Daniel Cham MD

Sunday, August 30, 2026

Giovanni “Gio” Cruz Was Playing Baseball. Then a Rare Disease Changed Everything.

What one family’s journey through medical uncertainty can teach physicians, clinic owners, and healthcare leaders about the human cost of fragmented care—and why better information may be healthcare’s most overlooked innovation.

“The relationship between a patient and their primary care doctor is, at its best, one of the most protective forces in human health – and, right now, we’re systematically dismantling it.” — Lucy McBride, MD, The Guardian, How do you define good health – and achieve it in a broken US medical system?

There is a sentence in a recent healthcare story that I cannot stop thinking about.

Not because it involves artificial intelligence.

Not because it involves a billion-dollar healthcare company.

Not because another payer announced another policy.

It is much simpler than that.

It is about an 11-year-old boy named Giovanni “Gio” Cruz.

In May, Gio was dancing, smiling and playing baseball.

A few weeks later, he was in a children's hospital fighting a rare disease.

His mother, Patricia Cruz, was trying to understand what was happening to her son.

Gio developed severe chest pain. His symptoms escalated. He experienced sweating, back pain and vomiting. He was hospitalized, underwent testing and was eventually transferred to Banner Diamond Children's Medical Center.

Specialists diagnosed him with idiopathic multicentric Castleman disease, a rare immune disorder that can cause widespread inflammation and affect multiple organs.

He is reportedly receiving intensive treatment, including IL-6-directed therapy, high-dose steroids and dialysis.

But the part that stayed with me was not the name of the disease.

It was Patricia's description of uncertainty.

“They want answers from me when I don’t even know what to say or what’s going on.”

Think about that sentence.

Now think about your medical practice.

Because here is my contrarian view:

Healthcare does not have a billing problem.

Healthcare has an information problem.

Billing is simply where the problem becomes expensive.

The patient can live with uncertainty. Your revenue cycle shouldn't have to.

Medicine is complicated.

Rare diseases are complicated.

Patients do not always read the textbook.

Symptoms do not always arrive in neat diagnostic categories.

Sometimes the first test does not solve the mystery.

Sometimes the second test doesn't either.

Sometimes the physician has to say:

“We don't know yet.”

That is not necessarily bad medicine.

Sometimes it is honest medicine.

But imagine the administrative side of the practice saying:

“We don't know why this claim failed.”

Then:

“We don't know where the documentation went.”

Then:

“We don't know who is working on it.”

Then:

“We don't know why the payer didn't pay it.”

Then, six weeks later:

“Let's have a meeting about it.”

That is not clinical uncertainty.

That is operational failure.

And there is a difference.

A very expensive difference.

Here's the uncomfortable question

If a physician can manage a rare disease with a differential diagnosis, why can't the revenue cycle manage a claim with a risk score?

Think about it.

A physician doesn't look at a complex patient and say:

“Everything looks normal. Good luck.”

The physician asks:

What do we know?

What don't we know?

What is most likely?

What could be dangerous?

What information do we need next?

What should we do now?

What requires specialist input?

That is clinical reasoning.

Why shouldn't revenue-cycle technology work the same way?

Instead of:

Claim denied.

Imagine:

Potential authorization problem detected.

Documentation may not support the submitted service.

Payer-specific rule identified.

Missing information detected.

High-risk claim. Human review recommended.

That's a very different philosophy.

It moves billing from:

react → repair

to:

detect → prevent → learn

And that is where I believe healthcare technology is heading.

The great billing myth: “We just need to collect more.”

No.

We need to create fewer problems to collect in the first place.

This is where the healthcare industry sometimes makes me smile.

We build enormous machinery to repair mistakes we could have prevented.

A claim gets denied.

Someone opens it.

Someone researches it.

Someone calls someone.

Someone sends something.

Someone documents the phone call.

Someone updates the account.

Someone resubmits it.

Someone waits.

Someone checks again.

And then someone asks:

“Why is billing so expensive?”

Well...

Look at what we just did.

We turned one preventable problem into a small administrative opera.

Five people.

Three systems.

Two phone calls.

One spreadsheet.

And a partridge in a pear tree.

That's not innovation.

That's organized rework.

The hidden cost isn't the denial

This is the part I wish more physician-owners would measure.

The denial itself is not always the biggest problem.

The bigger problem can be everything surrounding it.

Staff time.

Rework.

Delayed cash.

Management attention.

Payer follow-up.

Appeals.

Patient confusion.

Employee frustration.

Opportunity cost.

And sometimes, the claim is never recovered.

Recent healthcare revenue-cycle analysis has continued to show pressure from denials and revenue leakage. Kodiak Solutions reported that provider organizations experienced a 25% increase in net revenue losses from final denials and bad debt in 2025 in its proprietary analysis.

The precise financial impact varies by organization.

But the direction is hard to ignore.

Revenue leakage is not merely a finance problem.

It is an operational problem.

And here's where small practices get hurt

A large health system can have departments for:

Revenue cycle.

Compliance.

Coding.

Analytics.

IT.

Contracting.

Denials.

Payer relations.

Authorization.

Data science.

A physician-owned clinic?

Maybe it has:

One office manager.

Two billers.

A front-desk team.

A physician who is already seeing patients.

And someone named Linda who somehow knows how the entire practice works.

Every practice has a Linda.

And when Linda leaves?

Suddenly everybody discovers that Linda was the EHR administrator, billing expert, payer encyclopedia, compliance historian and keeper of the sacred spreadsheet.

This is not a technology strategy.

It's institutional memory held together by one exhausted human being.

The future of medical billing isn't about hiring more people to chase more claims.

It is about building systems that require fewer people to chase preventable problems.

That is a very different goal.

And it changes what we should expect from software.

Traditional billing software asks:

What happened?

Modern systems should ask:

What is happening?

The next generation should ask:

What is likely to happen next?

And the really useful system should tell you:

What should we do about it?

That's the difference between a reporting system and an intelligent operating system.

Three experts. One surprisingly consistent lesson.

The most interesting part of the Gio Cruz story is not just the disease.

It is what rare disease teaches us about information.

1. David Fajgenbaum, MD: collaboration changes what is possible

Dr. David Fajgenbaum is a physician-scientist whose personal experience with Castleman disease helped drive the development of the Castleman Disease Collaborative Network.

His work illustrates a powerful principle:

Complex problems improve when information stops living in isolated silos.

Patients.

Physicians.

Researchers.

Families.

Clinical data.

Research data.

The goal is not simply to collect information.

It is to make the information useful to the people who need it.

That lesson translates directly to medical practices.

If your EHR knows something your billing system doesn't...

If your scheduling system knows something your authorization workflow doesn't...

If your billing system knows something your physician-owner doesn't...

You don't have an information system.

You have information islands.

And islands are beautiful on vacation.

They're terrible for healthcare operations.

2. Frits van Rhee, MD, PhD: complexity demands structure

Dr. Frits van Rhee is another major figure in Castleman disease research.

The broader lesson from rare-disease medicine is that complex clinical problems require structured approaches to diagnosis, treatment and collaboration.

The same principle applies to revenue cycle.

A complex claim should not simply be dumped into the same queue as everything else.

It should be recognized as complex.

It should receive appropriate attention.

It should be prioritized.

And, when appropriate, it should be escalated to a human.

Complexity should trigger intelligence, not confusion.

3. Andrew Knight and colleagues: you don't need perfect certainty to act

Recent rare-disease guidance emphasizes practical coordination and support for clinicians caring for patients with uncommon conditions.

That principle is important.

Healthcare often operates before complete certainty exists.

That's normal.

The mistake is believing that uncertainty means the system should become disorganized.

It doesn't.

A good system says:

We don't know everything yet.

But we know what we know.

We know what we don't know.

And we know what we're doing next.

That is exactly how good clinical teams operate.

It should also be how good administrative systems operate.

The lesson hidden inside Gio Cruz's story

Gio's physicians faced a difficult clinical problem.

The disease was rare.

His symptoms were serious.

The diagnosis was not immediately obvious.

The family was frightened.

Nobody could simply press a button and make certainty appear.

But there was still a process.

Evaluation.

Testing.

Specialist involvement.

Diagnosis.

Treatment.

Monitoring.

Adjustment.

That process matters.

Now compare that with a typical denied claim.

What happens?

Usually:

Denial.

Queue.

Research.

Correction.

Resubmission.

Wait.

Maybe payment.

Maybe another denial.

Where is the learning?

Often, it is buried somewhere in a report.

That is the missed opportunity.

A denial should be a teacher

This is one of my favorite contrarian ideas.

Stop treating every denial as a task.

Treat it as a lesson.

If one claim is denied because authorization was missing, that's a task.

If 37 claims are denied because authorization was missing, that's a process failure.

If the same payer creates the same authorization problem every month, that's a systems problem.

And if you continue fixing each claim individually?

Congratulations.

You have built a very efficient machine for repeating the same mistake.

The real question is:

Where did the error begin?

Maybe it started at scheduling.

Maybe eligibility.

Maybe authorization.

Maybe documentation.

Maybe coding.

Maybe charge capture.

The denial is simply where the problem became visible.

Visibility is not origin.

That distinction is crucial.

The billing equivalent of a differential diagnosis

Here is a framework I think physician-owned practices should adopt.

Instead of looking at a denial as:

“Why didn't they pay?”

Ask:

1. What happened?

The claim was denied.

2. What are the possible causes?

Eligibility?

Authorization?

Coding?

Documentation?

Payer policy?

Timely filing?

Technical rejection?

3. What evidence supports each possibility?

Look at the data.

4. What is the most likely root cause?

Find the pattern.

5. What action should happen next?

Assign responsibility.

6. How do we prevent recurrence?

Fix the upstream process.

That's basically clinical reasoning applied to operations.

And frankly, healthcare should be better at this than almost any other industry.

We invented the differential diagnosis.

Why are we still treating billing problems like random acts of nature?

Statistics: don't fall in love with the dashboard

Healthcare leaders love dashboards.

I understand why.

Dashboards look intelligent.

Lots of numbers.

Lots of graphs.

Lots of green.

Very reassuring.

Until the cash doesn't arrive.

A dashboard should not exist to make executives feel informed.

It should exist to help someone make a better decision.

For a physician-owned practice, I would focus on a relatively small group of operational signals:

First-pass acceptance

Denial rate

Denial reason

Days in A/R

A/R aging

Net collection rate

Charge lag

Payment lag

Underpayments

Rework hours

Exception volume

And I would add one more:

Preventable error rate

Because that's where the future is.

Not:

“How many problems did we fix?”

But:

“How many problems did we prevent?”

The metric nobody puts on the wall

Here's another contrarian metric:

Human touches per claim.

How many times does a human have to touch a claim?

One?

Two?

Five?

Ten?

Twenty?

Nobody celebrates this number.

Maybe we should.

Because every human touch has a cost.

Not just payroll.

Attention.

Context switching.

Fatigue.

Opportunity cost.

And error risk.

If a routine claim requires six human touches, don't immediately ask:

“How can we make our staff faster?”

Ask:

“Why does this claim need six human touches?”

That's a better question.

AI is not the answer

Now I am going to disappoint the AI crowd.

AI is not the answer.

There.

I said it.

AI is a tool.

A potentially powerful one.

But if you put AI on top of a broken workflow, you may simply create a faster broken workflow.

That's not transformation.

That's turbocharged dysfunction.

The useful question is:

Where does intelligence actually belong?

AI can help with:

Pattern recognition

Anomaly detection

Claim validation

Risk prediction

Denial classification

Work prioritization

Documentation comparison

Payer-rule analysis

Workflow recommendations

But AI should not become an excuse to remove human accountability.

The ideal relationship is simple:

AI finds.

AI explains.

AI prioritizes.

Human validates.

Human decides when judgment matters.

The most dangerous AI in healthcare

It isn't necessarily the hallucinating chatbot.

It's the AI that looks confident.

Imagine a system that says:

“This claim is fine.”

And everyone believes it.

That's dangerous.

A better system might say:

“This claim appears low risk based on the available information. Here are the validation checks completed.”

And:

“These two elements remain uncertain.”

That's much more useful.

Healthcare does not need artificial confidence.

It needs transparent assistance.

Why I built OnnX

This is the problem that led me to build OnnX.

I am a physician.

I have spent enough time around healthcare to see how often good clinicians are forced to operate inside bad administrative systems.

The problem is not that people don't care.

Usually, they care enormously.

The problem is that the workflow asks them to compensate for system weaknesses manually.

My philosophy is simple:

Capture better information.

Validate earlier.

Surface risk sooner.

Reduce repetitive work.

Give physicians visibility.

Keep people involved where judgment matters.

OnnX is being built around that idea.

Not:

“Let's put AI on billing.”

But:

“Let's rethink what billing should have been doing all along.”

That's a much bigger ambition.

The industry asks the wrong question

The usual question is:

“How do we get more claims paid?”

I think the better question is:

“Why are we creating claims that need fixing?”

That sounds like a small difference.

It isn't.

The first question creates a reactive organization.

The second creates a preventive organization.

The first rewards recovery.

The second rewards learning.

The first measures output.

The second measures system quality.

And that is a fundamental shift.

Five “best practices” I would challenge

Best practice #1: “Denials are inevitable.”

Some are.

Many patterns are not.

If the same preventable denial happens repeatedly, calling it “inevitable” is just a polite way of saying we stopped looking for the cause.

Best practice #2: “Just hire another biller.”

Sometimes staffing is absolutely necessary.

But if the workflow is broken, adding people can hide the problem.

Before adding labor, ask:

Can we eliminate the work?

Then:

Can we automate the work?

Then:

Can we simplify the work?

Only then:

Do we need more people?

Best practice #3: “Billing starts after the encounter.”

No.

Billing starts when information enters the system.

Registration matters.

Eligibility matters.

Scheduling matters.

Authorization matters.

Documentation matters.

Coding matters.

Charge capture matters.

The claim is the final product of a much larger information chain.

Best practice #4: “The physician doesn't need to know.”

The physician doesn't need to know every claim.

But the physician-owner needs to know whether the practice is leaking money.

That's not micromanagement.

That's ownership.

Best practice #5: “More software means better technology.”

Absolutely not.

Sometimes the best technology is the technology that removes a screen.

Or removes a spreadsheet.

Or removes a phone call.

Or removes a manual handoff.

The best workflow is often the one you no longer need.

The front desk may be your most important revenue-cycle department

Here's another uncomfortable truth.

Some billing problems are born before the biller ever sees the claim.

An incorrect demographic field.

An outdated insurance card.

A missing authorization.

A scheduling mismatch.

A payer-specific requirement nobody noticed.

By the time the biller receives the claim, the mistake has already matured.

The biller is now asked to perform archaeology.

This is why I believe revenue-cycle technology needs to move upstream.

Don't wait until the claim becomes a problem.

Catch the problem when it is still cheap to fix.

Fix problems where they begin

This principle is simple:

The farther downstream a mistake travels, the more expensive it becomes.

Consider a missing authorization.

At scheduling:

Easy to correct.

At check-in:

Still manageable.

After the encounter:

More complicated.

After claim submission:

More expensive.

After denial:

Now someone has a work queue.

After appeal:

Now management may be involved.

After timely filing expires:

Maybe the revenue is simply gone.

Same mistake.

Different price.

That's why prevention is so powerful.

The 20-minute physician-owner audit

You don't need a consulting firm.

Start with 20 minutes.

Ask your team:

Question 1

What are our three most common denials?

Question 2

Where does each one actually begin?

Question 3

How many staff hours do we spend fixing them?

Question 4

Which one could we prevent first?

Question 5

What information would have allowed us to catch it earlier?

Question 6

Can technology help us catch it?

Question 7

Who owns the fix?

That's it.

You don't need a 74-slide PowerPoint.

You need answers.

A seven-step revenue-cycle reset

Step 1: Map the journey

Patient registration.

Scheduling.

Eligibility.

Authorization.

Encounter.

Documentation.

Coding.

Charge.

Claim.

Adjudication.

Payment.

Denial.

Appeal.

Do not assume everyone sees the same journey.

Map it.

Step 2: Identify the friction

Where are people copying information?

Where are they switching systems?

Where are they waiting?

Where are they calling?

Where are they manually checking?

Those are opportunities.

Step 3: Find the repeat offender

Which denial keeps coming back?

That's your first target.

Step 4: Trace it upstream

Find where the problem began.

Do not stop at the denial code.

Step 5: Create a rule

If a predictable problem can be identified before submission, create a validation rule.

Step 6: Create an exception path

Not every claim needs human attention.

Not every claim should bypass human review.

Build a smart middle.

Step 7: Measure prevention

Track whether the same problem actually declines.

If it doesn't, your intervention didn't work.

That's okay.

Learn.

Change it.

Try again.

Legal and compliance reality

Here is where enthusiasm needs a seatbelt.

Healthcare technology cannot turn compliance into a checkbox.

Practices must consider applicable requirements involving:

HIPAA

Privacy

Security

Coding

Documentation

Payer contracts

Fraud and abuse

False claims risk

Auditability

Vendor agreements

AI governance

AI does not transfer responsibility to the algorithm.

If an automated recommendation is wrong, the practice still needs appropriate controls.

That is why explainability, audit trails, access controls and human oversight matter.

The goal is not:

“The AI told us to do it.”

The goal is:

“The system identified an issue, showed us why, and the appropriate person made the decision.”

That's defensible.

Ethical question: what happens when revenue optimization wins?

This deserves more discussion.

Healthcare organizations have a legitimate responsibility to collect appropriate payment for services provided.

But optimization can become dangerous when the financial objective becomes more important than clinical truth.

Technology should never encourage:

Unsupported coding.

Misleading documentation.

Unnecessary services.

Aggressive interpretation of clinical facts.

Gaming payer rules.

The goal is not:

maximize the bill.

The goal is:

accurately represent the care that was provided and get appropriately reimbursed for it.

That's a very different philosophy.

Myth Buster

Myth: AI will eliminate billing staff.

Reality: The better objective is to eliminate unnecessary billing work.

Myth: A low denial rate means you're doing great.

Reality: Look at underpayments, aging, write-offs, rework and net collections too.

Myth: The billing department owns the revenue cycle.

Reality: Revenue-cycle performance is influenced by the entire practice.

Myth: Every denial should be appealed.

Reality: The right response depends on the reason, documentation, economics and likelihood of recovery.

Myth: Small practices can't use sophisticated technology.

Reality: Small practices may benefit disproportionately because they have less administrative capacity to absorb inefficiency.

Myth: AI makes billing objective.

Reality: AI reflects the data, rules and assumptions built around it. Human oversight still matters.

The future is not “AI billing”

I don't think that's the right phrase.

I think the future is:

Predictive revenue-cycle management.

The system should know enough to say:

“This looks normal.”

Or:

“This looks unusual.”

Or:

“This claim is likely to encounter a problem.”

Or:

“Here is the reason.”

Or:

“Here is what you should check.”

Or:

“This problem has happened 18 times this month. You should probably fix the process rather than the claims.”

Now we're getting somewhere.

The real competitive advantage for physician-owned practices

It won't necessarily be having the largest staff.

It won't necessarily be having the most software.

It won't necessarily be having the fanciest AI.

It may be something much less exciting:

Knowing what is happening sooner.

That sounds boring.

Good.

Boring is underrated.

Predictable cash flow is boring.

Accurate claims are boring.

Fewer denials are boring.

Employees not spending Friday afternoon fixing the same error for the 400th time is boring.

Boring is wonderful.

Healthcare has enough excitement.

What healthcare founders should learn

If you're building healthcare technology, here's my challenge.

Stop asking:

“Where can we add AI?”

Start asking:

“Where is a human repeatedly compensating for a system failure?”

That's the opportunity.

Find the spreadsheet.

Find the workaround.

Find the person everyone calls when something breaks.

Find the task employees complain about but have accepted as normal.

Find the process that requires three systems.

Find the task someone does every morning because “that's just how we've always done it.”

That's where the innovation is hiding.

A word about middlemen

I am deliberately provocative here.

Healthcare has accumulated layers.

EHR.

Practice management.

Clearinghouse.

Billing company.

Consultant.

RCM vendor.

Coding vendor.

Analytics platform.

Payer portal.

Another portal.

Another login.

Another spreadsheet.

Every layer may have a legitimate purpose.

But every layer also creates a question:

Who actually owns the outcome?

Physician-owned practices should have more visibility and control.

That does not mean every practice must insource everything.

It means outsourcing should not require surrendering intelligence.

If a vendor is managing your revenue cycle, you should still be able to understand:

What is happening.

Why it is happening.

What is being done.

What is being recovered.

What keeps recurring.

And what is being done to prevent it.

Transparency should not be an upgrade.

It should be the baseline.

The OnnX thesis

My thesis behind OnnX is simple:

Healthcare billing is a data-quality problem before it is a billing problem.

Bad information creates bad claims.

Bad claims create denials.

Denials create rework.

Rework consumes labor.

Labor increases cost.

Delayed payment affects cash flow.

And the cycle repeats.

So why start at the end?

Start upstream.

Capture.

Validate.

Predict.

Submit.

Monitor.

Learn.

That's the loop.

What I would do if I owned a small practice tomorrow

Monday morning, I would ask for the last 90 days of denial data.

Not the dashboard.

The actual reasons.

Then I would sort them.

Top 10 causes.

Then I would ask:

Which three are preventable?

Then:

Which one costs us the most?

Then:

Where does it begin?

Then:

Can we catch it before the claim is submitted?

Then:

Can technology help?

And finally:

Who owns the change?

That's it.

No massive transformation project.

No six-month committee.

No 300-page strategy.

Just one problem.

Then another.

The human lesson

Let's return to Gio.

Because this article should not lose the human being who started it.

Giovanni “Gio” Cruz is not a billing problem.

He is not a statistic.

He is not a workflow.

He is not a claim.

He is a child.

His mother, Patricia Cruz, is not a data point.

She is a mother trying to understand what is happening to her son.

The medical team is dealing with a rare and serious condition.

There is uncertainty.

There are questions.

There are decisions.

There is fear.

Healthcare cannot eliminate all of that.

But it can decide how much additional friction to create.

That matters.

Because every unnecessary administrative problem consumes someone's attention.

And attention is one of the most valuable resources in healthcare.

The bigger idea

Maybe the real measure of healthcare technology isn't how much technology we deploy.

Maybe it is how much unnecessary work disappears.

How many clicks?

How many calls?

How many spreadsheets?

How many duplicate entries?

How many preventable denials?

How many hours spent searching?

How many times does a patient have to repeat the same information?

How many times does a physician have to explain something already documented?

How many times does a biller have to repair something that could have been prevented?

That is the technology scorecard I care about.

Not:

How intelligent does the software look?

But:

How much unnecessary friction did it remove?

The next healthcare advantage may be boring

Healthcare loves breakthroughs.

The new drug.

The new device.

The new AI model.

The new platform.

The new billion-dollar startup.

But some of the biggest improvements may look remarkably ordinary.

A claim that never becomes a denial.

A prior authorization caught before the appointment.

A missing field identified before submission.

A physician who can see the practice's financial health in five minutes.

A biller who no longer spends half the day fixing the same error.

A patient who receives a clear statement.

A small practice that keeps more of the revenue it legitimately earned.

None of that makes a flashy conference keynote.

But it matters.

Three questions every physician-owner should ask

1. What problem are we repeatedly fixing instead of preventing?

If you don't know, find out.

2. Where does that problem actually begin?

The denial may not be the origin.

3. What information would have allowed us to catch it earlier?

That question points directly toward better workflow and better technology.

Three actions for this week

Action 1: Find your most expensive recurring problem.

Not the most annoying.

The most expensive.

Action 2: Trace it upstream.

Find the first point where the problem could have been detected.

Action 3: Ask whether the next occurrence can be prevented.

If yes, build the rule.

If no, build the exception workflow.

Then measure what happens.

Final Thoughts: Medicine can tolerate uncertainty. Systems shouldn't manufacture it.

Gio Cruz's story began with uncertainty.

A child became sick.

His family needed answers.

The diagnosis was difficult.

The road ahead remained unclear.

That's medicine.

We should not pretend otherwise.

But there is another kind of uncertainty that physicians and clinic owners do not have to accept.

Where is the claim?

Why was it denied?

Who is fixing it?

Why does this keep happening?

How much money are we losing?

Which problem should we address first?

Those are not mysteries.

They are information problems.

And information problems can be designed better.

That's where I believe healthcare technology has an enormous opportunity.

Not to replace physicians.

Not to replace billers.

Not to make another dashboard.

Not to sprinkle AI dust over an old workflow and call it innovation.

But to make the right information available earlier, to the right person, with the right next action.

That is what intelligent healthcare operations should look like.

And if we get that right, something interesting happens.

The technology becomes almost invisible.

The physician gets more time.

The staff gets less rework.

The practice gets better visibility.

The patient gets a smoother experience.

And healthcare becomes just a little more human.

That is the kind of innovation worth building.

Get Involved: Don't Just Read This. Challenge It.

Here is my question for physicians and clinic owners:

If your practice could identify a preventable billing problem before the claim was submitted, why would you wait for the denial?

Maybe you disagree.

Good.

Tell me in the comments.

What is the biggest source of administrative friction in your practice today?

Denials?

Authorization?

Coding?

Eligibility?

Documentation?

Payer rules?

Or something nobody is talking about?

Saturday, August 29, 2026

Tyler Tabor Was Once on the Gurney. Now He’s Becoming the Doctor: What His Story Reveals About the Human Cost of Healthcare

A cancer survivor’s journey from patient to medical student reveals what healthcare technology should really be designed to protect: the physician’s attention and the human connection at the heart of care.

“AI should be a tool that empowers physicians, restores time with patients, and ultimately humanizes healthcare.” — Dr. Imamu “Mu” Tomlinson, emergency physician and CEO of Vituity

At 18, Tyler Tabor was not thinking about artificial intelligence.

He wasn't thinking about medical billing.

He wasn't thinking about healthcare innovation.

He was thinking about cancer.

Tabor had been diagnosed with Stage IIB Hodgkin's lymphoma involving his neck. He began treatment at The University of New Mexico Hospital in Albuquerque, New Mexico.

He was frightened.

He was young.

And suddenly, instead of preparing for whatever comes next in life, he was learning how to survive it.

His pediatric oncologist was Jessica Valdez, MD, MPH, FAAP.

But something unusual happened during that relationship.

Tabor didn't just become Valdez's patient.

He told her he wanted to become a doctor.

And Valdez apparently took that dream seriously.

She told him they would get him there.

Years later, after treatment, relapse and a stem-cell transplant in Colorado, Tabor survived cancer.

Then came the next chapter.

On July 24, 2026, Tabor put on a white coat as a first-year medical student at the University of New Mexico School of Medicine.

The former cancer patient was now entering the profession that had once cared for him.

He described it beautifully:

He was moving to “the other side of the gurney.”

Think about that for a second.

The patient became the medical student.

The frightened teenager became the future physician.

And the physician who treated him became one of the people who helped him imagine that future.

That's healthcare at its best.

Not a dashboard.

Not an algorithm.

Not a billing platform.

A relationship.

And that is exactly why I think we are asking the wrong question about healthcare technology.

We keep asking:

“What can AI do?”

Maybe we should be asking:

“What are we making physicians do that they never should have been doing in the first place?”

The uncomfortable healthcare problem nobody wants to call by its real name

We have a strange habit in healthcare.

We say we want physicians to spend more time with patients.

Then we give them more administrative work.

We say we want clinicians to practice at the top of their license.

Then we ask them to chase documentation.

We say patient experience matters.

Then we make patients navigate confusing billing systems.

We say physicians should listen.

Then we interrupt them with tasks that software could potentially handle.

We say burnout is a physician wellness problem.

Sometimes it looks suspiciously like a workflow problem.

The American Medical Association continues to identify administrative burden as something that directly interferes with the physician-patient relationship.

That's the part we should pay attention to.

Because administrative burden isn't simply annoying.

It competes with something scarce:

human attention.

And attention is one of the most valuable resources in medicine.

What if physician attention were a clinical resource?

We measure everything.

Visits.

RVUs.

Claims.

Denials.

Days in A/R.

Patient satisfaction.

Length of stay.

Readmissions.

Productivity.

But we rarely ask:

How much of the physician's attention did the system consume today?

Imagine a physician begins the morning with 100 units of attention.

A patient needs 20.

Another patient needs 15.

A complicated diagnosis needs 25.

A family needs 10.

A trainee needs 5.

Now add:

Three payer messages.

Two coding questions.

A rejected claim.

A prior authorization.

A documentation clarification.

An insurance portal.

A form.

Another form.

Suddenly the physician isn't short on intelligence.

They're short on attention.

And no productivity seminar can manufacture more of it.

The irony of healthcare AI

Here's my contrarian take:

Healthcare doesn't necessarily need more AI.

It needs less unnecessary work.

That's different.

We have become fascinated with AI because it sounds futuristic.

But the most valuable application of AI in a medical practice may be remarkably unglamorous.

Find the denied claim.

Explain why it was denied.

Find the relevant documentation.

Identify the likely correction.

Prepare the next step.

Ask a human to approve it.

Move on.

No robot doctor.

No holographic physician.

No sci-fi soundtrack.

Just fewer things sitting in someone's inbox.

And honestly?

That might be more useful.

The AI arms race may be missing the point

Healthcare organizations are racing to announce AI initiatives.

AI scribes.

AI assistants.

AI copilots.

AI agents.

AI coding.

AI claims.

AI documentation.

AI everything.

But here's the question I would ask before buying any of it:

What human work disappears?

If the answer is:

“None, but now we have an AI dashboard,”

we haven't solved the problem.

We've added another dashboard.

Congratulations.

The inbox has acquired artificial intelligence.

The inbox is still an inbox.

Tyler Tabor's story exposes something technology cannot replace

Tabor's story is powerful precisely because it is so human.

He was sick.

Someone cared for him.

Someone encouraged him.

He survived.

He remembered.

And now he wants to become that person for somebody else.

His oncologist, Jessica Valdez, became more than the person treating his cancer. She became a mentor and role model.

That distinction matters.

Because healthcare isn't simply an exchange of information.

It's an exchange of trust.

A patient is often asking a physician something deeper than:

“What is my diagnosis?”

They're asking:

“Am I going to be okay?”

No software can completely answer that question.

Even when the medical answer is uncertain, the physician can still say:

“We're going to work through this together.”

That's not inefficiency.

That's medicine.

And yet we're spending enormous amounts of clinician time on things that aren't medicine

This is where I see the opportunity.

The administrative machinery surrounding medicine has become incredibly complex.

A typical revenue-cycle journey can look something like:

Patient → documentation → coding → claim → payer → denial → correction → appeal → payment → A/R

Every arrow is a handoff.

Every handoff is an opportunity for delay.

Every delay can create more work.

And every additional manual step consumes someone's attention.

The problem isn't that billing exists.

Billing has to exist.

Clinics have payroll.

They have rent.

They have supplies.

They have staff.

They have technology.

They need revenue to continue providing care.

The problem is that we sometimes confuse necessary administration with necessary human labor.

Those aren't the same thing.

This is where I think the industry has it backward

We often ask:

“Can AI replace the biller?”

I think that's the wrong question.

Ask instead:

“Which parts of the billing workflow should never have required a person to perform them manually?”

That's a much more interesting question.

Maybe the future isn't:

AI versus billers.

Maybe it's:

AI + billing expertise.

Let AI search.

Let AI organize.

Let AI identify patterns.

Let AI flag exceptions.

Let AI prepare.

Let humans decide.

That is a much more defensible model.

What I'm building with OnnX

This is the philosophy behind OnnX, the AI-powered medical billing SaaS I founded.

The goal isn't to replace the people who understand healthcare.

It's to reduce unnecessary friction in the workflow.

Think about a denied claim.

Traditional workflow:

Someone notices it.

Someone opens the payer portal.

Someone reads the denial.

Someone finds the chart.

Someone checks the documentation.

Someone asks what happened.

Someone figures out the correction.

Someone resubmits it.

Someone tracks it.

Someone follows up.

And eventually someone wonders:

“Why did we spend this much human time on one claim?”

That's the opportunity.

OnnX is built around the idea that AI can help analyze the problem, retrieve relevant information, recommend the next action and prepare the work for human review.

The human remains accountable.

The workflow becomes smarter.

And ideally, the physician sees less of it.

That's the point.

Don't automate the mess

Here's another unpopular opinion:

Don't automate a broken workflow.

Fix it first.

If three people manually enter the same information into three systems, don't immediately build AI to do the same thing faster.

Ask why the information has to be entered three times.

If a denial requires five people to understand, don't immediately build a five-person AI workflow.

Ask why the process is so difficult.

If a physician is being asked to clarify the same documentation issue repeatedly, don't just add another notification.

Fix the source.

Automation without workflow redesign is just faster bureaucracy.

That's the sentence I would put on the wall.

Three questions every clinic owner should ask

Before buying another healthcare technology product, ask:

1. What problem are we actually solving?

Not:

“Where can we use AI?”

Ask:

“Where are we losing time, money or attention?”

2. What happens today?

Map the actual workflow.

Not the PowerPoint version.

The real version.

Who touches the work?

How many times?

How many systems?

How many handoffs?

How many exceptions?

3. What should disappear?

That's the most important question.

Not:

“What new feature should we add?”

But:

“What work should no longer exist?”

The statistics tell an uncomfortable story

Recent healthcare reporting reinforces the pressure.

The latest AMA data show physician burnout has declined overall, but some specialties continue to report burnout rates above 40%.

Meanwhile, a recent Healthcare IT News report found that nearly two-thirds of surveyed practice employees said manual data entry consumes at least an hour of their workday.

That's a lot of human attention.

And here's the interesting part.

We don't necessarily have to solve it with more people.

We can redesign the work.

That's where automation becomes interesting.

But AI doesn't automatically save time

This is where I'm intentionally skeptical.

AI vendors love saying:

“Save hours every week.”

Maybe.

Maybe not.

Recent medical commentary has questioned whether AI tools, including clinical documentation systems, always deliver the promised time savings in real-world practice.

Why?

Because implementation matters.

A tool can generate a draft.

Someone still has to review it.

A tool can identify a claim problem.

Someone still needs to validate it.

A tool can automate one task.

But if it creates two new tasks, you've gone backward.

That's why I don't think AI adoption should be the metric.

Work eliminated should be.

The new ROI question

Forget:

“How many AI features did we deploy?”

Ask:

How many unnecessary human touches did we remove?

Then ask:

What happened to the time we recovered?

That's where ROI becomes interesting.

If staff saved 100 hours and used them to process 100 more claims, that's one outcome.

If those 100 hours allowed them to answer patient calls faster, coordinate care, reduce delays and support physicians, that's another.

And if physicians recovered meaningful time with patients?

Now we're talking about something bigger than revenue-cycle efficiency.

We're talking about care capacity.

Three experts. Three lessons.

Jessica Valdez, MD, MPH, FAAP: don't underestimate the power of believing in a patient

Valdez treated Tabor when he was a teenager.

But she also encouraged his ambition to become a physician.

The lesson is profound:

A clinician's influence can outlive the clinical encounter.

Tabor didn't just survive cancer.

He carried something from that experience into his future profession.

Healthcare leaders should think about that.

Every patient encounter has an emotional component.

Every clinician has an opportunity to shape how a patient sees the future.

That cannot be reduced to a CPT code.

Richard Holt and the lesson of “boring” quality

Recent commentary in The Permanente Journal emphasizes patient perceptions of timeliness, coordination, clarity and kindness as meaningful aspects of cancer care.

Notice something.

None of those words sounds particularly futuristic.

That's the point.

Healthcare doesn't always need to become more complicated to become better.

Sometimes it needs to become easier to navigate.

Dr. Imamu “Mu” Tomlinson: AI should give physicians time, not take judgment away

Recent discussion around AI in healthcare has emphasized a boundary worth preserving:

AI can support physicians.

It should not casually replace human judgment in consequential medical decisions.

That is particularly important when decisions involve life-changing treatment, patient context or end-of-life care.

The principle applies to administrative AI too.

Automate the repetitive. Escalate the uncertain. Keep humans accountable.

Simple.

Not always easy.

But simple.

The myth of “full automation”

Here's a myth I would like healthcare leaders to retire:

The best AI system is the one that requires the least human involvement.

Not necessarily.

The best system is the one that puts human involvement where it creates the most value.

If AI can check 10,000 routine data points, let it.

If a physician needs to decide whether a complicated case is adequately supported, let the physician decide.

If a biller needs to handle an unusual payer situation, let the biller handle it.

The goal isn't zero humans.

The goal is humans doing human work.

Another myth: “Billing is just back-office work”

I disagree.

Billing is back-office infrastructure.

But infrastructure affects the front office.

A clinic with poor cash flow may delay hiring.

A practice with chronic denial problems may lose resources.

Administrative overload can contribute to burnout.

Billing confusion can frustrate patients.

So yes, billing is administrative.

But administrative doesn't mean irrelevant to patient care.

The legal and ethical line

AI-powered billing has another reality that should not be ignored.

Healthcare data is sensitive.

Claims are consequential.

Coding has compliance implications.

Documentation matters.

Payer contracts matter.

Privacy matters.

Auditability matters.

So a responsible system needs more than a clever model.

It needs:

security

access controls

audit trails

human review

data governance

clear accountability

appropriate vendor agreements

transparent workflows

The question isn't merely:

“Can AI do this?”

It is:

“Can AI do this safely, explainably and accountably?”

The workflow I want to see

Imagine this.

A claim is submitted.

The system notices a potential problem.

Instead of waiting for the payer to reject it, the system flags the issue.

It explains why.

It identifies supporting information.

It gives the billing professional a recommendation.

The professional approves.

The claim moves forward.

If the system isn't confident?

It escalates.

If the issue involves clinical judgment?

It doesn't pretend otherwise.

If the action has significant consequences?

A human remains in control.

That's not AI replacing healthcare workers.

That's AI respecting healthcare workers' time.

A six-step playbook for clinic owners

Step 1: Find your ugliest workflow

Not your most exciting one.

Your ugliest.

The one everyone complains about.

Step 2: Count the human touches

How many people touch it?

How many times?

Step 3: Identify the exception

What causes the workflow to break?

Step 4: Separate routine from judgment

Automate the routine.

Protect the judgment.

Step 5: Pilot one workflow

Don't transform the entire practice on Monday morning.

Start small.

Step 6: Measure what matters

Track:

Denial rate

Clean claim rate

Days in A/R

A/R aging

Staff touches

Time to resolution

First-pass resolution

Administrative minutes

And one more:

Physician attention returned.

What failure looks like

Let's be honest.

Some AI projects will fail.

Some integrations will be painful.

Some models will make mistakes.

Some staff will hate the first version.

Some workflows will turn out to be more complicated than expected.

That's normal.

The real failure is pretending otherwise.

The better approach is to build feedback loops.

Ask staff:

What did the system get wrong?

Ask physicians:

What interrupted you?

Ask billing teams:

What still requires manual work?

Ask patients:

Did anything actually become easier?

Innovation isn't the absence of failure.

It's learning faster than the failure costs you.

The funniest thing about healthcare innovation

We sometimes spend $500,000 trying to save five minutes.

Then discover the five minutes were spent because someone had to log into three different systems.

I'm exaggerating.

But only slightly.

Healthcare has accumulated layers of software over decades.

EHR.

Payer portal.

Clearinghouse.

Scheduling system.

Billing platform.

Fax.

Email.

Spreadsheet.

Password manager.

And, somewhere in the corner:

one person who knows how everything actually works.

That person is usually the real operating system.

And everyone is terrified they'll take a vacation.

That's not digital transformation.

That's institutional memory with a login.

The real opportunity for healthcare founders

If you're building healthcare technology, I would challenge you to stop asking:

“Where can we insert AI?”

Ask:

“Where is human attention being wasted?”

That is a better startup question.

Find repetitive cognitive work.

Find fragmented workflows.

Find expensive handoffs.

Find exception-heavy processes.

Find tasks physicians hate.

Find tasks nurses hate.

Find tasks billing teams hate.

Then design around the workflow.

Not the technology.

The technology is the means.

The workflow is the product.

Why small and midsize clinics matter

Large health systems can absorb inefficiency differently.

Small and midsize practices often cannot.

One employee leaving can matter.

One prolonged denial can matter.

One broken workflow can matter.

One hour of physician time can matter.

One unnecessary software subscription can matter.

That's why I believe healthcare automation needs to become more practical.

Less:

“Look what our AI can do.”

“Here is the work we removed.”

That's a much harder claim.

It's also much more useful.

What I would measure at OnnX

If you're building an AI medical billing platform, vanity metrics are easy.

Number of claims processed.

Number of AI interactions.

Number of users.

Number of recommendations.

Those are interesting.

But I care more about:

How many claims required human intervention?

How quickly were denials identified?

How often were recommended corrections accepted?

How much manual work disappeared?

How much revenue moved through the system?

How much physician or staff attention was returned?

That is where the value lives.

The bigger idea: attention is infrastructure

We normally think of infrastructure as roads, buildings, networks and software.

I think healthcare has another form of infrastructure:

human attention.

And we're consuming it faster than we're replenishing it.

Every unnecessary click takes a little.

Every redundant form takes a little.

Every avoidable denial takes a little.

Every unnecessary notification takes a little.

Every poorly designed workflow takes a little.

Eventually, you have a clinician who is physically present but mentally fragmented.

That's dangerous.

Because the opposite of patient-centered care isn't necessarily cruelty.

Sometimes it's distraction.

Tyler Tabor gives us the test

Tyler Tabor's story gives healthcare technology a simple test.

Imagine Tabor's future patient.

Imagine that patient sitting across from him.

Imagine Tabor trying to listen.

Now imagine someone interrupts him with an unnecessary administrative task.

Then another.

What should technology do?

Not make Tabor faster at multitasking.

Not give him another dashboard.

Not turn him into a more efficient administrator.

Give him his attention back.

Because someday, there may be another 18-year-old sitting on the other side of that gurney.

And that patient deserves the doctor.

Not the inbox.

The future of medical billing shouldn't look like more billing

This may sound strange coming from someone who founded an AI medical billing company.

But I don't want the future to be about making physicians better at billing.

I want it to be about making billing less visible to physicians.

That is the difference.

The physician should understand the economics of the practice.

Absolutely.

The physician should understand documentation and coding.

Yes.

But they shouldn't have to become a human middleware layer connecting every broken administrative system.

That's what software should be for.

Here's the argument I would make to healthcare leaders:

Stop trying to make physicians more efficient at doing unnecessary work.

Instead, eliminate the work.

That's harder.

It requires redesign.

It requires uncomfortable conversations.

It may require changing contracts, workflows, staffing models and software.

But that's where real innovation lives.

Not in adding another tool.

In removing a step.

Final Thoughts: What if the future of healthcare is actually less technological?

Tyler Tabor survived cancer.

His physician encouraged him.

His mother supported him.

His community helped him.

Now he's learning to become the physician on the other side of the gurney.

There is something almost beautifully old-fashioned about that story.

One human being helped another human being.

Then the first person decided to help someone else.

That's healthcare.

Technology should support that chain.

It shouldn't interrupt it.

So perhaps the question isn't:

“How much AI will healthcare use?”

Perhaps the better question is:

“How much unnecessary work can we remove before AI even becomes necessary?”

And when AI is useful, let's use it.

Let it search.

Let it summarize.

Let it detect patterns.

Let it organize.

Let it predict.

Let it prepare.

But when a patient looks across the room and asks:

“Doctor, what happens now?”

I don't want the system answering.

I want the physician to have enough attention left to answer.

That's the standard.

And perhaps that's what Tyler Tabor's story really teaches us.

The future physician may have better technology.

But the patient will still need a human being.

Get Involved

Here's the question I want to leave with physicians and clinic owners:

If you could permanently eliminate one administrative task from your practice tomorrow, what would it be?

Don't give me the polished answer.

Give me the task that makes you think:

“Why are we still doing this?”

Tell me in the comments.

And if this perspective resonates with you, repost it and bring another physician, practice owner, administrator, or healthcare innovator into the conversation.

Because healthcare doesn't need another slogan about transformation.

It needs fewer unnecessary steps.

Find one workflow. Fix one bottleneck. Give one clinician some attention back.

That's where meaningful change starts.

About the Author

Dr. Daniel Cham is a physician, medical consultant and healthcare entrepreneur focused on the intersection of medical technology, healthcare operations, medical billing and workflow automation.

He is the founder of OnnX, an AI-powered medical billing SaaS focused on helping small and midsize medical practices reduce unnecessary administrative friction.

His approach is deliberately practical:

Don't automate everything. Automate what shouldn't require human attention.

Connect with Dr. Cham on LinkedIn to learn more.

Continue the Conversation

Healthcare innovation is not just about building smarter technology.

It is about asking better questions about the work surrounding patient care.

For more perspectives on healthcare operations, medical billing, AI, workflow automation and medical practice innovation, explore:

Visit the personal website
Listen to the podcast on Spotify
Subscribe and watch on YouTube
Follow updates on X (Twitter)
Follow on Facebook
Discover AI-powered medical billing solutions for busy physicians

· Connect professionally on LinkedIn

Knowledge creates leverage. Better questions create better healthcare. Start there.

Check the Featured section of my LinkedIn profile for a free resource available without an email signup.

And if this article made you rethink the relationship between physician attention, administrative burden and healthcare technology, consider reposting it so another physician or clinic owner can join the conversation.

Disclaimer

This article is intended for general educational and informational purposes only. It is not legal, medical, compliance, financial or professional advice. Healthcare organizations should consult appropriately qualified professionals regarding their specific clinical, legal, regulatory, privacy, billing and technology circumstances.

References

1. University of New Mexico Health Sciences — “From Cancer Patient to Healthcare Provider”
The August 28, 2026 story chronicles Tyler Tabor's journey from Hodgkin's lymphoma patient at UNM Hospital to first-year medical student and highlights his relationship with pediatric oncologist Jessica Valdez.

2. American Medical Association — Administrative Burdens
The AMA identifies administrative burden as a factor that can interfere with the physician-patient relationship and provides resources aimed at reducing unnecessary administrative work.

3. Healthcare IT News — Healthcare practices and AI automation
Recent reporting highlights growing interest in AI and automation to reduce manual administrative work, while also noting fragmented software environments and privacy concerns.

#Healthcare #MedicalBilling #HealthcareAI #Physicians #ClinicOwners #HealthcareInnovation #MedicalPracticeManagement #RevenueCycleManagement #WorkflowAutomation #HealthTech #PhysicianBurnout #PatientCare #ArtificialIntelligence #HealthcareLeadership #DigitalHealth #IndependentPractice #ResponsibleAI #HealthcareOperations #MedicalTechnology #OnnX

The future of healthcare isn't about putting more technology between physicians and patients.

It's about using technology to remove the things that shouldn't be between them in the first place.

The goal isn't to make physicians better administrators. It's to give them more room to be physicians.