
Blog
3 Outcomes To Expect From Autonomous Inpatient Coding
A revenue cycle veteran on what autonomous inpatient coding should deliver for revenue integrity and what to ask before you buy.
The Gist
Health systems have spent years adding coders, contractors, and review tools to the mid-cycle, yet cash still sits in DNFB. Industry veteran Linda Schatz explains why the usual fixes fall short, the three outcomes leaders should expect from autonomous coding, and the questions she’d ask any vendor before signing.
I started my career as a nurse, and over the past 40+ years I’ve advised health systems across the country on improving their revenue cycle. In that time, I’ve watched the cost of the middle revenue cycle climb steadily. Staffing needs have grown across every area: coding, CDI, revenue integrity, and internal and external audits.
To keep up, our industry has continued to add resources: more coders, more outsourcers, more review layers, and more tools. Each one was meant to fill a gap and improve revenue capture and efficiency.
But when I talk with CFOs and other revenue cycle leaders today, the problems haven’t changed. Revenue capture is not only delayed, but the cost to collect keeps climbing. Earned revenue still leaves the building without being captured.
The reason is that almost every fix we’ve tried still relies on someone reading every encounter. But the complexity of these encounters is only growing.
Autonomous coding is the first approach that removes that dependency. If you’re evaluating it for inpatient, these are the three outcomes you should expect, and you should hold any vendor to them.
1. Revenue that moves faster
Every inpatient encounter waiting to be coded is revenue you’ve earned but can’t bill yet. And inpatient claims are your highest-dollar claims.
Here’s a quick way to size it: take the DNFB days you can attribute to coding and multiply by your average daily inpatient revenue. That number is working capital stuck in a queue.
Autonomous coding works on encounters as soon as the documentation is ready, without waiting for a coder to reach them. It may take one of your coders 3–4 days to get to an encounter. Autonomous coding can do it in seconds.
Claims go out sooner, DNFB and AR days come down, and cash arrives more predictably, even in months when you’re short-staffed.
What to ask vendors: How soon after discharge are encounters coded? What share of our volume can be coded autonomously?
2. A lower cost to collect, without adding headcount
Today, coding costs rise with volume. About half of inpatient coders are eligible for retirement, and experienced replacements are hard to find. So the usual answer has been contractors, overtime, offshoring of resources, and more review layers.
Those review layers cost more than most leaders realize. At one organization I led, my team accepted only 8% of the opportunities flagged by the tool. That means 92 out of every 100 reviews took up expert time and produced nothing but increased costs.
Autonomous coding takes on the routine volume, so your experienced coders and CDI specialists can spend their time on the complex cases that need their judgment. Your volume can grow without your costs growing at the same rate.
What to ask vendors: How does it fit with our current tools and workflow? How will our team’s time be redeployed?
3. Results that are accurate and defensible
The average inpatient encounter runs about 60 documents and 50,000 words, roughly the length of The Great Gatsby. Expecting a coder to read several of those a day without missing anything isn’t realistic.
When AKASA analyzed coding errors, most were one-off misses, not patterns that more training could fix. There’s just too much.
The goal isn’t more codes. It’s the right codes. A missed secondary diagnosis leaves earned revenue on the table. A code the documentation doesn’t support creates denial and compliance risk. And, even more importantly, it may impact patient care, transitions of care, and outcomes.
Our new autonomous inpatient solution reads the full record and codes only what the documentation supports. It catches what people miss without reaching for what isn’t there. And because every code links back to its evidence in the chart, you can see exactly why each one is on the claim. That matters more every year as payers look harder at inpatient claims.
What to ask vendors: Has accuracy been tested blind against expert coders? Is the model trained on our own documentation and coding practices? Can every code be traced back to the chart?
Getting there without putting accuracy at risk
None of this happens overnight, and it shouldn’t. The health systems I see doing this well follow the same path:
Start with a human in the loop on every encounter
Prove the accuracy on your own data
Expand autonomy as your confidence grows
In AKASA’s blinded head-to-head tests, autonomous coding matched or beat expert coders on key accuracy measures. That gives you a measured way to start, not a leap of faith.
The margin pressure isn’t going away. For health system leaders, the question is no longer whether to automate the mid-cycle, but how soon. The systems that start now will see benefits in accuracy, cash, cost, and capacity long before everyone else.
Learn more about autonomous inpatient coding

Linda Schatz
Linda Schatz, RN, BSN, CCDS, is the director of CDI at AKASA. With more than 40 years of experience in healthcare, and the last 15 focused on documentation and coding, Schatz has built a comprehensive career. She started her CDI career as a Medicare auditor and then dove into the CDI space, moving into consulting roles across multiple firms. Schatz has implemented CDI programs across the country, provided CDI education, and spent a large portion of her consulting career in provider education related to documentation integrity at the Advisory Board Company. She has also been the corporate director at a large health system, moving the organization to a centralized program focused on quality. In her current role at AKASA, Schatz acts as a CDI subject matter expert, working with clients and contributing to the development of generative AI-powered CDI tools. She works closely with machine learning teams to improve coding suggestions, manage and audit CDI staff within AKASA, and ensure quality and productivity standards are met.






