Health tech company AKASA is pushing deeper into AI-powered revenue cycle workflows, launching an autonomous AI platform for inpatient medical coding and clinical documentation integrity.
AKASA, which provides generative AI solutions for hospital RCM, is expanding from AI-powered prebill review to target autonomous mid-cycle for healthcare’s most complex and resource-intensive workflows, according to executives.
The company provided Fierce Healthcare with a first look at the new autonomous AI platform for the mid-cycle.
Inpatient medical coding is highly complex process due to the depth of clinical documentation required, stringent regulatory rules and the financial impact on hospital reimbursement. The mid-cycle is where a patient’s clinical record is translated from documentation into codes that drive reimbursement, quality reporting, risk adjustment and the integrity of the patient record.
Unlike outpatient coding, which focuses on individual procedures, inpatient coding captures the entire patient stay from admission to discharge. The work also is resource-intensive, and still predominantly manual with many hospitals understaffed in this area.
Autonomous mid-cycle has long been considered the "holy grail" in healthcare due to its complexity, Malinka Walaliyadde, CEO and co-founder of AKASA, told Fierce Healthcare.
"The typical inpatient stay consists of about 60 documents, 50,000 words," he noted. Medical inpatient coders have to take those documents and convert that patient encounter into a set of medical codes chosen from a set of 150,000 codes, he noted.
Typically, a coder takes 30 to 60 minutes to code an inpatient encounter. Coding workforce shortages and capacity constraints can leave health systems waiting several days after discharge for a coder to even start working on an account.
"You're going to have this perfect storm where the volume of care delivered is going to increase, the complexity of care delivered is going to increase, so the work is going to get harder, more of it is going to happen, and there's just not going to be enough people to do it," Walaliyadde said in an interview. "Fortunately, we're now in a moment where AI has advanced enough that it can actually deliver against this exact need."
AKASA’s AI can complete coding in approximately 90 seconds after the patient is discharged, which can speed up billing process and reduce accounts receivable days. The company asserts that its autonomous mid-cycle AI is designed to fully code highly complex inpatient cases across all specialties with no human intervention.
The company conducted rigorous third-party, blinded evaluations comparing AI performance against medical coder performance. In these studies, the company tested inpatient encounters representing 65% to 85% of all inpatient volume for health systems and demonstrated that its AI performance matched or exceeded that of human coders on key accuracy measures. These measures included MS-DRG assignment, principal diagnosis, clinical quality capture and present-on-admission accuracy.
"There's a number of evaluation mechanisms that we're working on with health system partners on how do you continue monitoring post-go-live," Walaliyadde said. Following both internal testing and evaluation with a health system alpha partner, AKASA plans to make the autonomous AI platform for inpatient medical coding available in the next several months.
AKASA's vision is that autonomous capabilities can take on a significant portion of the workload during periods of high demand, enabling experienced staff to concentrate on the most complex and specialized tasks.
The company customizes and fine-tunes the AI for individual health systems, allowing its models to account for differences in patient populations, clinical criteria, documentation practices, and care complexity, Walaliyadde noted.
"We found that using generic approaches does not work well for very complex activities. We actually build a custom model for each health system that we work with. Every health system we work with gets their own model, and then that powers a number of different applications at that health system. That approach, we found, enables us to solve much harder problems than other people are able to solve," he said.
Beyond medical coding, AKASA extends these capabilities upstream with CDI, helping ensure documentation completeness and creating a unified AI layer across clinical documentation, coding, and prebill review, according to executives.
AKASA also plans to launch outpatient facility encounters to the platform as well.
The company is seeing rapid expansion of its AI products across its health system customer base. It reported that inpatient volume processed by the company has grown almost 6x in the last year. AKASA’s customers represent more than $180 billion in aggregate net patient revenue and roughly 10% of the country’s inpatient discharges.
Cleveland Clinic is using AKASA's AI-powered prebill review products across coding and CDI and plans to deploy the autonomous mid-cycle solutions to improve the accuracy and completeness of its clinical records and enhance operational efficiency.
“Our revenue cycle work is especially time-intensive because we care for many medically complex patients,” Rohit Chandra, Ph.D., chief digital officer at Cleveland Clinic, said in a statement to Fierce Healthcare. “With autonomous coding, we seek to improve speed and precision in these challenging processes under a compliance-first approach to this work.”
Julie Yoo, general partner at Andreessen Horowitz (a16z), a primary backer and investor in AKASA, assets that the company is differentiated by its fully autonomous, large language model-based approach to medical coding, rather than relying on legacy automation tools or human-in-the-loop workflows.
AKASA's autonomous AI inpatient coding solution is "truly first in class," she told Fierce Healthcare.
She also asserts that AKASA "leans into the hardest things" before other companies. "You'll see the majority of the startup players in this market tend to start on the outpatient side, where it's a much simpler case mix and lower stakes use cases. Whereas AKASA has solely focused on the inpatient setting, where there's a much higher bar for clinical complexity and accuracy," she said.
AKASA takes a clinical approach to revenue cycle, Yoo said. "What I've learned from AKASA along the way is you actually need to have a tremendous amount of clinical rigor in the underlying technology and your overall approach. AKASA has taken an almost academic-level of rigor in terms of publishing their research, publishing their findings, putting their product through rigorous validation studies as one would see in the clinical AI world," Yoo noted.
She added, "I think that approach is paying off in the form of market adoption and recognition from some of the most prominent health systems that in order to actually get revenue cycle right, you actually need to have a tremendous amount of clinical credibility as opposed to just viewing it as an admin set of tasks."
The company's expansion into mid-cycle automation builds on years of work helping health systems modernize revenue cycle operations.
“AKASA has been a pioneer in generative AI and in revenue cycle solutions,” said Jeff Francis, chief financial officer and vice president of finance at Nebraska Methodist Health System, in a statement. The health system has worked with AKASA since 2019.
“We’ve moved with them step by step, and it shows up in the metrics I care about: revenue integrity, denials, write-offs, and how quickly we get paid. Autonomy feels like the natural next step as health systems look for ways to increase capacity and make complex revenue cycle tasks more efficient.”
There is a growing tension between health systems and payers over the use of AI in medical billing. A recent claims analysis from the Blue Cross Blue Shield Association found that hospital adoption of AI coding and revenue cycle management tools drove $942 million in additional healthcare costs over a two-year period. BCBSA is the national trade association and lobbying organization for the federation of independent, locally operated Blue Cross and Blue Shield companies.
It's critical that health systems get full credit for the care they deliver, Walaliyadde said. To do that, organizations are turning to technology to navigate the healthcare industry's highly complex billing system.
Yoo asserts that AI-powered RCM solutions like AKASA could help reduce friction between payers and providers as AI can help close information gaps that drive administrative burden and reimbursement disputes. "Companies like AKASA are positioned right at that strategic node of sitting at that interface and potentially being able to leapfrog, from a first principles perspective, all the friction that currently exists around utilization management and prior auth in a way that operates as a clinical source of truth," she said.
"So much of the friction is a lack of trust and information asymmetry between the payer and the provider, but AKASA has the opportunity to become that source of truth layer for clinical decisions that are made in a way that could eliminate a lot of that overhead," she said.