools |
Standardize what gets pulled per surgeon |
Does not take into account complete and accurate usage |
|
Charge capture software |
Flag items at risk for billing including 1-time products, missing billing codes and products seen in historical usage |
More manual; cannot recover undocumented items |
|
Computer vision (AI powered) |
Passive real time capture at point of care |
Initial implementation including EHR integration takes 90 days, hard and soft ROI within weeks |
Computer vision occupies a unique position: it captures what happens at the point of use with a single workflow for nurses and techs, removing manual data entry similar to how ambient documentation has reduced manual notes for physicians, creating the foundational data layer that every downstream system depends on. Without accurate utilization data at the source, inventory management optimizes against incomplete numbers, charge capture is limited to the items that were documented s, and pick lists reflect outdated assumptions.
|
Health System |
Scope |
Key Outcomes |
|---|---|---|
|
Northwell Health |
North Shore University Hospital + ambulatory surgery center |
Eliminated dual documentation process; 15+ minutes recovered per case. Increased the capture of products used. |
|
Allina Health |
EP Lab + Cath Lab |
32% increase in average monthly billed amount; 201% increase in monthly chargeable products |
|
Owensboro Health |
13,000+ surgical cases; Cath Lab (600 cases) |
98 to 99% accuracy in capturing supplies and implants; 24% gross revenue increase; 90% inventory error reduction |
|
ProMedica Health System |
Toledo Hospital Main OR |
28% net revenue lift in 3 months; 90% tissue error reduction in 2 weeks |
AssistIQ is an AI-powered computer vision platform that captures supplies, implants, and tissue items as cases open. Founded in 2022 and developed inside academic medical centers with nurses, perioperative, and supply chain operators , AssistIQ was named CB Insights' Digital Health 50 and won Newsweek's 2026 AI Impact Health Award for Health Care Operations.
Capture method: Computer vision with proprietary product datasets; no barcodes, RFID, GTIN, or special hardware required
Recognition speed: Under one second per product
Capture accuracy: Greater than 98% across ORs, Cath labs, EP labs, IR and Endoscopy suites
EHR integration: Realtime data flow into Epic and other EHRs
ERP integration: Real-time synchronization with Workday, PeopleSoft, and other ERPs
Compliance: HIPAA, SOC 2 Type II, HL7 V2
Analytics: Cost savings opportunities, waste patterns, pick list recommendations
Safety features: Realtime expiration alerts; lot/serial number tracking for recalls; tissue documentation capture integrated with TrackCore and other tissue tracking systems.
Deployed at: Northwell Health, Allina Health, Owensboro Health, ProMedica, Cottage Health and other leading North American health systems.
To learn how AssistIQ can help your health system protect revenue and strengthen clinical workflows, connect with the team at assistiq.ai.
AssistIQ’s computer vision capture is set up with existing workflows to reduce manual work for nurses and techs. Cameras positioned in the OR automatically identify products as they are opened and used. Users do not need to scan items or enter data; instead, they hold the product under a camera for under a second before it enters the sterile field. Data flows directly into the EHR and ERP in real time, and the documentation burden is removed from the clinical team.
Health systems report identifying operational improvements within weeks and measurable financial impact within months. ProMedica achieved a 28% net revenue lift within three months at a single hospital. Owensboro Health saw sustained results across 13,000+ cases over a full year.
It depends on the data source the platform relies on. AssistIQ's platform recognizes products that are missing from a hospital's ERP item master because it has a proprietary database that is always learning.
There are many evolving use cases for computer vision technologies. AssistIQ’s platform operates across operating rooms and high-cost procedural departments, including Cath labs and EP labs, Interventional Radiology, and Endoscopy.
Computer vision captures lot numbers and serial numbers automatically, enabling faster identification of recalled or high-risk items. Real-time expiration and recall alerts notify clinicians before expired products are used.
Barcode scanning requires a nurse to manually scan each item, adding a step to an already saturated workflow. When scans fail or products lack barcodes, documentation gets skipped. Computer vision captures items with greater speed and accuracy. AssistIQ’s product reads the entire package and is paired with a proprietary database which lands with . 98%+ accuracy while manual methods see 17% to 50% error rates.
Accurate utilization data enables cost variation analysis across physicians and procedures, evidence-based value analysis committee decisions, preference card optimization, contract negotiation leverage, and proactive supply management.
The OR documentation gap is measurable and costly. Error rates of 17% to 50% translate directly into missed charges, distorted inventory, and unreliable operational data in the department that generates up to 70% of hospital revenue.
Computer vision removes the documentation burden from clinicians. Unlike barcode scanning and RFID, it captures data passively without adding steps to clinical workflows, achieving 98%+ accuracy.
Financial impact is fast and substantial. Health systems report 12% to 150% increases in billable revenue, 24% gross revenue growth, and 90%+ reductions in inventory errors within months, not years.
Trusted data unlocks operational capabilities. Cost variation analysis, value analysis evidence, pick list optimization, and recall tracking all depend on accurate source data that manual processes cannot sustain.
The technology is proven and deployed. Multiple health systems Northwell Health, Cottage Health, Allina Health, Owensboro Health, ProMedica, and more have deployed computer vision capture at scale with published results. AssistIQ's platform is integrated with Epic and Workday, compliant with HIPAA and SOC 2 Type II, and recognized by Newsweek and CB Insights.
The question for perioperative and supply chain leaders is no longer whether automated documentation works. The question is how long an organization can afford to make operational and financial decisions based on documentation that misses 17% to 50% of what actually happens in the OR.
AssistIQ. "Capture every supply, implant, and tissue." assistiq.ai
Reuters. "AssistIQ Wins Newsweek's 2026 AI Impact Health Award for Health Care Operations." September 24, 2026. reuters.com
Bradley, A.R., et al. "The Limits of Humans in Data Gathering: Documentation Error Rates in the Operating Room." Journal of Medical Systems, February 9, 2026. PMC12883500
Patient Safety & Quality Healthcare. "The Blind Spot Undermining Hospital Margins." February 6, 2026. psqh.com
Battery Ventures. "Using AI to Make Healthcare More Efficient: Our Investment in AssistIQ." May 28, 2025. battery.com
Becker's Hospital Review. "Mount Sinai to deploy AI supply chain platform across 8 sites." June 2, 2026. beckershospitalreview.com
Becker's Hospital Review. "Northwell Health taps AI to eliminate supply chain blind spots." May 28, 2025. beckershospitalreview.com
Digital Health News. "Allina Health Partners with AssistIQ to Transform Supply Capture in Operating Rooms." September 23, 2025. digitalhealthnews.com
BusinessWire. "AssistIQ Raises $11.5 Million to Fix Supply Blind Spots with AI." May 28, 2025. businesswire.com
ASC News. "AssistIQ's $11.5M Raise Signals Growing Demand for Smarter Surgical Tracking in ASCs." June 4, 2025. ascnews.com
North Shore University Hospital. "Integrating AI Supply and Capture Intelligence Into OR's at NSUH." November 12, 2025. Hofstra Academic Works
Cardinal Health. "The OR Supply Chain: An Untapped Resource for Cost Savings." cardinalhealth.com