From five hospital visits to a rare diagnosis, Evan's story exposes a deeper healthcare problem: patients experience one continuous journey, while information gets fragmented across the system.
“Safe care for life!” — World Health
Organization, World Patient Safety Day 2026
A sixth-grader, a mosquito, five hospital visits—and a
question healthcare should be uncomfortable answering
Evan was supposed to be starting sixth grade.
Instead, he ended up on a medical journey his mother, Summer
Rose, never expected.
It started with a really bad headache.
Then came a high fever.
Then his speech became so slurred that his mother said he
sounded as though he had suffered a stroke.
Then came seizures.
Summer took Evan to the hospital.
Then she took him again.
And again.
According to current reporting, Evan was taken to the
hospital five times over approximately three weeks before a final spinal
tap led to the diagnosis of La Crosse virus, a rare mosquito-borne
disease that can cause severe neurological illness in children.
The virus caused encephalitis and brain swelling. His
recovery may involve speech therapy and could take months or longer.
Summer is now speaking publicly because she does not want
another parent to watch a child go through what Evan experienced.
And that is where this story becomes bigger than one rare
virus.
Because there is a question hiding underneath Evan's story:
What happens when the patient experiences one continuous
story—but the healthcare system experiences five separate encounters?
That question bothers me.
Not because I think every difficult diagnosis is a systems
failure.
It isn't.
Medicine is hard.
Rare diseases are rare for a reason.
Symptoms overlap.
Patients do not arrive with diagnostic labels attached to
their foreheads.
Physicians make decisions with incomplete information.
And sometimes the correct diagnosis only becomes obvious
after time reveals the pattern.
That is medicine.
But there is another problem.
Information can exist—and still be effectively lost.
It can be buried in a note.
Trapped in a different system.
Recorded in a different format.
Duplicated across encounters.
Separated from the clinical context that gives it meaning.
Delayed until it is no longer useful.
Or simply never transformed into an actionable signal.
And when that happens, the system can begin making decisions
with an incomplete version of reality.
That is not just a clinical problem.
It is also an operational problem.
A documentation problem.
A workflow problem.
And, eventually, a revenue-cycle problem.
Because the same fragmentation that can make clinical
information harder to connect can also make downstream administrative
information harder to connect.
A physician documents the encounter.
A coder interprets it.
A biller translates it into a claim.
A clearinghouse processes it.
A payer applies its rules.
A denial comes back.
Then someone starts investigating what happened.
By that point, the organization may be trying to reconstruct
a story that should have been coherent from the beginning.
The claim becomes the crime scene.
But the real problem may have started much earlier.
And perhaps we have been looking at both problems backward.
Maybe healthcare does not primarily have a billing
problem.
Maybe it has an information problem that billing simply
makes visible.
The uncomfortable question: Did healthcare lose the
thread?
Let me ask the question differently.
Did anyone lose the information?
Maybe.
But that is not actually the most interesting question.
The more important question is:
Did the system lose the thread?
Because those are two very different things.
The information may still exist.
The hospital may have it.
The laboratory may have it.
The emergency department may have it.
The primary-care physician may have it.
The specialist may eventually have it.
The payer may have a completely different version of it.
The billing department may have yet another version.
The problem is that having information is not the same as
connecting information.
And healthcare has become exceptionally good at collecting
information.
We are much less consistently good at turning that
information into a continuous, actionable narrative.
The patient has one story. Healthcare has a thousand
boxes.
Think about what happens to a patient during a complicated
medical journey.
There is an appointment.
Then a note.
A laboratory result.
An imaging study.
A medication.
A referral.
A telephone call.
An emergency-department visit.
Another physician.
Another note.
Another diagnosis.
Another claim.
Another payer response.
Another authorization.
Another denial.
Another appeal.
Another payment.
From the patient's perspective, these are not separate
events.
They are one life.
But the healthcare system often represents them as separate
transactions.
That distinction matters.
A patient does not think:
“Today I am Encounter #4.”
The patient thinks:
“I have been sick for three weeks. Why isn't anyone
figuring this out?”
That is the difference between a transactional system
and a longitudinal story.
And it is one of the central problems healthcare must solve.
Information fragmentation is not always an absence of
data.
This is where I think we have misunderstood the healthcare
data problem.
We often say:
“We need more data.”
I'm not convinced.
Healthcare already produces an enormous amount of data.
The harder problem is making existing information:
available, connected, structured, contextual, timely, and
actionable.
A physician may have the information.
But is it visible at the exact moment it matters?
A biller may have the documentation.
But does the documentation clearly support the claim?
A practice may know that a payer repeatedly denies a
particular service.
But is that pattern being converted into an operational
rule?
A patient may have mentioned the same symptom three times.
But did anyone—or any system—recognize the pattern across
encounters?
These are not simply data-storage problems.
They are information-intelligence problems.
And that distinction changes how we should think about AI.
Healthcare Doesn't Have a Data Problem. It Has a Context
Problem.
This may be one of the most uncomfortable ideas in
healthcare technology.
We have spent enormous amounts of time asking:
How do we collect more information?
Perhaps we should be asking:
How do we preserve context as information moves?
Because information without context can be surprisingly
dangerous.
A diagnosis without the clinical reasoning behind it.
A laboratory result without the surrounding symptoms.
A claim without the documentation supporting medical
necessity.
A denial without understanding the upstream event that
created it.
A physician note without knowing what information the payer
will ultimately require.
A payer rule without translating it into something useful at
the point of workflow.
In each case, the information may technically exist.
But the meaning has become fragmented.
That is information friction.
And information friction creates downstream work.
Someone has to investigate.
Someone has to call.
Someone has to resend.
Someone has to clarify.
Someone has to appeal.
Someone has to reconstruct what should have been obvious.
And someone eventually pays for all that friction.
Usually the practice.
Sometimes the payer.
Sometimes the patient.
And ultimately, the healthcare system.
The Claim Is Often the Crime Scene—Not the Crime
This is where the connection to revenue-cycle management
becomes particularly interesting.
A claim gets denied.
What happens next?
The conventional response is usually:
Work the denial.
Open the claim.
Read the payer response.
Identify the denial code.
Correct something.
Resubmit.
Appeal if necessary.
Move to the next denial.
It is understandable.
The revenue-cycle team has to get paid.
But there is a problem with this model.
It treats the downstream symptom as the primary
problem.
What if the denial actually started much earlier?
Maybe the authorization was incomplete.
Maybe the patient's eligibility changed.
Maybe the diagnosis did not adequately support the service.
Maybe documentation lacked a critical element.
Maybe a payer-specific rule was missed.
Maybe the physician's documentation contained the necessary
information, but it was not structured in a way that survived the
clinical-to-claims translation.
Maybe the workflow itself created the error.
By the time the denial reaches the billing department, the
original cause may be several steps upstream.
So the billing team is asked to solve a problem that was created
somewhere else.
That is expensive.
And it is repetitive.
And it is largely preventable.
The denial is simply where the organization finally notices
the failure.
The claim is the crime scene.
But the crime may have happened upstream.
The Healthcare Version of “Garbage In, Garbage Out”
Artificial intelligence has made the phrase “garbage in,
garbage out” famous again.
But healthcare has a more complicated version.
It is not simply:
Bad data → bad output.
It is:
Incomplete information → fragmented context → uncertain
interpretation → downstream rework → financial leakage.
And sometimes:
Incomplete information → delayed recognition → delayed
action.
This is why simply adding AI to an existing workflow may not
solve the underlying problem.
If the workflow is broken, AI can make the broken workflow
faster.
If the information is poorly structured, AI can process
poorly structured information faster.
If the context is missing, AI cannot magically recreate
reality.
The question therefore should not be:
“Where can we put AI?”
The better question is:
“Where does information first become unreliable,
incomplete, ambiguous, or disconnected?”
That is where intelligence belongs.
Move Intelligence Upstream
This is the principle behind the next generation of
healthcare infrastructure.
Don't wait for the denial.
Don't wait for the rejected claim.
Don't wait for the appeal.
Don't wait for the physician to discover that documentation
was incomplete.
Don't wait for the biller to spend twenty minutes
reconstructing what happened.
Move intelligence upstream.
At the point where information is created.
At the point where clinical documentation is generated.
At the point where coding decisions begin.
At the point where payer requirements can still influence
the workflow.
At the point where an error is still cheap to fix.
This is fundamentally different from simply automating
billing.
It is about creating a clinical-to-claims intelligence
layer.
The goal is not to make the billing department better at
cleaning up mistakes.
The goal is to make fewer mistakes reach the billing
department in the first place.
That distinction may sound subtle.
It isn't.
It changes the architecture.
Instead of:
Encounter → Claim → Denial → Investigation → Rework →
Appeal
Imagine:
Encounter → Intelligence → Validation → Clean Claim →
Payment
That is a very different healthcare operating model.
But There Is a Problem With AI
Here is where I want to challenge another assumption.
Healthcare does not need AI everywhere.
Healthcare needs trustworthy intelligence in the right
places.
An AI system that generates a recommendation nobody
understands will eventually become another source of friction.
A physician should be able to ask:
Why did the system flag this?
A coder should be able to ask:
What documentation is missing?
A biller should be able to ask:
Why is this claim likely to be denied?
A practice owner should be able to ask:
Where are our preventable revenue leaks coming from?
And the system should be able to answer.
Not with:
“Because the algorithm said so.”
But with something closer to:
“Here is the information we found. Here is the rule or
pattern involved. Here is what appears inconsistent. Here is what you can
verify before proceeding.”
That is the difference between automation and decision
support.
And in healthcare, that distinction matters.
The Black Box Problem Is Not Going Away
Healthcare professionals have spent decades learning to
question information.
That is part of medicine.
Why is this diagnosis being considered?
What evidence supports it?
What are the alternatives?
What changed?
What did we miss?
AI should not eliminate those questions.
It should make them easier to answer.
The future of healthcare AI cannot simply be:
“Trust the model.”
It should be:
“Here is what the model sees. Here is why it matters.
Here is what you should verify.”
That requires explainability.
It requires auditability.
It requires human oversight.
And it requires systems designed around the actual workflow
of physicians, administrators, coders, billers, and patients.
The objective should never be to replace clinical judgment
with an opaque machine.
The objective should be to give professionals better
information at the moment decisions are made.
What Evan's Story Teaches Us About Information
We should be careful here.
I do not know whether information from Evan's earlier
encounters could have changed his ultimate outcome.
I do not know whether an earlier diagnosis was realistically
possible.
I do not know what each clinician knew at each moment.
And neither should anyone pretend to know those things from
a news article.
But we can still learn something important.
A healthcare journey is not a series of isolated
transactions.
It is a continuous story.
And when that story becomes fragmented, the system becomes
more dependent on individual people remembering, recognizing, reconciling, and
reconstructing information.
That does not scale.
And it creates vulnerability.
For patients.
For physicians.
For hospitals.
For practice administrators.
For billing teams.
And for the financial health of medical practices.
This is why I believe the next generation of healthcare
infrastructure should focus less on simply moving data and more on preserving
meaning as data moves.
Interoperability matters.
But interoperability alone is not enough.
Data exchange matters.
But data exchange alone is not enough.
AI matters.
But AI alone is not enough.
The real objective is:
Right information.
Right context.
Right moment.
Right action.
That is where healthcare begins to move from reactive
systems toward intelligent systems.
A Simple Experiment Every Practice Can Run
You do not need to build an AI platform tomorrow.
Start with 100 claims.
Take the last 100 denials from your practice.
Do not simply count them.
Investigate them.
For each denial, ask five questions:
1. What was the immediate reason for the denial?
Was it eligibility?
Authorization?
Coding?
Documentation?
Medical necessity?
Payer policy?
Timeliness?
Something else?
2. Where did the problem actually originate?
At billing?
At coding?
During documentation?
During scheduling?
During authorization?
At the point of care?
3. When could the problem first have been detected?
This is the question most organizations skip.
Because the goal should not simply be to determine when
the denial was discovered.
The goal is to determine when the denial became
preventable.
4. Could the problem have been detected before the claim
was submitted?
If yes, you have identified an opportunity for upstream
intelligence.
5. Can the organization prevent the same problem next
week?
If the answer is no, you have identified a systems problem
rather than an individual error.
That is the difference between denial management and denial
prevention.
Measure Prevention, Not Just Recovery
Most revenue-cycle dashboards tell you how much work was
done.
How many claims were submitted.
How many were denied.
How many were appealed.
How much was collected.
How many days in A/R.
Those numbers matter.
But they tell you what happened after the system produced
the problem.
Add another layer.
Measure:
Clean-claim rate
Preventable denial rate
First-pass resolution
Manual touches per claim
Rework hours
Documentation-related denials
Authorization-related denials
Average time from error creation to error detection
That last metric is particularly interesting.
Because the shorter the interval between error creation
and error detection, the cheaper the correction usually becomes.
The ultimate goal is not:
“How quickly can we fix this?”
It is:
“How early can we detect this?”
And eventually:
“How often can we prevent it?”
What Healthcare Should Stop Doing
We should stop treating every downstream problem as a
downstream problem.
Stop asking the billing department to solve documentation
problems.
Stop asking coders to reverse-engineer clinical intent after
the encounter.
Stop asking practice administrators to manually remember
payer-specific rules.
Stop treating denials as inevitable.
Stop measuring AI success by how much work it automates
without asking whether it is preventing work altogether.
And stop assuming that more data automatically means better
healthcare.
Sometimes the answer is not more information.
Sometimes the answer is better-connected information.
What Healthcare Should Start Doing
Start identifying where information first breaks.
Start connecting clinical and operational workflows.
Start converting recurring patterns into actionable rules.
Start using AI to surface missing information before it
becomes expensive.
Start giving physicians explanations instead of unexplained
alerts.
Start giving practice owners visibility into the upstream
causes of revenue leakage.
Start designing workflows around prevention rather than
repair.
And start asking a different question about every recurring
problem:
“How far upstream can we move the intervention?”
That may be one of the most important questions healthcare
technology can ask.
The Future Is Not More Automation. It Is Better Timing.
We often talk about the future of healthcare as though the
answer is simply more automation.
More AI agents.
More software.
More dashboards.
More integrations.
More algorithms.
I think that misses something fundamental.
Timing matters.
An alert after the claim has already been denied is useful.
An alert before the claim is submitted is better.
A documentation suggestion after the patient has left is
useful.
A documentation prompt while the clinical story is still
being captured is better.
A denial report at the end of the month is useful.
A system that identifies the pattern before the denial
occurs is better.
That is what moving intelligence upstream means.
It is not about doing more.
It is about knowing sooner.
From Reactive Billing to Deterministic Revenue
This is ultimately the opportunity.
Today's healthcare revenue cycle is often reactive.
Something goes wrong.
Someone investigates.
Someone fixes it.
Someone resubmits it.
Someone appeals it.
Someone waits.
Someone follows up.
Someone gets paid.
Then the cycle starts again.
The alternative is a system that becomes increasingly predictive,
preventative, and deterministic.
The encounter creates structured information.
The system validates it.
Potential problems are surfaced.
The physician or staff member resolves them while the
information is still fresh.
The claim is generated with fewer unknowns.
The payer receives a cleaner representation of the clinical
event.
And the organization learns from every transaction.
That creates a feedback loop.
Encounter → Intelligence → Validation → Clean Claim →
Payment → Learning
Every completed claim makes the next claim smarter.
That is a fundamentally different philosophy from simply
hiring more people to work more denials.
The Bigger Lesson
Evan's story began with something frighteningly simple.
A headache.
A fever.
Slurred speech.
Seizures.
Five hospital visits.
A rare diagnosis.
A family trying to understand what was happening.
We should not turn that story into a simplistic indictment
of physicians or hospitals.
That would be unfair.
Instead, we should use it to ask a harder systems question.
How do we make sure the healthcare system remembers the
story while the story is still unfolding?
That question reaches far beyond rare diseases.
It reaches into every part of healthcare.
Diagnosis.
Documentation.
Care coordination.
Prior authorization.
Coding.
Claims.
Denials.
Payments.
Patient safety.
And artificial intelligence.
The technology may change.
The underlying challenge will not.
Information must become useful before it becomes
expensive.
That may be the real frontier.
Not more data.
Not more software.
Not more automation.
Better intelligence, earlier in the workflow.
A Note to Physicians, Practice Owners, Administrators and
Billers
If you run a medical practice, you already know where the
friction lives.
You see it every week.
The missing authorization.
The incomplete note.
The payer rule nobody knew had changed.
The claim that should have been clean.
The denial that takes 30 minutes to understand.
The staff member who spends hours chasing information that
should have been available immediately.
The physician who finishes seeing patients and then spends
part of the evening completing documentation.
These problems are often treated as isolated annoyances.
They are not.
They are signals.
Every repeated administrative problem is telling you
something about the architecture of your workflow.
The question is whether you are listening.
My Challenge to Healthcare
Take one recurring problem in your practice.
Not ten.
One.
Find the moment it begins.
Find the moment it becomes detectable.
Find the moment it becomes expensive.
Then ask:
Why are we waiting until this point to act?
That question may reveal more about your practice than
another dashboard ever will.
Because the future of healthcare revenue is not going to be
won by the organization that becomes best at cleaning up yesterday's mistakes.
It will be won by the organization that becomes best at preventing
tomorrow's mistakes.
Expert Perspective: What This Means for Patient Safety
The broader patient-safety movement is increasingly
emphasizing the importance of safer systems rather than relying exclusively on
individual vigilance.
The World Health Organization's 2026 World Patient Safety
Day campaign uses a simple but powerful message:
“Safe care for life!”
That framing is important.
Patient safety is not merely about whether one person made
one mistake.
It is about whether the system consistently creates
conditions for safe care.
The same systems thinking applies to healthcare operations.
If the same denial happens repeatedly, it may not be a
staff-performance problem.
It may be a workflow-design problem.
If the same documentation gap repeatedly creates rework, it
may not be a physician-compliance problem.
It may be a feedback problem.
If the same payer rule repeatedly surprises the practice, it
may not be a billing problem.
It may be an information-timing problem.
Fix the system, and you reduce dependence on heroics.
Myth Buster
Myth: “We just need more staff to work the denials.”
Sometimes additional staff are necessary.
But more people working a broken process can simply create more
expensive rework.
The better question is:
How many of those denials could have been prevented
before they ever reached the billing queue?
Myth: “Interoperability solves the information problem.”
Interoperability helps systems exchange information.
But exchanging information is not the same as understanding
it.
The next challenge is turning exchanged data into contextual,
actionable intelligence.
Myth: “AI will solve the administrative burden.”
AI can reduce burden.
But only if it is applied to the right problem.
Automating a poorly designed workflow can accelerate the
wrong behavior.
Myth: “Every denial is a billing problem.”
No.
Some denials originate upstream—in documentation,
authorization, eligibility, coding, payer policy, workflow, or
clinical-to-claims translation.
The billing department may simply be the first place where
the financial consequence becomes visible.
Frequently Asked Questions
Is Evan's story evidence of medical negligence?
No conclusion like that should be drawn from the publicly
reported story alone.
The fact that a rare disease was diagnosed only after
multiple healthcare encounters does not, by itself, establish negligence or
malpractice.
Could better information have changed Evan's outcome?
We do not know.
That would require a detailed review of the actual medical
records, clinical findings, timing, differential diagnoses, testing, and
treatment decisions.
The point of using the story here is not to claim a
particular clinical outcome.
It is to examine the broader problem of information
continuity.
What does this have to do with medical billing?
More than it initially appears.
A claim is ultimately a structured financial representation
of a clinical encounter.
If critical information is incomplete, inconsistent,
delayed, or poorly translated upstream, the downstream claim can become
vulnerable.
What does “move intelligence upstream” mean?
It means identifying problems before they become
denials, rework, appeals, delayed payments, or operational failures.
The earlier an error can be detected, the more opportunities
exist to correct it.
Does that mean AI should replace physicians or billers?
No.
The goal should be to augment professionals with better
information, better timing, explanations, and workflow support.
Human judgment remains essential.
Get Involved
If you are a physician, practice owner, administrator,
coder, biller, healthcare executive, or healthcare technology builder, I
want to hear from you.
Forget the buzzwords for a moment.
Forget AI.
Forget automation.
Forget digital transformation.
Think about your actual workday.
If you could eliminate ONE repetitive administrative
problem from your practice tomorrow, what would it be?
Prior authorization?
Documentation?
Coding?
Denials?
Eligibility?
Claim status?
Payer rules?
Something else?
Tell me in the comments.
Your answer may reveal where healthcare's next great
opportunity actually lies.
And if this perspective resonates with someone running a
medical practice, repost this article and start the conversation with
them.
The goal is not to sell another piece of software.
The goal is to identify the friction worth eliminating.
Final Thoughts
Stop measuring how quickly you repair the problem. Start
measuring how often you prevent it.
Stop asking where you can add AI. Start asking where
information first becomes unreliable.
And stop accepting fragmented information as the
inevitable cost of modern healthcare.
The patient has one story.
Healthcare needs to learn how to keep the thread.
About the Author
Dr. Daniel Cham is a physician, healthcare technology
entrepreneur, and medical consultant focused on the intersection of medicine,
artificial intelligence, healthcare operations, and revenue-cycle
infrastructure.
His work examines a fundamental question:
How can healthcare organizations move from reactive
processes to intelligent systems that prevent problems before they become
expensive?
He is the founder of OnnX, an AI-powered healthcare
revenue-cycle platform focused on helping physician-owned practices move
intelligence upstream—from claims and denials toward the point where clinical
and operational information is first created.
Connect with Dr.
Daniel Cham on LinkedIn
Disclaimer
This article is intended for general educational and
informational purposes only. It does not constitute medical, legal,
medical-malpractice, billing, coding, financial, or professional advice.
The discussion of Evan and his family is based on
publicly reported information and is not an independent medical or legal
assessment of his care. No conclusion regarding negligence, malpractice,
causation, or preventability should be inferred from this article.
Individual clinical, billing, coding, legal, and
operational questions should be evaluated by appropriately qualified
professionals with access to the relevant records and circumstances.
Continue the Conversation
LinkedIn: Follow the conversation on healthcare AI,
physician entrepreneurship, revenue-cycle transformation, and the future of
healthcare infrastructure.
Featured Resource: Watch for my upcoming practical
resources on identifying and preventing upstream revenue-cycle problems in
physician practices.
The question remains:
What information problem in healthcare would you fix
first if you had the power to redesign the workflow?
Continue the Conversation
Explore practical insights, evidence-based strategies,
and behind-the-scenes perspectives that help physicians and clinic leaders
navigate complex challenges.
- 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 drives progress — start your journey today.
References
1. People — Evan’s La Crosse virus story
Ingrid Vasquez, People, September 9, 2026. This is the primary source
for the reported facts about Evan, his mother Summer Rose, the five hospital
visits, initial laryngitis diagnosis, spinal tap, La Crosse virus,
encephalitis, and neurological complications.
People — Ohio 6th Grader Diagnosed with Rare Mosquito-Borne
Illness
2. World Health Organization — World Patient Safety Day
2026
WHO, World Patient Safety Day 2026: Safe care for noncommunicable diseases.
The 2026 campaign specifically highlights fragmented care, risks across
the continuum of care, and the need for stronger, integrated health
systems. Its slogan is “Safe care for life!”
WHO — World Patient Safety Day 2026
3. Agency for Healthcare Research and Quality —
Diagnostic Documentation & Information
AHRQ, Documenting Diagnosis: Exploring the Impact of Electronic Health
Records on Diagnostic Safety. AHRQ notes that accurate, complete
documentation improves access to meaningful clinical information,
communication, coordination, and decision support, while its diagnostic-safety
research identifies communication barriers, lack of coordination,
information overload, interoperability/usability problems, and missing
follow-up as contributors to diagnostic-process breakdowns.
AHRQ — Diagnostic Safety and Quality Resources
These three are sufficient for the article. I would
not overload the piece with 10–20 references; the three
#Healthcare #HealthcareAI #MedicalBilling
#RevenueCycleManagement #RCM #PatientSafety #DiagnosticSafety #HealthTech
#ArtificialIntelligence #PhysicianEntrepreneur #HealthcareInnovation
#MedicalTechnology #PhysicianOwnedPractice #ClinicalDocumentation #MedicalCoding
#DenialManagement #HealthcareOperations #HealthcareLeadership #OnnX
