Artificial intelligence (AI) is increasingly being incorporated into ophthalmic ambulatory surgery center (ASC) coding, documentation review, and reimbursement workflows, offering new opportunities to improve efficiency. As these tools become more widely available, ASC leaders are increasingly evaluating how AI can support workflow optimization without introducing additional compliance risk. The question for most ASCs is no longer whether AI will be incorporated into coding workflows, but how to incorporate it responsibly.
ASC AI Oversight Checklist
The following checklist highlights key review points ASC teams should consider when incorporating AI into coding and claim submission workflows.
- Verify operative report is complete and finalized
- Confirm CPT codes accurately reflect procedures performed
- Check NCCI edits and bundling rules
- Validate modifier usage, diagnosis linkage, and code order
- Ensure diagnosis codes support medical necessity
- Review claim prior to submission
In the ASC setting, the margin for error is small. Coding inaccuracies, incomplete documentation, or incorrect modifier use can result in claim denials, delayed reimbursement, or audit exposure. AI in the ASC can directly affect how procedures are reported and reimbursed. As a result, its use must be carefully controlled and integrated into structured workflows that prioritize accuracy, validation, and oversight.
Accurate ophthalmic ASC coding already depends on a structured review process that includes documentation analysis, CPT selection, modifier application, diagnosis linkage, payer policy review, and claim validation. As outlined in prior ASC coding guidance, these review steps are essential to ensure compliant reimbursement and reduce audit risk. AI may assist with portions of this process, but it cannot independently replace the clinical judgment, coding interpretation, and oversight required throughout each stage of surgical coding and billing.
AI Across the ASC Surgical Workflow
AI tools are not confined to a single task. In most ASCs, they influence multiple points in the surgical process, including preoperative planning, intraoperative documentation, coding review, and claim submission.
In the preoperative stage, AI may assist with identifying procedures that require prior authorization, generating documentation checklists, or suggesting initial CPT codes. During documentation, AI may summarize operative reports or extract key details. In the postoperative phase, AI may suggest final CPT and ICD-10 codes, while at the billing stage it may assist with sequencing and claim preparation.
Understanding how AI interacts with each stage is essential. Errors introduced early in the workflow—such as incomplete documentation—can carry forward and affect final coding accuracy. A workflow-based approach ensures that validation occurs throughout the process rather than at a single endpoint.
Outputs should also be evaluated using the “ABCs” of AI: accuracy, bias, and context. AI-generated coding guidance may contain outdated or incorrect information, particularly when payer policies or coding rules have changed. In addition, AI systems are trained on broad data sets that may not fully reflect ophthalmic ASC workflows or payer-specific coding nuances. Most importantly, AI lacks full procedural and clinical context. A coding recommendation may appear technically correct while failing to accurately represent the surgical service performed, payer requirements, or documentation standards. Applying the ABCs during review can help identify potential coding and reimbursement errors before claim submission.
Human Oversight Remains Crucial
Successful AI integration still depends on human oversight throughout documentation, coding, and claim submission workflows. Although AI may assist with summarizing operative reports or suggesting CPT codes, these outputs must still be evaluated within the full clinical, procedural, and payer context. CPT descriptors, NCCI edits, modifier requirements, and payer-specific rules continue to require human interpretation to ensure accurate coding and compliant reimbursement.
Case Example #1: Dislocated IOL With PPV and IOL Exchange
A patient with a dislocated intraocular lens (IOL) secondary to zonular weakness undergoes pars plana vitrectomy (PPV) with removal of the dislocated lens and placement of a scleral-fixated IOL. After reviewing the operative report, AI suggests CPT 67121, CPT 67036-59, and CPT 66985 based on the documented procedural steps. Further review of the CPT descriptors and bundling edits is required to determine the most accurate coding.
CPT 67121 is bundled with CPT 67036 when performed as part of the same vitreoretinal procedure, and modifier -59 is not appropriate because the procedures were not distinct services. CPT 66985 also does not accurately describe the procedure because it applies to secondary IOL placement in an aphakic patient.
The operative report documents removal of a dislocated IOL with implantation of a new IOL during the same surgical session, which is more appropriately reported as an IOL exchange using CPT 66986. The PPV is separately reportable with CPT 67036. Because CPT 66986 carries a slightly higher relative value than the vitrectomy, it should be submitted as the primary procedure.
This example highlights how AI may identify individual procedural elements without fully recognizing procedural intent, bundling edits, or the distinction between secondary lens placement and IOL exchange. Accurate coding depends not only on the procedures documented, but also on understanding CPT definitions.
Case Example #2: Retinal Detachment
A patient presents with a retinal detachment with proliferative vitreoretinopathy (PVR) centrally and inferiorly. The operative report documents repair of the retinal detachment with PPV, endolaser photocoagulation, and gas tamponade.
After surgery, AI reviews the operative report and suggests CPT 67113 for complex retinal detachment repair. AI may also separately identify CPT 67036 for the vitrectomy or CPT 67105 for the endolaser photocoagulation based on the documented procedural steps. Further review of the operative report and full CPT descriptor is required to determine whether the coding accurately reflects the procedure performed.
Reading the full descriptor for CPT 67113 is critical, as it describes repair of complex retinal detachment “with vitrectomy and membrane peeling,” including, when performed, gas tamponade and endolaser photocoagulation.
In this case, the operative note documents vitrectomy, laser treatment, and gas tamponade, but it does not describe membrane peeling. Although PVR is present, which defines the complexity of the disease, the required membrane peel was not performed or documented. Because CPT 67113 requires both vitrectomy and membrane peeling, along with the specific diagnosis, the procedure does not meet the full CPT definition for complex retinal detachment repair.
In addition, CPT 67036 and CPT 67105 are not separately reportable because vitrectomy and endolaser photocoagulation are included components of CPT 67108 when performed as part of retinal detachment repair and are bundled under National Correct Coding Initiative (NCCI) edits. It is inappropriate to unbundle these services when performed in the same eye.
After review of the operative report, CPT descriptors, and included procedural components, the procedure is more appropriately reported with CPT 67108: Repair of retinal detachment; with vitrectomy, any method including, when performed, air or gas tamponade, focal endolaser photocoagulation, cryotherapy, drainage of subretinal fluid, scleral buckling, and/or removal of lens by same technique.
This example highlights the importance of accuracy and context in AI-assisted coding. AI may associate diagnoses such as proliferative vitreoretinopathy with higher-level CPT codes alone or identify individual procedural steps separately without fully recognizing how comprehensive retinal detachment repair codes are defined within CPT. Accurate coding depends not only on diagnosis terminology, but also on careful review by a well-trained coding expert to verify the complete CPT descriptor and the documented surgical procedure match.
Conclusion
AI offers meaningful opportunities to improve efficiency in ophthalmic ASC coding and revenue cycle workflows, but accurate reimbursement still depends on structured review and oversight. AI-generated recommendations must be evaluated within the full clinical, procedural, and payer context to ensure coding accurately reflects the services performed. ASCs that use AI most effectively will be those that integrate it strategically within established documentation, coding, and claim validation processes to improve efficiency while maintaining compliance and reimbursement integrity. OASC
