The billing department may be where the problem appears. But it may have started long before the claim was ever submitted.
“Healthcare delivery inherently requires a human touch.”
— Dr.
John Whyte, CEO, American Medical Association
A patient. A couch. A physical therapist. And a lesson
that may completely change how physicians think about medical billing.
It started with a couch
Ericka Akoto lives in Hopewell, Virginia.
Her home is her safe place.
Her 2-year-old Silky Terrier, Hazel, helps with that.
But in March, Ericka had to leave home and go to VCU
Medical Center in Richmond after fluid began building up in her leg and
foot because of congestive heart failure.
She had already been through hospitalizations before.
She knew the routine.
Hospital bed.
Monitors.
Questions.
Rounds.
Waiting.
Discharge.
And, as she described it, she had previously gone home while
still struggling with symptoms.
Then VCU Health offered her something different.
Hospital at Home.
Instead of bringing Ericka into a hospital and trying to
make her fit into the hospital's environment, the care team brought
hospital-level care into hers.
And something interesting happened.
The clinicians could see things they might never have seen
inside a hospital.
Including a problem with her favorite couch.
Ericka had experienced two strokes, and moving around
her home had become more difficult.
A physical therapist noticed that she struggled to get up
from the couch.
The solution wasn't a $500,000 medical device.
It wasn't generative AI.
It wasn't blockchain.
It wasn't a new digital health platform with 47 features and
a dashboard nobody opens.
It was a cushion and handles near the armrest.
Simple.
Practical.
Human.
And Ericka said something that should make every healthcare
executive stop scrolling:
“It just made me, as a patient, feel valued and that my
healthcare was truly important.”
That sentence is bigger than hospital-at-home care.
It is a lesson about the entire healthcare system.
And it has an uncomfortable implication for medical billing.
The most important information about a patient isn't
always in the chart.
Sometimes it's on the couch.
And sometimes the information determining whether your
practice gets paid isn't in the billing system either.
It is upstream.
Here's my contrarian take
Most medical practices don't have a billing problem.
They have a visibility problem.
And then they hire someone to work harder inside the billing
problem.
That's different.
A claim gets denied.
Someone investigates it.
Someone calls the payer.
Someone opens a portal.
Someone sends records.
Someone appeals.
Someone waits.
Someone follows up.
Someone sends another fax.
Eventually, someone gets paid.
Everyone celebrates.
Until the same denial happens again.
Congratulations.
You successfully repaired the symptom.
The disease is still there.
We have built an industry around fixing yesterday
Think about how absurd this is.
A patient sees a physician today.
The encounter creates information.
That information becomes documentation.
Documentation becomes coding.
Coding becomes a claim.
The claim reaches the payer.
The payer rejects it.
And weeks later, somebody discovers that something
was missing.
Then we call the billing department.
Why?
Because the billing department is where the problem became
visible.
But that doesn't mean the billing department created it.
This is one of the biggest mistakes in revenue cycle
management.
We confuse the location where a problem is discovered
with the location where it was created.
Those are not necessarily the same place.
Ericka's couch is actually a data-quality story
Stay with me.
The couch matters because the physical therapist saw
something that a traditional clinical encounter might have missed.
The patient could describe the problem.
But seeing the problem was different.
Observation created context.
And context changed the intervention.
This is precisely what happens in revenue cycle management.
A billing system might tell you:
Claim denied.
That's observation.
Useful.
But incomplete.
The real question is:
Why?
Maybe eligibility was wrong.
Maybe authorization was missing.
Maybe the diagnosis didn't support the service.
Maybe documentation was incomplete.
Maybe the wrong modifier was used.
Maybe the payer's policy changed.
Maybe the information was entered incorrectly three steps
earlier.
The denial is the symptom.
The workflow is the environment.
The $10 billion question
Here's the question I would ask every physician-owner:
How much revenue are you losing because your practice
discovers errors after the point where they were cheapest to fix?
I don't mean one dramatic billing mistake.
I mean thousands of tiny leaks.
An eligibility error here.
A missing authorization there.
A documentation mismatch.
A coding inconsistency.
A claim submitted late.
A payer-specific rule nobody noticed.
A denial that gets appealed instead of prevented.
None of these individually looks catastrophic.
Together?
They can become a serious operating problem.
And the irony is that practices often respond by hiring more
people to manually chase the consequences.
That's like putting another person at the bottom of a leaky
boat.
Useful?
Maybe.
But eventually someone should probably look for the hole.
The healthcare industry's favorite phrase: "That's
just how it works"
Physicians hear it.
Practice managers hear it.
Billers hear it.
Patients hear it.
“That's just how insurance works.”
“That's just how prior authorization works.”
“That's just how the payer portal works.”
“That's just how the EHR works.”
“That's just how billing works.”
I have a problem with that phrase.
Because sometimes “that's just how it works” really means
“we've stopped questioning the workflow.”
The physician's hidden second shift
There is another problem.
Physicians aren't only practicing medicine anymore.
They are increasingly becoming unpaid operations staff.
The physician sees the patient.
Then documents the encounter.
Then responds to messages.
Then reviews results.
Then handles prior authorization.
Then deals with a coding question.
Then answers the billing team's question.
Then goes home.
And opens the laptop.
Again.
The phrase “pajama time” has become almost normal in
healthcare.
That's not normal.
It's just familiar.
There is a difference.
Here's where I disagree with conventional RCM thinking
The traditional revenue-cycle conversation often sounds like
this:
How do we collect more?
My question is different:
Why did we create so much work to collect it in the first
place?
That's not semantics.
It's architecture.
If the workflow produces bad information, the billing
department becomes a cleanup operation.
If the workflow produces good information, billing becomes
much more predictable.
The goal should not be to build the world's most efficient
cleanup crew.
The goal should be to create less mess.
Expert #1: Dr. Julia Breton
Dr. Julia Breton, co-medical director of VCU Health
Hospital at Home, describes an important difference between hospital-based care
and home-based care.
At home, clinicians can see how patients actually live.
They can see medication organization.
They can see mobility challenges.
They can involve family.
They can understand the environment.
Breton describes the goal beautifully:
The job is not to make the home more like a hospital. It
is to make the hospital more like home.
That principle has implications for technology.
Healthcare software shouldn't force physicians to behave
like data-entry clerks.
It should adapt to clinical workflows.
The technology should work around the physician.
Not the physician around the technology.
Expert #2: Chris Walker, R.N.
Chris Walker, R.N., a VCU Hospital at Home nurse,
describes another advantage.
In a hospital, clinicians can be pulled in multiple
directions.
At home, he can focus on one patient.
That sounds simple.
But it reveals a powerful operational principle:
Attention is a resource.
The same is true in a medical practice.
If your staff spends hours manually checking claims,
portals, eligibility, spreadsheets, and denial queues, those people have less
attention available for higher-value work.
Automation isn't really about eliminating humans.
It's about deciding where human attention is worth
spending.
That is a much more useful definition of AI.
Expert #3: The patient herself
The third expert isn't a CEO.
Isn't a consultant.
Isn't a technology founder.
It's Ericka Akoto.
Her lesson is the most important one.
She said that being treated at home allowed clinicians to
see what she dealt with every day.
That changed the care she received.
The implication for healthcare technology is profound:
Patients don't experience healthcare as a series of
databases.
They experience it as life.
A couch.
A medication bottle.
A worried spouse.
A difficult staircase.
A confusing bill.
A phone call nobody returned.
A physician who listened.
A physician who didn't.
Healthcare technology that ignores this context can be
technically sophisticated and still clinically stupid.
What physician-owned practices should steal from this
story
Not the Hospital at Home model.
The design philosophy.
See the environment.
Find the signal.
Understand the context.
Fix the problem where it starts.
Don't wait for the failure report.
That is exactly how I think about revenue cycle.
The denial is not the problem
This may be the most important sentence in the article:
A denial is an outcome, not a root cause.
Yet we often manage denials as though they are the disease.
The claim appears.
The denial appears.
The biller works it.
Done.
But what happens next?
Another claim.
Same payer.
Same service.
Same problem.
Another denial.
Another work queue.
Another phone call.
Another afternoon.
Eventually somebody says:
“Why do we keep getting these?”
Exactly.
That's the question that should have been asked first.
Denial management vs. denial prevention
There is nothing wrong with denial management.
You need it.
Claims will fail.
Payers will make mistakes.
Patients will change insurance.
Rules will be misunderstood.
Technology will fail.
Humans will make mistakes.
But if your entire RCM strategy is built around repairing
denials, you're operating downstream.
A better model is:
Detect → Understand → Prevent → Monitor
Instead of:
Deny → Work → Appeal → Wait → Repeat
One is a learning system.
The other is a hamster wheel with a clearinghouse login.
The five signals I'd watch first
If I owned a physician practice today, I'd start here.
1. Eligibility exceptions
How many claims are affected because coverage wasn't
properly verified?
Don't just measure the number.
Find the pattern.
2. Authorization failures
Which procedures, payers, physicians, or locations generate
the most authorization problems?
If the same pattern repeats, you don't have an employee
problem.
You have a workflow problem.
3. Documentation-related denials
Are physicians repeatedly being asked for the same missing
information?
If yes, ask why the workflow doesn't surface that
requirement earlier.
4. First-pass claim performance
The first submission tells you something.
A claim that succeeds immediately is operationally different
from one that requires three touches.
Track the difference.
5. Denial concentration
If 70% of your denials come from a small number of causes,
stop treating 100 denial codes as 100 separate problems.
Find the few causes creating the majority of the pain.
Don't build another dashboard
I can already hear someone saying:
“Great. We'll build a dashboard.”
No.
Please don't.
Healthcare has enough dashboards.
We have dashboards looking at dashboards.
The real question isn't:
Can we see the problem?
It's:
Can we do something about it before it becomes a problem?
A dashboard tells you the house is on fire.
An intelligent workflow should ideally notice smoke.
AI's biggest opportunity in RCM isn't writing emails
Generative AI can write emails.
Wonderful.
It can summarize documents.
Great.
It can draft an appeal letter.
Useful.
But those aren't necessarily the highest-value applications.
The bigger opportunity is pattern recognition across
fragmented workflows.
Imagine an AI system noticing:
“This payer has rejected 18 similar claims over the past 30
days. The common factor is a missing documentation element. Most originated
from two physicians and one scheduling workflow.”
That's useful.
Now imagine it identifies the issue before submission.
That's better.
Now imagine it automatically alerts the appropriate team.
That's even better.
Now imagine the system learns whether the intervention
worked.
Now we're getting somewhere.
The AI test I would use
Don't ask:
“Does your platform use AI?”
Almost everybody says yes.
Ask:
“What decision does the AI improve?”
Then ask:
“What happens differently because of it?”
Then:
“Can you measure the result?”
If the answer is vague, you may be looking at AI decoration.
And healthcare does not need more decoration.
My own failure lesson
I have spent enough time around healthcare technology to
know that building a technically impressive product is not the same as solving
a meaningful problem.
Healthcare founders love features.
We love integrations.
We love architecture.
We love saying “AI-powered.”
We love the demo.
But physicians don't wake up thinking:
“I hope someone gives me another dashboard today.”
They wake up thinking:
“I have 37 patients, six messages, two prior
authorizations, a full clinic, and somehow I still need to finish yesterday's
notes.”
That is the product problem.
Not the demo.
This is why I founded OnnX
My thesis behind OnnX is deliberately simple.
Medical billing should not require physicians to manage
an ecosystem of disconnected middlemen, portals, spreadsheets, phone calls, and
manual work queues.
For small and medium-sized physician-owned clinics,
administrative complexity can become disproportionately expensive.
The answer isn't necessarily another employee.
And it isn't necessarily another piece of software.
The answer is a more intelligent operating layer.
One that connects the signals.
Finds the exceptions.
Automates repetitive work.
Surfaces root causes.
And helps prevent avoidable problems.
The goal is not to make physicians better billers.
The goal is to make billing less of the physician's
problem.
A practical experiment for your practice
Don't buy anything.
Don't call a vendor.
Don't launch an AI project.
Do this first.
Take your top 20 recent denials.
Put them on a table.
For each one, ask:
Where was the problem created?
Not:
“Who fixed it?”
Ask:
Where did it begin?
Then categorize the answer:
Scheduling.
Registration.
Eligibility.
Authorization.
Clinical documentation.
Coding.
Claim creation.
Payer processing.
Appeal.
Patient responsibility.
You may discover something uncomfortable.
The billing department may be fixing problems it never
created.
That's valuable information.
Then ask the $64,000 question
Not literally $64,000.
Unless that's what you're losing.
Ask:
What would happen if we prevented half of these problems
before they reached billing?
Calculate:
Labor saved.
Revenue accelerated.
Appeals avoided.
Patient calls avoided.
Physician interruptions avoided.
A/R reduced.
Staff capacity recovered.
Now you have a business case.
Not a technology case.
The statistics tell part of the story
Current healthcare trends reinforce this larger shift.
VCU Health says its Hospital at Home program has cared for more
than 1,000 people since launching in 2023. Patients receive virtual
physician visits, in-person visits, and 24/7 virtual nursing support.
The broader hospital-at-home model is also expanding. The
American Hospital Association reports that hundreds of hospitals across dozens
of health systems and states have received approval to provide hospital-level
care at home, while CMS has found generally positive outcomes and patient
experiences in its evaluation.
At the same time, healthcare demand is shifting toward
outpatient and home-based care. The AHA's 2026 forecast projects 20% growth
in outpatient volumes by 2036, while post-acute care is projected to grow 31%.
The direction is clear.
Healthcare is becoming more distributed.
Which means information becomes more distributed too.
That makes interoperability, context, and workflow
intelligence more important—not less.
And here's the uncomfortable part
The more healthcare moves outside the hospital, the less
useful a healthcare architecture becomes if it assumes the hospital is the
center of everything.
The same applies to billing.
If your RCM architecture assumes the billing department is
the center of the revenue cycle, you're already looking backward.
The center is the patient journey.
Billing is one consequence of that journey.
The new RCM model
I believe we should think about revenue cycle differently.
Old model
Patient.
↓
Encounter.
↓
Claim.
↓
Denial.
↓
Biller.
↓
Appeal.
↓
Payment.
Better model
Patient.
↓
Signal.
↓
Validation.
↓
Clinical documentation.
↓
Intelligent claim preparation.
↓
Payer response.
↓
Continuous learning.
↓
Prevention.
The difference is subtle.
But strategically enormous.
Myth Buster
Myth: “More billers means fewer billing problems.”
Sometimes.
But more people can also mean more handoffs.
Headcount is not the same thing as capacity.
Myth: “Our EHR already contains all the information.”
Technically?
Maybe.
Operationally?
That's a different question.
Information trapped inside a system isn't necessarily usable
information.
Myth: “AI will solve our denials.”
Not automatically.
If the workflow is broken, AI can simply help you process
the broken workflow faster.
AI needs good architecture.
Myth: “Physicians don't care about revenue cycle.”
Physicians may not want to spend their evenings thinking
about claims.
That's different.
Physician owners absolutely care about whether their
practice can remain financially healthy.
Myth: “The billing department owns the revenue cycle.”
No.
The revenue cycle crosses the entire organization.
Billing is where the financial consequences become visible.
The legal and compliance reality
There is a serious side to all this.
Automation should never become an excuse to manufacture
documentation, manipulate coding, or optimize reimbursement at the expense of
clinical truth.
AI should support legitimate healthcare operations.
It should not invent medical facts.
It should not create false justification.
It should not turn a reimbursement goal into a clinical
decision.
And practices need to understand how vendors handle
protected health information.
HHS notes that functions such as billing, claims
processing, practice management, and data analysis may involve
business-associate obligations under HIPAA when performed on behalf of covered
entities.
Before implementing technology, practices should ask about:
HIPAA
Business Associate Agreements
PHI access
Audit logs
Encryption
Data retention
Subcontractors
AI training and data use
Human oversight
Data portability
Incident response
The cheapest vendor is not necessarily the cheapest
decision.
The ethical question
Here's the ethical line I would draw:
Use technology to reduce friction around legitimate care.
Don't use it to manipulate the meaning of care.
That distinction matters.
We should optimize:
Workflow.
Accuracy.
Speed.
Transparency.
Patient experience.
Administrative burden.
We should never optimize the clinical story to fit the
reimbursement.
What healthcare founders should learn from Ericka
If you're building healthcare technology, go watch
healthcare happen.
Not a conference.
Not a pitch deck.
Not a demo.
Go where the patient is.
Watch the nurse.
Watch the physician.
Watch the scheduler.
Watch the biller.
Watch what happens after the patient leaves.
You will probably discover that the workflow in your product
roadmap isn't the workflow in real life.
That's not a failure.
That's research.
The couch taught VCU something.
The billing queue can teach you something too.
What physician leaders should learn
Don't ask only:
“How much did we collect?”
Ask:
“What made collecting it difficult?”
That question changes the conversation.
It moves you from finance to operations.
From operations to workflow.
From workflow to root cause.
And from root cause to prevention.
That is where the leverage lives.
A 30-day RCM challenge
Week 1: Observe
Map the entire revenue cycle.
Don't change anything.
Just observe.
Week 2: Find
Identify the five most expensive recurring failure points.
Week 3: Fix
Choose one.
Fix the upstream workflow.
Not the downstream symptom.
Week 4: Measure
Compare:
Denial rate.
First-pass acceptance.
A/R.
Staff touches.
Appeals.
Administrative hours.
Then ask:
Did we actually make the system better?
The metrics I would put on the wall
Not 50 metrics.
Five.
First-pass claim rate
Denial rate
Top denial cause
Days in A/R
Administrative hours per 100 encounters
And one more:
Physician hours spent on billing-related work.
Because that number has a human cost.
Future outlook: the practice becomes an intelligent
system
The next generation of physician-owned practices will not
necessarily be defined by size.
A small practice with excellent technology and clean
workflows can potentially operate with far less administrative friction than a
larger organization built on disconnected systems.
The future practice may look something like this:
The patient schedules.
Eligibility is checked.
Potential authorization issues are identified.
The encounter occurs.
Relevant documentation requirements are surfaced without
interrupting clinical reasoning.
The claim is prepared.
Exceptions are routed to humans.
Routine work is automated.
Payer responses are analyzed.
Recurring problems are detected.
The system learns.
The practice improves.
The physician gets more time back.
That is what useful healthcare AI should feel like.
Not magical.
Just less annoying.
And honestly, that would be a pretty impressive healthcare
breakthrough.
The bigger lesson from Ericka Akoto
Ericka didn't need her healthcare team to know everything.
She needed them to notice something important.
That's the difference.
Healthcare has no shortage of information.
What it lacks is connected attention.
The same is true of revenue cycle.
We don't necessarily need another report telling us that
claims are being denied.
We need to understand why.
We need to connect the dots.
And we need to act before the failure becomes expensive.
That's the opportunity.
Final Thoughts: Stop Fixing What You Could Prevent
Ericka Akoto's favorite couch had nothing to do with medical
billing.
That's precisely why I like this story.
It reminds us that healthcare isn't a collection of
transactions.
It's a collection of human experiences.
The physical therapist didn't need another database.
They needed to see.
The nurse didn't need another dashboard.
They needed time.
The patient didn't need another workflow.
She needed someone to understand her reality.
And physicians don't need another billing chore.
They need a system that understands the signals created by
their work.
The best denial is the one that never happens.
The best administrative task is the one the system
quietly removes.
And the best healthcare technology is the technology that
gives clinicians more room to be clinicians.
Get Involved
So here's the question I want to leave with you:
If your practice could permanently eliminate one
administrative headache tomorrow, what would you choose?
Eligibility?
Prior authorization?
Documentation?
Coding?
Denials?
A/R?
Payer calls?
Or the endless “Can you just check this one thing?” messages
that somehow become a second career?
Tell me in the comments.
Your answer may reveal where the next major healthcare
innovation opportunity actually is.
Share this post with a physician or clinic owner who is
still spending too much time fixing problems that should have been prevented
upstream.
And if you believe healthcare can be both more human and
more operationally intelligent, join the conversation.
Continue the Conversation
I write about the messy intersection of medicine,
healthcare operations, AI, medical billing, entrepreneurship, and the future of
physician-owned practices.
The goal isn't to predict the future from a conference
stage.
It's to understand what is actually happening inside
practices—and figure out what should change.
Knowledge is useful. Applied knowledge is leverage.
Explore more:
Website: Dr.
Daniel Cham
Podcast: Spotify
YouTube: Dr.
Cham
Facebook: Dr. Daniel Cham
Free Resource
There is a free resource waiting in the Featured section
of my LinkedIn profile.
No complicated funnel.
No need to schedule a call.
No sales pitch disguised as a white paper.
Just something useful you can take back to your practice.
Start there if you want a practical next step.
About the Author
Dr. Daniel Cham is a physician, healthcare
consultant, entrepreneur, and founder of OnnX, an AI-powered medical
billing platform focused on reducing administrative friction for small and
medium-sized physician-owned practices.
His work focuses on the intersection of clinical
medicine, healthcare management, medical billing, AI, healthcare technology,
and operational design.
His central belief is simple:
Healthcare technology should give physicians more time to
practice medicine—not create another system they have to manage.
Connect with Dr. Cham on LinkedIn.
Disclaimer
This article is intended for general educational and
informational purposes and should not be interpreted as medical, legal,
coding, reimbursement, compliance, or financial advice.
Healthcare laws, payer requirements, contracts, coding
rules, and technology regulations change. Practices should consult
appropriately qualified professionals for guidance specific to their
circumstances.
References
1. VCU Health — Ericka Akoto and Hospital at Home:
The August 24, 2026 patient story provides the human foundation for this
article, including Ericka's experience, the VCU care team, and the couch
intervention.
2. American Hospital Association — Hospital at Home:
Current AHA reporting describes the growth of hospital-at-home programs and the
broader shift toward care delivered outside traditional inpatient settings.
3. American Hospital Association — 2026 Healthcare Demand
Forecast: Current projections point toward continued growth in outpatient,
virtual, and post-acute care, reinforcing the need for healthcare organizations
to connect information across settings.
Final Three Sentences
Stop treating every denial as a billing problem.
Start looking upstream for the signal that created it.
And build healthcare technology that helps physicians see
what matters before the system asks them to clean it up.
#HealthcareAI #MedicalBilling #RevenueCycleManagement #RCM
#PhysicianLeadership #PhysicianOwnedPractice #IndependentPractice
#HealthcareInnovation #HealthTech #MedicalPracticeManagement #DenialPrevention
#HealthcareAutomation #ClinicalOperations #HealthcareEntrepreneur
#PhysicianEntrepreneur #PatientCenteredCare #HealthcareTechnology
#PracticeManagement #AIinHealthcare #OnnX
