Healthcare has a vibrant startup and innovation ecosystem, but that doesn’t mean everyone shares the perks that come with technological developments. Historically, payors have often been ahead of the game in adopting and benefiting from new tech, forcing providers to play catch up.
But artificial intelligence (AI) is changing the game. A persistent trend I’ve witnessed is the steady rise of providers prioritizing technology – especially AI – to inform strategic priorities and address chronic headwinds such as staff shortages, increasing cost pressures, and slow reimbursement times, to name a few.
As healthcare leaders catch on to the enormous potential of AI to combat thorny issues, AI will take center stage next year, reshape the larger healthcare ecosystem, and begin to even the playing field between payor and provider.
As the end of the year approaches, here’s how I see this playing out in 2024:
Autonomous medical coding will be widespread — if not the norm.
The latest health IT report from Bain & Company and KLAS Research highlights the increasing importance of software and technology. Per the report, 70% of providers think AI will have a more significant impact on their organizations this year vs. last year, and an impressive 56% of those surveyed view software and technology as one of their top three strategic priorities, with revenue cycle management (RCM) coming in at a resounding first place. With many health systems focused on reducing administrative burdens for clinicians and a continued shortage of medical coders, autonomous coding adoption will surge.
Large language models (LLMs) like ChatGPT won’t work as advertised.
There’s plenty of commotion about the capabilities of language models, but they will likely disappoint when functioning as the core of autonomous coding engines. However, they will be enormously valuable in solving smaller pieces and edge cases, pushing coding automation rates to 100% for all the high-volume outpatient specialties.
Autonomous coding enjoys a high level of trust among healthcare finance professionals who use or plan to use the technology, with 45 percent indicating it often works well and 16 percent placing complete trust in it. Yet despite its emergence as a powerful tool for streamlining and improving error-prone manual coding processes, autonomous coding suffers from an awareness problem, with 52 percent saying they do not know what it is.
Those are the findings of a new survey from the Healthcare Financial Management Association (HFMA) on behalf of AGS Health, a leading provider of tech-enabled revenue cycle management (RCM) solutions and strategic growth partner to healthcare providers across the U.S. More than 450 healthcare finance professionals were surveyed during the 2023 HFMA Annual Conference on their knowledge of and value expectations for autonomous coding, including 60 percent that use or plan to use autonomous coding.
More than half (52%) of respondents said they don’t know what autonomous coding is and 30 percent either did not or were unsure if it could be trusted.
Thomas Thatapudi
“Despite high expectations around its potential to increase coder productivity and coding accuracy, reduction in denials, missed charges and low-risk scores, and accelerated provider decision-making, autonomous coding suffers from a knowledge gap that must be closed if we are to see broader adoption,” said Thomas Thatapudi, CIO of AGS Health. “Until we can fully educate finance leadership on the potential autonomous coding holds for improving the healthcare revenue cycle, we are unlikely to see an acceleration in use cases for AI-powered technology which includes autonomous coding.”
Among the key benefits of autonomous coding is its ability to eliminate the potential for human errors that result in missed reimbursement opportunities, backlogs, delays, and claims errors, and its ability to push accuracy levels to near perfect percentages. All of which can be achieved in near real time with the right integration pipelines. Autonomous coding is also faster than its human counterparts – it can complete charts in seconds – yet it also understands what it does not know, flagging it for human review.
