Inside Angle
From 3M Health Information Systems
Debunking myths in AI medical coding automation
There are many pressures that health care revenue cycle departments face today, including recruiting enough qualified medical coding staff, complying with ever-changing regulatory requirements and coming up with productivity gains in a “do more with less” environment. Escalating demands to streamline coding and billing often leads revenue cycle departments to think about coding automation and leveraging artificial intelligence (AI). However, there are misconceptions about AI technology when it comes to its application in health care automation, so let’s dispel some popular myths.
Myth #1: AI will make medical decisions instead of a physician
Reality: AI should assist and expedite decisions, but should not have the last word
AI should not have the last word in health care decisions. AI solutions can automate simple tasks to free up time for physicians to make informed medical decisions, but it should not act independently. In the same way, for coders, AI solutions can automate simpler medical coding tasks, which then frees up time for coders to focus on the more complex cases.
Myth #2: AI will replace medical coders. Bots are coming for your jobs!
Reality: Coder’s roles will evolve, but their unique skills remain essential
Smart automation is upon us, but a medical coder’s expertise is very much needed for moderate to complex visits and oversight of AI results. Coder expertise, as well as having guardrails in place, can provide a built-in “safety net” for smart automation systems. This will ensure those visits that do not qualify for full automation can be easily moved to a semi-autonomous or traditional computer-assisted coding (CAC) workflow for coders to finalize.
AI and automation in coding will continue to evolve. The introduction of CAC decades ago helped automate workflows where it matters the most. As coding teams get even closer to a fully autonomous coding experience, it’s important to consider the realities of how far technology can take us, where do we need humans to step in and what can be safely and compliantly automated.
See the full infographic (PDF, 99 KB) to debunk more AI myths.
David R Bacon is the director of revenue cycle innovation for 3M Health Information Systems.
One thought on “Debunking myths in AI medical coding automation ”
Leave a reply
David Bacon
Business manager
David R Bacon is the director of revenue cycle innovation for 3M Health Information Systems. In this role, David’s current specific focus is on autonomous coding. He joined 3M as…
I learned more than 30 years ago about computer systems and data analytics that the first rule is GIGO, garbage in, garbage out. Computers need good data or the output is corrupt. In any form of coding, computer-assisted or not, there must be adequate data in the form of documentation to support the assignment of codes. For a variety of reasons there is generally a lack of thoroughness and specificity in provider documentation. That exists with inpatients, outpatients and HCCs. It is improving, slowly. Until the documentation/data improves greatly, coders have no need to worry. As stated in the article, humans need to validate the output much of which, we find, is not valid. there will be a long transition.