Revenue cycle management performance has never been more important. And recent technological developments, especially artificial intelligence, offer great potential for healthcare administrative functions.
The RCM function could lay the foundation for using technology to contribute to better performance of hospitals and healthcare systems, says Jay Aslam, co-founder and chief data scientist at CodaMetrix. Aslam was part of the team that developed Massachusetts General Brigham’s original AI medical coding system in 2016 and has an insider’s perspective on the role AI plays in driving impact in RCM today.
We interviewed Aslam, who has more than 30 years of experience developing AI, machine learning and natural language processing technologies, to talk about his AI efforts at Mass General Brigham that evolved into CodaMetrix, and his views on the role that generative AI can play in RCM. , and what he thinks the next five to ten years will look like in AI healthcare.
Q. You helped create Mass General Brigham’s original AI for medical coding, which created the spinoff CodaMetrix, your company today. Please share the story of your AI efforts at Mass General Brigham, what the AI does, and how the spinoff came about.
A. The origins of founding CodaMetrix in 2019 began ten years earlier, in 2009, when I signed on as a consultant with a company (VOBA Solutions) that partnered with the Massachusetts General Physicians Organization (MGPO), part of what is now Mass General Brigham . VOBA developed custom systems and performed systems integration for several revenue cycle functions at Mass General, including medical coding.
As is the case with most healthcare systems, the burden of medical coding often falls on the physicians themselves (for example, for the CPT or procedure codes) and/or professional medical coders (often for the ICD or diagnosis codes). mainly want to reduce the burden of coding for physicians, but also improve the efficiency of their professional medical coding staff.
VOBA and the MGPO knew they had a wealth of data to make their systems ‘intelligent’, but they did not have the expertise to do so.
I was hired as a consultant because of my expertise in AI, natural language processing, machine learning and statistics, and given that I had worked with a VOBA member in the past.
To ease the coding burden on physicians, we set out to build an AI-based system that could reduce the universe of CPT codes to just a handful of likely codes that a physician must consider when faced with a medical coding task.
Essentially, we could learn from historical billing data that, for example, a knee and shoulder surgeon performing an operation with a particular scheduling description would have a high probability of performing one or more of just a handful of procedures – and we would list the CPTs that match with the most likely procedures, together with their descriptions, that the surgeon can use as a starting point in his coding efforts.
The AI-based system continuously learned over time, and with enough data it could learn to tailor the results to a particular surgeon (in our example), expanding the space of the most likely codes that a doctor must assess is enormously limited. This significantly reduced the burden on physicians when faced with medical coding tasks. This system was deployed at Mass General Brigham in 2010 and has been in use ever since – continuously learning.
In that system, we trusted that the physician—who knew which procedure(s) he was performing—would ultimately choose the correct CPT code, but we gained efficiency by giving the physician a good starting point and the right information to make this task easy. to be carried out. .
If we were to rely on the clinical note instead, we could potentially completely eliminate the involvement of physicians in CPT coding and/or professional medical coders for CPT and ICD coding by predicting codes directly from the clinical note itself.
Such an AI-based system would need to learn the patterns of words and phrases in a clinical note that correspond to a particular CPT or ICD code, along with the numerous and varying coding rules dictated by various governing bodies and payers.
Furthermore, if the AI-based system could accurately self-assess its confidence in those predictions, it could perform autonomous medical coding – sending cases directly to the bill without human intervention when such cases, based on the AI’s self-assessed confidence , are guaranteed. a certain level of accuracy, while the remaining cases, along with the AI’s predictions, are sent for human review.
We developed just such a system and deployed it in 2015 at Mass General Brigham, where it has been running successfully and continuously learning ever since – automating medical coding, easing the burden on physicians and increasing the efficiency of the professional coding staff from Mass General Brigham.
Given the success of this internally developed and implemented system, Mass General Brigham ultimately decided to explore the feasibility of this technology in the broader healthcare market. Once it was determined that this technology could be used and useful beyond the boundaries of Mass General Brigham, the decision was made to form a company dedicated to developing and deploying this technology to the broader healthcare industry. This is how CodaMetrix was born in 2019.
Q. Today, you’re big on integrating generative AI into the administrative functions of revenue cycle management. Please describe your vision.
A. Our vision is to increase efficiency and reduce costs in the American healthcare system; to reduce the burden on physicians and medical coders; and to provide autonomous medical coding with the accuracy and clinical specificity needed for fee-for-service care, value-based care, public health and beyond. Let me describe each one in turn.
First, estimates vary, but administrative and revenue cycle functions account for about 20-25% of US healthcare spending – dollars that could instead be spent on patient care – and medical coding is the most expensive part of the revenue cycle. Our vision is to apply AI to increase efficiency and reduce costs in the US healthcare system, starting with autonomous medical coding.
But those same AI techniques can deliver insights and solutions that go far beyond autonomous medical coding; These techniques and the analysis of their results can also be used to optimize the routing of cases requiring manual review to the most appropriate medical coders, identify opportunities for clinical documentation improvement, and pave the way for payer certified coding algorithms, automated review, automated pre-authorization and more – all driving efficiency and reducing costs in healthcare.
Second, our goal is to leverage AI to reduce the burden on physicians and enable professional medical coders to operate at the top of their licensure. Regarding the first, I would like to start with two anecdotes. My father was a practicing physician until his retirement about twelve years ago. I remember my father making home visits as a child in the 1970s – and I would come along occasionally – because he had the time and could provide that level of care.
But by the time my father retired from private practice, he was spending many hours every day dealing with the paperwork required for reimbursement, pre-authorization, and the like—and he wasn’t the only one exposed to this ever-increasing physician burden that spending time with patients causes burnout among doctors.
Second, I have a family member who recently completed a radiology residency and internship program at one of the most prestigious medical institutions in the US. He told me the story of how the residents drew straws every week to see who would do the medical coding for all the radiology cases that week, while the others could spend their time learning radiology.
Our vision is to use AI to reduce the burden on physicians and enable physicians to learn and practice their profession.
Even for professional medical coders whose job is to perform medical coding, the medical coding task can be tedious. Routine cases such as chest x-rays or screening mammograms without findings do not require the considerable skills that professional medical programmers have learned. Our goal is to automate all such cases – and more – so that these professionals can operate at the highest level.
Finally, medical coding is the language used to abstract and describe patient encounters, reimbursement, and more. Currently, in a reimbursement use case, medical coding only needs to meet a lower standard for “medical necessity,” where clinically comprehensive coding is unwarranted and often undesirable.
However, for value-based care, public health, clinical trials, longitudinal analyzes and more, there is a great need for much more accurate and comprehensive coding, and our vision is to leverage AI to provide that level of coding accurately and efficiently.
Q. What will the next five to 10 years look like in healthcare in the areas of artificial intelligence, machine learning and natural language processing?
A. First a general comment. I think that the AI revolution in the future will be a lot like the smartphone revolution, in the sense that AI will be seen as a universal and indispensable tool that improves our daily lives, but that we must learn to use it wisely.
Think about your smartphone and think about how much of your daily life – usually for the better, but sometimes for the worse – revolves around this indispensable device. AI will be like this – both universal and indispensable – and it is up to us to learn to harness its benefits while minimizing costs.
Within healthcare, autonomous medical coding is just one application of AI. And while just a handful of years ago autonomous medical coding was seen as the province of large academic medical centers that could afford to experiment with cutting-edge technology, it is quickly being seen as a necessary and indispensable tool required by all healthcare systems – in much the same way that the original smartphones were once seen as cutting-edge technology for early adopters, but quickly became indispensable tools for everyone.
AI will be like this for all aspects of healthcare, including diagnostics, treatment planning, drug discovery and design – virtually everything. The combination of massive amounts of data, computing resources and the latest AI algorithms will enable rapid improvements in all these areas, and we are seeing such improvements today.
And my parting comment and vision for the future is that AI will not completely replace human effort, but rather augment humans, and that through humans-in-the-loop, AI-augmented systems can achieve better results than AI or humans alone . AI is a powerful tool that can and will be used by, for and alongside people in healthcare to increase efficiency and achieve performance.
Follow Bill’s HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.