AKASA launches AI for autonomous mid-cycle coding

Healthcare technology company AKASA is pushing AI beyond the bounds of a revenue cycle assistant to a fully autonomous tool for clinical documentation and medical coding.

The company, known for its generative AI for hospital revenue cycle management, announced on Friday the launch of its "autonomous AI platform for the mid-cycle." The technology is designed to fully code inpatient cases with no human intervention in as little as 90 seconds after a patient is discharged, according to the announcement.

The new solution will add to AKASA's mid-cycle technology portfolio, which includes pre-bill optimization and clinical documentation improvement. The company currently serves health systems representing about 10% of U.S. inpatient discharges and over $180 billion in aggregate net patient revenue.

This is a "holy grail in our industry," said Malinka Walaliyadde, CEO and co-founder of AKASA, in a press release.

The middle revenue cycle or mid-cycle is where clinical care is translated into medical codes for billing and reimbursement, as well as public health tracking, population health management and clinical communications.

However, moving cases through this phase of the revenue cycle has been a challenge for healthcare providers, given the complexity of clinical documentation improvement, charge capture and pre-billing reviews in the mid-cycle. AKASA reports on its website that an inpatient encounter typically contains 60 documents and 50,000 words, indicating 150,000 possible codes.

Many providers also still use manual processes to complete mid-cycle functions, relying heavily on human coders amid a massive workforce shortage.

"The incredible demand for healthcare is finally being addressed by advancements in AI," Walaliyadde stated. "Multiple parts of the healthcare ecosystem will need to scale up, with documentation and coding being critical components."

With autonomous mid-cycle technology, AKASA seeks to expand workforce capacity, accelerate billing and improve quality performance.

The company reported that an evaluation of their AI's performance matched or exceeded that of human coders on key accuracy measures for the bulk of inpatient cases for health systems. Those measures included MS-DRG assignment, principal diagnosis, clinical quality capture and present-on-admission accuracy.

The technology also reduced the time to code an inpatient case to seconds, compared with the typical 30-60 minutes, although coding can take as long as three to four days, depending on coding workloads, according to AKASA.

The latest medical coding productivity benchmarks from MedCodex Health show an industry standard of 4-6 charts per day for certified coders working on complex inpatient cases. At high-acuity teaching hospitals, coding speed was as high as 3-4 charts per day.

AI tools for the mid-cycle have recently come under scrutiny from healthcare payers. A recent analysis from the Blue Cross Blue Shield Association linked AI coding and billing solutions to nearly $1 billion in additional costs over two years.

However, healthcare AI companies and their clients contend that the tools are reducing the administrative burden on hospitals while ensuring providers receive all the revenue owed to them. In fact, the technologies could ease friction in the medical billing process by submitting more complete, accurate claims in the first place, they say.

Adoption of AI coding and billing continues to grow as providers reap the benefits of automated mid-cycle functions. As the technology advances, autonomy is offering even greater advantages in terms of reimbursement speed, coding accuracy and workload capacity.

AKASA is already planning to extend its autonomous platform to outpatient facility encounters, a move that could further accelerate adoption as providers seek comprehensive mid-cycle automation.

Still, healthcare industry leaders warn about the potential risks of autonomous revenue cycle management tools, including large-scale errors and data security challenges.

For now, vendors like AKASA are threading the needle, deploying autonomous AI while keeping human experts in the loop to catch errors and ensure compliance. But it begs the question: Can revenue cycle management technology be truly autonomous, or will this hybrid model become a permanent safeguard?

Jacqueline LaPointe is an Executive Editor at Xtelligent Healthcare Media, covering revenue cycle management, healthcare payers, health policy, and health IT since 2016.