Revenue cycle management performance has never been more important. And recent advances in technology, especially artificial intelligence, offer great potential for healthcare management functions.
Jay Aslam, co-founder and chief data scientist at CodaMetrix, said RCM capabilities can lay the foundation for leveraging technology to help hospitals and health systems improve performance. Aslam was part of the team that developed the original medical coding AI system at Massachusetts General Brigham in 2016, and he is an insider's expert on the role AI is playing in increasing his RCM's impact today. Have a point of view.
We spoke to Aslam, who has more than 30 years of experience developing AI, machine learning, and natural language processing technologies, to learn more about his efforts at MassGeneral Brigham AI, which spun off to become CodaMetrix, and how generative AI is working with RCM. We spoke to him about his views on the role he can play. , and what he thinks the next 5 to 10 years will look like in his AI medicine.
Q. You helped create Mass General Brigham's original medical coding AI, which spawned the spinoff CodaMetrix, which is your company today. Tell us about your approach to AI at Mass General Brigham, how it works, and how the spin-off happened.
A. The origins of CodaMetrix in 2019 date back to 10 years ago, in 2009, when I was contracted as a consultant to a company (VOBA Solutions) working with the Massachusetts General Physicians Organization (MGPO), now part of Massachusetts General Brigham. It started when. VOBA developed custom systems and performed system integration for Mass General's various revenue cycle functions, including medical coding.
As is true in most healthcare systems, the burden of medical coding often falls on the physicians themselves (e.g., CPT or procedure codes) and/or the specialized medical coders (often ICD or diagnostic codes). MGPO is particularly keen to reduce the coding burden for physicians as well as improve the efficiency of professional medical coding staff.
VOBA and MGPO knew they had a wealth of data to make their systems “intelligent,” but they didn't have the expertise to do so.
I was hired as a consultant due to my expertise in AI, natural language processing, machine learning, and statistics, as well as the fact that I have worked with VOBA members in the past.
To reduce the coding burden for physicians, we have narrowed down the world of CPT codes to just a handful of codes that physicians may need to consider when faced with the task of medical coding. We started by building an AI-based system that could.
Essentially, historical claims data tells us that, for example, knee and shoulder surgeons who perform surgeries according to a particular schedule description will have a high probability of performing one or more of a handful of surgeries. You can know that. His list of CPTs and their descriptions corresponding to the most likely surgeries that surgeons can use as a starting point for their coding efforts.
The AI-based system continuously learns over time and, given enough data, it learns to tailor the results to a specific surgeon (in this example), potentially allowing the physician to review them. was able to significantly limit the space of the highest code. This has significantly reduced the burden on doctors facing medical coding tasks. This system was introduced in his 2010 at General Brigham, Mass., and has been in use and continually learned ever since.
That system relied on the physician knowing what procedure he or she performed to ultimately select the appropriate CPT code; Efficiency has been improved by providing physicians with a starting point and pertinent information. .
If we instead relied on clinical notes, predicting codes directly from the clinical notes themselves could completely eliminate physician involvement in CPT coding and professional medical coder involvement in CPT and ICD coding.
Such AI-based systems must learn the patterns of words and phrases in clinical notes that correspond to a particular CPT or ICD code, along with a myriad of different coding rules dictated by different governing bodies and payers. there is.
Additionally, if an AI-based system can accurately self-assess its confidence in those predictions, it can perform autonomous medical coding, where such cases are warranted based on the AI's self-assessed confidence. can send cases directly to billing without human intervention. Submit the remaining cases along with the AI's predictions for human review while maintaining the specified level of accuracy.
We developed just such a system and implemented it at Massachusetts General Brigham in 2015. Since then, the system has been operating successfully and is continuously learning. We automate medical coding, reducing the burden on physicians and increasing the efficiency of Massachusetts General Brigham's professional coding staff.
Given the success of this internally developed and deployed system, Army General Brigham ultimately decided to explore the feasibility of this technology in the larger medical market. Once it was determined that this technology could be used and useful outside of the scope of Commander-in-Chief Brigham, the decision was made to spin out a company dedicated to developing and deploying this technology for the larger medical industry. Ta. Thus, CodaMetrix was born in his 2019 year.
Q. You are currently working hard on incorporating generative AI into the management functions of revenue cycle management. Please describe your vision.
A. Our vision is to increase efficiency and reduce costs in the U.S. health care system. To reduce the burden on doctors and medical coders. And it delivers autonomous medical coding with the precision and clinical specificity needed for reward-based care, value-based care, population health, and more. Let's discuss each in turn.
First, although estimates vary, administrative and revenue cycle functions account for approximately 20% to 25% of U.S. healthcare spending, money that could be spent on patient care instead, and medical coding is a is the most expensive element. Our vision is to apply AI to increase efficiency and reduce costs in the U.S. healthcare system, starting with autonomous medical coding.
But these same AI technologies can generate insights and solutions far beyond autonomous medical coding. Analysis of these technologies and their results will optimize the routing of cases requiring manual review to the most appropriate medical coder, identify opportunities for clinical documentation improvement, and provide a pathway to payer-approved coding algorithms and automated adjudication. It can also be used to open. Automated pre-authorization and more all drive efficiency and cost savings for the healthcare industry.
Second, our goal is to leverage AI to reduce the burden on physicians and enable professional medical programmers to operate with the highest licenses. Regarding the former, let's start with two anecdotes. My father was a practicing physician until he retired over ten years ago. I remember when I was a kid in the 1970s, his father would call home and I would sometimes tag along. Because my father has the time to do so and he can provide that level of care.
However, by the time my father retired from private practice, he was spending hours every day completing paperwork for reimbursements, pre-approvals, and more. And he wasn't the only one under this ever-increasing burden on doctors. This increases the amount of time spent with patients, leading to physician burnout.
Second, I have a relative who recently attended a radiology residency and internship program at one of the most prestigious medical institutions in the United States. He told me a story about how the medical residents would draw straws every week to decide who would do the medical coding. I was responsible for all radiology cases that week, allowing others to focus their time on radiology studies.
Our vision is to leverage AI to reduce the burden on physicians and empower them to learn and practice their skills.
Even for professional medical coders whose job it is to perform medical coding, medical coding tasks can be tedious. Routine cases such as chest X-rays or mammograms with no findings do not require the advanced skills of professional medical coders. Our goal is to automate all such cases and more, enabling these professionals to perform at the highest level.
Finally, medical coding is the language used to abstract and describe reimbursement and subsequent patient encounters. Currently, in fee-for-service use cases, medical coding only needs to meet a lower “medical necessity” standard, and clinically comprehensive coding is not warranted and is often undesirable.
However, there is a critical need for more accurate and comprehensive coding in value-based care, population health, clinical trials, longitudinal analysis, and more, and our vision is to leverage AI to reach that level. It's about delivering coding accurately and efficiently.
Q. What do you think will happen in the next 5-10 years in the medical field with artificial intelligence, machine learning, and natural language processing?
A. First, a general comment. In the future, the AI ​​revolution will be much like the smartphone revolution in the sense that AI will be a universal and essential tool to improve our daily lives, but at the same time we must learn to use it wisely. will be seen as such.
Think about your smartphone and consider how much of your daily life, mostly for the better and sometimes for the worse, revolves around this essential device. That's what AI is, and it will become universal and essential. It's up to us to take advantage of the benefits while minimizing costs.
In healthcare, autonomous medical coding is just one application of AI. And while just a few years ago, autonomous medical coding was considered the preserve of large academic medical centers that could afford to experiment with cutting-edge technology, it is rapidly becoming an essential part of every healthcare system. It is seen as a tool. Similarly, the original smartphone was once considered cutting-edge technology for early users, but quickly became an essential tool for everyone.
AI will be in every aspect of healthcare, including diagnosis, treatment planning, drug discovery, and design – virtually everything. The combination of vast amounts of data, computational resources, and modern AI algorithms has enabled rapid improvements in all of these areas, and we are seeing such improvements today.
And my parting words and vision for the future are that AI will not completely replace human effort, but rather enhance it, and that human-involved AI-augmented systems will be better than AI or humans alone. This means that you can achieve the desired results. AI is a powerful tool that can be used by, for, and with humans to increase efficiency and achieve performance in the healthcare industry.
Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email: bsiwicki@himss.org
Healthcare IT News is a publication of HIMSS Media.