The emergence of autonomous medical coding solutions that could eliminate the need for care providers to do their own coding or human coders to review their codes or do the coding on their behalf can turn out to be a turning point in the digital revolution of healthcare. While digital revolution of healthcare sounds grand and exciting, and some of us actually believed it was around the corner 15 years ago, However, this has proven to be quite challenging and we ended up learning that the legacy healthcare information systems, rigid workflows, and limited budgets of the buyers means that the progress would be slow and reserved only for those solutions that can navigate these barriers. So, why is it that autonomous coding (and autonomous clinical documentation) can be the turning point?
Part of the reason is that so far many of the benefits seen in other industries from the digital technologies, internet, software, data, analytics, and more have not been observed in healthcare yet. Electronic health records did not lead to simplification and time- savings. Quite the opposite! They made doctors into data entry clerks and kept their eyes focused on their computer screen while visiting with patients. When will we then see automation of what has been done manually historically? when will the convenience and time-savings show up in healthcare? When will a digital technology remove steps from a process, rather than adding more to it? This is not to say that there has been no benefits from the digitization of health data. In fact, you can argue that while there has not been a sea change in how things are done in healthcare, what has been happening in the background may be setting the stage for some breakthrough use cases soon.
The reason this could be a turning point is that we are finally seeing applications that will automate somethings end to end. And, they are being tested in the real-world environment and showing the desired results. The new generation of ambient intelligence technologies are automating documentation of clinical encounters to a level that Kaiser Permanente recently announced that they are expanding the use Abridge to most of their California locations. This is a good sign that the new generations of these technologies are meeting their moment and showing solid results. Just a short couple of years ago when I was writing about this topic in my book, AI Doctor: The Rise of Artificial Intelligence in Healthcare, I wrote that ambient scribe was still work-in-progress and certain companies and use cases were showing some promise but the results where not quite there. I wrote about how Saykara optimized their technology for Orthopedics and they automated certain aspects of the workflows for the orthopedists. Now, with the emergence of the LLMs and the huge leap in NLP, we may finally be at a point where mass adoption will be a matter of time. The result will be better doctor-patient relationships and more effective providers.
Autonomous coding appears to be arriving at the same inflection point. There is now evidence of its high accuracy and direct to billing capabilities in the real world. I’ve recently spoken with several medical centers that are starting to use autonomous coding technology and they talked about their satisfaction with what they’re seeing. Companies like Maverick Medical, Nym, and Codametrix are blazing the path in this area. They are using deep learning technology and NLP to train models on specialty- and domain-specific data to take over the coding tasks from the care providers and human coders. It appears that not only they show great accuracy right out of the gate, they also keep improving and can reach accuracy rates of over 90%. The benefits are not limited to eliminating the work of finding the right codes and doing the coding manually. Well-trained autonomous coding can be less subjective than humans and provide more consistency. It can potentially catch issues with the documentation required to support the codes real-time and alert the care provider to fix those issues. This eliminates the need for re-work and doing addendums to the original note. Also, with greater consistency, there’s reduction in denials by insurance companies. Both of these benefits have been documented in the early rollout of these solutions.
In the final entry in this series, I’ll discuss how the autonomous coding solutions do what they do and what may be one of their key benefits that will drive their rapid adoption.
