Tech & Innovation in Healthcare

Remember: CAC-suggested codes are only recommendations.

With the sudden growth of medical coding supported by artificial intelligence (AI), it’s possible to confuse AI with computer-assisted coding (CAC). While the two are similar in their abilities, the overall capabilities are very dissimilar.

Tech & Innovation in Healthcare breaks down the difference between CAC and AI-assisted coding and shows you how CAC software can benefit your coding practice.

Separate CAC From AI-Assisted Coding

CAC software looks at the clinician-provided documentation and automatically generates a list of medical codes for professionals to review, validate, and use. “When we use the coding assist software, it actually generates codes that are based on some specific terms or phrases and that maps directly to those specific codes and it uses the natural language process (NLP), so it’s very similar to spell check,” said Christine Hall, CHC, CPC, CPB, CPMA, CRC, CPC-I, senior consultant and certified instructor of Stirling Global Solutions, during AAPC’s DOCUCON 2022 session, “CAC and Impact of Clinical Documentation.”

This software differs from AI-assisted coding in that the AI software uses machine learning (ML) and NLP to offer optimal accuracy when automatically selecting codes. The AI software can take coding guidelines into account, and the software learns from interactions with users to continually improve its code selections, too.

While CAC software also uses NLP, which can learn from the coder over time, the software is unable to make calculated decisions using the guidelines to make an accurate code recommendation, nor can it properly choose principal diagnosis or procedure codes.

Example: A provider sees an established patient for their moderate persistent asthma. However, in the encounter notes, the physician only notes the patient was seen for their “asthma.” The CAC software could analyze the encounter note and recommend the use of J45.909 (Unspecified asthma, uncomplicated), when in reality, a code from J45.4- (Moderate persistent asthma) would be assigned.

“What’s the differential between unspecified asthma or persistent asthma? And with persistent asthma, is it mild, moderate, or severe? If the providers aren’t reporting those specific words, then we could have codes that are mapped to unspecified codes because the software can’t pull any other information,” Hall added.

Note: Check out Tech & Innovation in Healthcare, Volume 3, Issue 4 to learn more about AI-assisted coding in “Add AI to Your Revenue Cycle to Proactively Prevent Denials.”

Learn Where CAC Benefits Your Practice

Implementing CAC into your coding workflows provides several benefits, including boosting your coders’ productivity. You can save valuable time by using the CAC software to find, review, and verify clinical electronic health record (EHR) documentation. You can also use the CAC software to make edits that may get flagged because of information in the documentation to improve code accuracy. As a result, your practice will file claims in a timely manner, which could result in better reimbursement turnaround time.

At the same time, the software may reduce backlogs and lessen staff burnout by tackling a lot of the tedious tasks that eat up valuable minutes and hours in the workday.

However, CAC software, like other developing technologies, requires human interaction to improve its accuracy. While the software can help increase productivity, the software’s limitation can be a hindrance while human coders work out the issues and “teach” the software.

“[Computer-assisted coding programs] are great databases and storehouses of information related to coded guidelines and RVU calculations, but they’re limited to the predetermined point-and-click user inputs. We consider this to be real-time analysis of documentation, but the dependence is still on human intelligence and analysis,” said Jessica Miller, CPC, MHA, of MediCodio, during AAPC’s HEALTHCON 2023 session, “Successful AI Integration for Healthcare.”

Realize That Coder Jobs Are Safe

Regardless of whether medical coding and billing practices implement AI or CAC software, remember that these technologies are tools to assist coders with their workflows — not to replace the human element. “CAC software is only a tool, and it assists us to be more efficient,” Hall said.

The CAC software’s results are merely coding recommendations based on the information provided by the provider’s documentation. These recommendations will still require careful review of a professional coder to ensure the codes are accurate for the patient, adhere to the official coding guidelines, and comply with payer policies. As the technology evolves, human review is pertinent to its success.

“To improve the functionality of the CAC codes, it’s essential that we are involved in reviewing and providing that additional information, so that we can make it the best possible tool that we can,” Hall added.