Healthcare can preserve a human heart outside the body. Why are we still struggling to preserve the context surrounding the human being who receives it?
“Physicians evaluate patients in context, drawing on
years of training and experience.” — American
Medical Association and leading physician organizations
Dr. Sanjana Kochhar once knew she was dying. A donor heart gave her another chance—and nearly four years later, that second chance became motherhood. Her story is a powerful reminder that behind every diagnosis, claim, and medical record is a human life waiting to continue.
There is something almost absurd about modern healthcare.
We can preserve a human heart outside the human body.
We can transport it.
Monitor it.
Keep it beating.
Put it inside another human being.
And save a life.
Yet somewhere between the exam room, the authorization
portal, the EHR, the billing system, the payer, and the claim…
we can lose the story of the patient.
That is the contradiction I can't stop thinking about.
And Dr. Sanjana Kochhar's story makes it impossible
to ignore.
Dr. Sanjana Kochhar Was Dying
In October 2022, Dr. Sanjana Kochhar was 29.
She was a physician.
She understood medicine.
She understood what it meant when the heart was failing.
But understanding medicine doesn't make you immune to
becoming its patient.
Kochhar had experienced symptoms for years.
She had palpitations.
She became short of breath.
Eventually, her heart deteriorated severely.
Then everything accelerated.
Her ejection fraction fell to 12 percent.
Her lungs filled with fluid.
Her kidneys and liver stopped working.
She became critically ill.
She was transferred to the transplant centre at Freeman
Hospital in Newcastle and placed on the urgent heart transplant list on
October 31, 2022.
She later described the experience in brutally simple terms:
She was running out of options.
Then came the heart.
Four weeks later, she received a donor heart.
Professor Stephen Clark and the transplant team at
Freeman Hospital performed the transplant.
Kochhar was also involved in a clinical study using the XVIVO
Heartbox, a system designed to preserve donor hearts during transportation.
She became the first patient in the United Kingdom to
receive a donor heart transported using the Heartbox.
The transplant worked.
Her body began functioning again.
She was discharged on Christmas Eve.
And then something happened that medicine cannot put neatly
into a billing code.
She got her life back.
Then Life Became Bigger
Kochhar returned to medical training.
She qualified as a GP in July 2024.
She began working as a GP.
She met members of her donor's family.
She married her partner, Paul.
And in August 2026, nearly four years after her transplant,
she welcomed her first child.
Think about that timeline.
2022: multiple organ failure.
2022: emergency heart transplant.
2024: qualified as a GP.
2026: became a mother.
The woman who once knew she was dying was now holding a
baby.
That is an extraordinary medical story.
But I think there is another story hiding inside it.
And it has almost nothing to do with transplantation.
It has to do with context.
The Patient Is Not the Chart
Healthcare loves labels.
Diagnosis.
Procedure.
Encounter.
Code.
Authorization.
Claim.
Denial.
Appeal.
Payment.
Useful labels.
Necessary labels.
But dangerous when we forget what they represent.
A patient is not a diagnosis.
A diagnosis is not an encounter.
An encounter is not a code.
A code is not a claim.
A claim is not the care.
And a denial is certainly not the whole explanation.
These are representations of reality.
They are not reality itself.
That's an important distinction.
Because when healthcare moves information from one system to
another, the facts may survive while the context disappears.
And when context disappears, humans have to reconstruct it.
Healthcare's Most Expensive Integration Layer
Here is my slightly uncomfortable theory:
The most expensive middleware in healthcare may be the
human being.
Think about what happens inside a typical practice.
The front desk checks eligibility.
The authorization specialist checks benefits.
The nurse checks the referral.
The physician checks the chart.
The biller checks the authorization.
The biller checks the claim.
Someone checks the payer portal.
Someone sends an email.
Someone calls the payer.
Someone leaves a voicemail.
Someone waits.
Someone calls again.
Someone says:
“I thought you already sent that.”
And someone else says:
“We did.”
There is a moment of silence.
Then somebody opens another browser tab.
Congratulations.
You've discovered healthcare interoperability.
It works perfectly—as long as humans are willing to be
the API.
Here's the Contrarian Part
The healthcare industry spends enormous energy talking about
denials.
I'm not convinced the denial is the real problem.
The denial is often the last visible symptom.
The interesting question is:
Where did the information failure actually begin?
Was eligibility wrong?
Was it never verified?
Was authorization required?
Was authorization obtained?
Was the authorization connected to the correct encounter?
Was the referral valid?
Was the benefit limitation known?
Did the clinical documentation support the service?
Did the information survive the handoff?
Did someone enter the wrong data?
Did the payer receive something different from what the
physician believed had been submitted?
By the time the claim is denied, the original mistake may be
weeks old.
The billing department simply gets the bill.
Literally.
The Denial Isn't the Problem. It's the Clue.
This is where I think healthcare needs a mental shift.
Stop asking only:
“How do we fix this denial?”
Ask:
“Why did this denial become possible?”
Those are completely different questions.
The first produces a corrected claim.
The second can produce a better system.
One is reactive.
The other is preventive.
One repairs the symptom.
The other searches for the cause.
And if the same denial happens 100 times, fixing 100 claims
is not process improvement.
It's cardio.
For the billing department.
We Don't Have a Data Shortage
Healthcare has data everywhere.
We have:
EHR data.
Claims data.
Eligibility data.
Authorization data.
Referral data.
Clinical notes.
Lab results.
Imaging.
Medication records.
Scheduling data.
Payment data.
Denial data.
Appeal data.
And now AI-generated data.
If anything, we're approaching a point where healthcare may
have the opposite problem.
We have too much information and too little continuity.
The question isn't:
“Do we have the data?”
The better question is:
“Does the right information arrive at the right place
with enough context to be useful?”
That's a much harder problem.
This Week's Warning From Washington
This isn't merely a philosophical argument.
On September 30, 2026, the Office of the National
Coordinator for Health Information Technology published new analysis of
administrative burden among more than 8,400 family physicians.
The findings are striking.
More than three-quarters experienced at least one
substantial burden involving external information retrieval, prior
authorization, or after-hours documentation.
And 12.2% experienced all three in 2026.
There was some progress.
Substantial after-hours documentation burden declined from
41% in 2024 to 34% in 2026.
The burden of tracking down outside information declined
from 43% to 38%.
But prior authorization moved in the opposite direction.
It increased from 54% to 58% over the same period.
That tells us something important.
Better interoperability can help.
But merely moving information electronically doesn't
automatically solve the workflow.
An electronic mess is still a mess.
It just loads faster.
The AI Industry Has Another Problem
Now we arrive at AI.
AI can summarize.
AI can classify.
AI can predict.
AI can draft.
AI can search.
AI can automate.
And physicians are increasingly using it.
The AMA's 2026 physician survey found that more than 80%
of physicians report using AI professionally.
That's a major shift.
But here's the question nobody should skip:
What happens when AI gets the wrong context?
It doesn't necessarily fail dramatically.
That's what makes it dangerous.
It may produce a beautiful answer.
A perfectly formatted summary.
A highly confident recommendation.
A neatly categorized claim.
A very impressive mistake.
AI doesn't eliminate bad context.
Sometimes it just processes bad context faster.
That's not intelligence.
That's high-speed confusion.
The AMA Just Put the Core Issue in Plain English
The AMA and five major physician organizations issued a
statement this week warning against treating AI as a replacement for physician
expertise.
Their point was direct:
“Physicians evaluate patients in context…”
That word matters.
Context.
Not just information.
Not just data.
Not just documentation.
Context.
Physicians combine history, examination, circumstances,
experience and judgment.
That is what makes the patient more than a collection of
fields in a database.
And this principle shouldn't stop at the exam room door.
It applies to the administrative side of healthcare too.
Context Should Follow the Patient
Imagine a patient needs a procedure.
At the scheduling stage, the system knows the appointment.
At eligibility verification, another system knows the
coverage.
At authorization, another system knows the authorization
requirement.
At the clinical encounter, the physician knows the medical
reason.
At coding, another person interprets the documentation.
At billing, someone assembles the claim.
At adjudication, the payer evaluates the submission.
Now ask:
Does one system understand the entire chain?
Usually not.
Pieces are everywhere.
The patient is the only thing connecting them.
And somehow we expect the patient to keep everything
straight.
The patient has become the interoperability layer.
That's not a technology strategy.
That's a hostage situation.
Dr. Kochhar's Story Makes This Human
Now go back to Sanjana.
You could summarize her story in a database:
29-year-old female.
Heart failure.
Multiple organ failure.
Heart transplant.
Recovery.
GP.
Mother.
Technically useful.
But it doesn't capture the story.
She had symptoms while training to become a doctor.
She continued her medical education while living with a
failing heart.
Her condition suddenly deteriorated.
Her heart function fell dramatically.
Her lungs filled with fluid.
Her kidneys and liver stopped working.
She waited for a donor heart.
She survived.
She struggled emotionally with the fact that another person
had died for her to receive the organ.
She eventually connected with members of the donor's family.
She returned to medicine.
She built a future.
She became a mother.
Those are not separate data points.
They are connected events.
The connection is the context.
And That's What Healthcare Keeps Losing
Healthcare doesn't necessarily lose data.
It loses relationships between data.
The authorization exists.
The clinical note exists.
The referral exists.
The payer requirement exists.
The procedure exists.
The claim exists.
But does the system understand how they relate?
That's the question.
Because a database can tell you:
A happened.
And another database can tell you:
B happened.
But healthcare often needs to know:
A caused B.
Or:
A was required before B.
Or:
B could not happen without A.
Or:
A changed after B was scheduled.
That's context.
Stop Building Bigger Buckets
Healthcare has become incredibly sophisticated at cleaning
up problems after they happen.
Denial management.
Appeals.
Work queues.
Exception queues.
Escalation queues.
Follow-up queues.
Correction queues.
You can build an entire organization around fixing problems.
And sometimes we do.
But consider the roof analogy.
If water is dripping through the ceiling, you can put down a
bucket.
That's useful.
You can put down ten buckets.
You can hire someone whose entire job is managing buckets.
You can buy bucket analytics.
You can create a bucket dashboard.
You can use AI to predict which bucket will overflow next.
Eventually someone should probably ask:
“Why don't we fix the roof?”
Healthcare may have a similar problem.
The Real Question Isn't “Where Can We Add AI?”
It's:
Where does the information first become unreliable?
That is a much better AI question.
Maybe it's scheduling.
Maybe eligibility.
Maybe authorization.
Maybe referral management.
Maybe documentation.
Maybe coding.
Maybe claims.
Maybe payer communication.
Maybe a handoff between two systems.
Find the first break.
Then ask:
Can we prevent it?
That's upstream thinking.
The OnnX Thesis
This is the thinking behind my work with OnnX.
The central thesis is simple:
Healthcare billing is a data-quality problem before it is
a tooling problem.
That doesn't mean software isn't important.
It means software cannot compensate indefinitely for missing
context.
The goal should not simply be:
Submit → Deny → Appeal → Repeat.
The goal should be:
Capture → Structure → Preserve → Propagate → Act → Learn.
That's a fundamentally different model.
Capture
Get the important information early.
Don't wait until the claim is rejected to discover something
that was knowable before the appointment.
Structure
Turn information into usable context.
A document is not necessarily structured information.
A note can contain the answer while still being difficult
for another system to use.
Preserve
Don't allow context to disappear during handoffs.
If the authorization was obtained for a particular reason,
that context should not vanish when the encounter moves downstream.
Propagate
Relevant information should travel with the workflow.
Not necessarily every piece of information.
The right information.
At the right time.
Act
Identify missing or conflicting information before it
becomes expensive.
That is where prevention begins.
Learn
When something fails, don't just correct it.
Learn from it.
A denial should create knowledge.
Not merely more work.
More AI Isn't Automatically Better
Here's another contrarian position:
The smartest AI system may be the one that knows when it
doesn't have enough context.
Imagine an AI billing system that doesn't confidently
produce an answer.
Instead it says:
“Authorization information is incomplete. The encounter
should not proceed to claim submission.”
That's not a failure.
That's intelligence.
A system that catches uncertainty upstream can be more
valuable than one that generates beautiful answers downstream.
Automation Shouldn't Mean “No Humans”
This is another misconception.
The goal isn't to eliminate people.
Healthcare still needs people.
Physicians need judgment.
Nurses need judgment.
Billers need judgment.
Practice managers need judgment.
Patients need someone who can explain what is happening.
The objective is different:
Stop spending human intelligence on work that doesn't
require human intelligence.
Let machines reconcile.
Let humans judge.
Let machines detect missing fields.
Let humans handle exceptions.
Let machines track relationships.
Let humans communicate.
That is augmentation.
Not replacement.
The Physician Should Not Become the Billing System
Physicians already spend too much time navigating
administrative systems.
The AMA's 2026 prior authorization survey found that
physicians complete an average of 40 prior authorizations per week,
consuming about 13 hours of physician and staff time weekly.
The same survey found:
95% said prior authorization delays necessary care.
94% said it contributes to burnout.
79% said patients sometimes abandon treatment because
of authorization challenges.
And 26% reported a serious adverse event associated
with prior authorization.
These are survey findings, not proof that every
administrative delay causes a particular clinical outcome.
But they reveal something important:
Administrative friction is not separate from clinical
care.
It surrounds it.
What Clinic Owners Should Measure
Most practices know their revenue.
Many know their denial rate.
Some know their days in accounts receivable.
Fewer know how much human reconstruction happens
between the encounter and the payment.
I would measure:
Clean-claim rate
First-pass payment rate
Denial rate
Denial root cause
Authorization turnaround time
Eligibility exceptions
Manual touches per claim
Staff minutes per claim
Days in accounts receivable
Appeal overturn rate
And one additional metric:
Context Reconstruction Events
How many times does somebody have to stop and ask:
“What happened here?”
That may be one of the most expensive questions in
healthcare.
A Seven-Day Experiment for a Clinic
You don't need a massive transformation project.
Try this.
Day 1 — Pull 20 denials.
Don't fix them yet.
Just categorize them.
Day 2 — Find the largest category.
Don't blame anyone.
Follow the information.
Day 3 — Trace three cases backward.
Find where the problem actually began.
Day 4 — Identify the missing context.
What did someone need to know?
Day 5 — Create one prevention rule.
Something simple.
Something measurable.
Day 6 — Measure human intervention.
How many calls?
How many messages?
How many minutes?
How many handoffs?
Day 7 — Ask the uncomfortable question.
Could technology have prevented this instead of merely
helping someone repair it?
That's where the real opportunity may be.
Five Questions Every Clinic Owner Should Ask
1. Where do our denials actually begin?
Not where we discover them.
Where do they begin?
2. How many times are we entering the same information?
Once?
Twice?
Five times?
If nobody knows, that's already useful information.
3. How often does someone ask, “What happened?”
Track it.
4. Which problems keep repeating?
Repetition is a signal.
5. What information should follow the patient
automatically?
That question can expose entire categories of workflow
failure.
Five Myths Worth Killing
Myth #1: “Denials are a billing problem.”
Not necessarily.
The billing department may simply be where an upstream
problem becomes visible.
Myth #2: “More documentation means better information.”
Not necessarily.
More information can mean more noise.
The goal is usable context.
Myth #3: “AI will eliminate administrative work.”
AI can automate tasks.
It cannot magically repair poorly structured workflows.
Myth #4: “Electronic means interoperable.”
It doesn't.
An electronic document sitting in the wrong system is still
difficult to use.
Myth #5: “The EHR contains the patient.”
It contains a representation of the patient.
That's different.
The Future Isn't a Bigger Billing Department
I don't think the future of medical billing should be:
more people → more queues → more software → more AI →
fewer denials.
That's incrementalism.
The more interesting possibility is:
better context → earlier intervention → fewer preventable
failures → fewer manual touches.
That changes the economics.
It changes the workflow.
And perhaps most importantly, it changes what humans spend
their time doing.
The Canonical Context Record
Imagine an encounter with a continuously updated context
record.
Not merely:
Patient.
Diagnosis.
Procedure.
Code.
Instead:
Coverage
Eligibility
Authorization
Referral
Benefit limits
Clinical context
Documentation
Procedure
Coding
Payer requirements
Submission
Payment
Denial
Appeal
And the relationships between them.
Now the system isn't just storing facts.
It understands how those facts connect.
That's the difference between a data warehouse and a story.
Why Independent Practices Should Care
Large health systems can sometimes absorb administrative
complexity.
Independent practices have less room for waste.
One recurring authorization problem.
One employee spending hours chasing information.
One physician staying late to finish paperwork.
One recurring denial category.
One patient abandoning treatment because of administrative
friction.
Individually, each seems manageable.
Together, they become infrastructure.
And infrastructure determines whether a practice feels lean
or exhausted.
Why Physicians Should Care
Physicians shouldn't have to become billing experts.
But physicians should understand what happens to their
clinical information after they sign the note.
Because their reasoning becomes:
Documentation.
Coding.
Authorization evidence.
Claim data.
Quality data.
Payment data.
Sometimes denial evidence.
The further that information travels, the greater the risk
that meaning gets lost.
The better the system preserves context, the less someone
downstream has to reinterpret the physician's work.
Why Healthcare Founders Should Care
Healthcare founders love technology.
Naturally.
Technology is exciting.
Workflow is messy.
Unfortunately, workflow is where the money is.
The best question isn't:
“What AI feature can we add?”
Try:
“What human reconstruction are we eliminating?”
That's a much harder question.
And potentially a much more valuable one.
The Human Story Is the Point
Return to Sanjana Kochhar.
A database can tell us:
29-year-old physician.
Heart failure.
Multiple organ failure.
Transplant.
Recovery.
GP.
Mother.
But that's not her story.
Her story is that a young doctor became critically ill.
That she understood enough medicine to know how serious it
was.
That she was running out of options.
That a donor heart became available.
That a transplant gave her another chance.
That she had to emotionally process the death of another
person whose organ saved her.
That she connected with members of that donor's family.
That she returned to medicine.
That she built a future.
And now she has a daughter.
Her story cannot be reduced to a claim.
And neither can yours.
Neither can mine.
Neither can the patient sitting in an examination room this
morning.
Maybe Healthcare Has Been Asking the Wrong Question
For years we've asked:
How do we collect more data?
Then:
How do we store more data?
Then:
How do we analyze more data?
Now:
How do we put AI on top of all that data?
Maybe the next question should be:
How do we keep the meaning intact?
That's different.
Because meaning requires context.
Context requires relationships.
Relationships require continuity.
And continuity requires systems that remember what happened
before.
Stop Optimizing the Mess
Healthcare has become very good at optimizing around
complexity.
We have consultants for it.
Software for it.
Dashboards for it.
Work queues for it.
Call centers for it.
AI for it.
And increasingly sophisticated ways to measure it.
But here's the provocative question:
What if we stopped optimizing the mess and started
removing the reasons the mess exists?
That's the shift.
From:
Denial → prevention
From:
Transaction → context
From:
Reconciliation → continuity
From:
More data → better-connected data
From:
AI everywhere → AI where it actually helps
From:
Human middleware → human judgment
One Last Look at Sanjana
Dr. Sanjana Kochhar's story is extraordinary because
medicine gave her something priceless:
time.
Time to recover.
Time to finish her training.
Time to become a GP.
Time to marry Paul.
Time to meet people connected to the donor who saved her
life.
Time to become a mother.
Healthcare exists to create more moments like those.
Not more claims.
Not more dashboards.
Not more work queues.
Not more administrative archaeology.
More life.
That's the point.
The Question I Want to Leave With You
If a healthcare system can preserve a human heart outside
the body...
why can't it preserve the context surrounding the human
being who receives it?
Maybe the future of healthcare isn't about collecting more
information.
Maybe it is about making sure the information we already
have doesn't lose its meaning.
Because when context disappears, somebody has to rebuild it.
Usually a human.
Usually manually.
Usually at the worst possible time.
And sometimes the person rebuilding the story is a physician
who should be seeing the next patient.
So perhaps the real innovation isn't another dashboard.
Or another AI model.
Or another RCM queue.
Perhaps it's something much simpler:
Keep the story connected.
Because the patient was never the chart.
The diagnosis was never the person.
The code was never the encounter.
The claim was never the care.
And the denial?
Maybe it was just the clue.
Your Turn
What is the one administrative problem in your practice that
everyone has learned to tolerate—but nobody has actually solved?
That's the conversation I want to have.
Share your experience in the comments.
And if this made you rethink how healthcare handles
information, repost it.
Another physician or clinic owner may be fighting the exact
same invisible problem.
Knowledge creates leverage.
Context creates clarity.
And better systems give clinicians more time for people.
Free Resource
Want more practical ideas about healthcare operations,
medical billing, healthcare technology and physician entrepreneurship?
Check the Featured section of my LinkedIn profile for a
free resource. No signup required.
Take what is useful.
Challenge what isn't.
Then bring the conversation back to your practice.
Listen to the
podcast on Spotify
Subscribe and
watch on YouTube
About the Author
Dr. Daniel Cham is a physician, medical consultant
and healthcare entrepreneur with experience across medical technology,
healthcare management and medical billing.
His work focuses on practical problems at the intersection
of medicine, healthcare operations, technology and intelligent automation.
As founder of OnnX, he is exploring how better
structured information and upstream workflow design can reduce unnecessary
administrative friction for physician-owned practices.
Connect with Dr. Cham on LinkedIn:
Dr.
Daniel Cham on LinkedIn
Sources
Dr. Sanjana Kochhar's heart-transplant story
The British Heart Foundation documents Kochhar's heart failure, urgent
transplant, recovery, return to GP practice, connection with her donor's family
and reflections on becoming a transplant recipient.
Sanjana Kochhar becomes a mother
People reported September 30, 2026 that Kochhar, a Liverpool GP who survived
multiple organ failure and an emergency heart transplant, welcomed her first
child in August 2026.
Current physician perspective on AI and context
The American Medical Association and five leading physician organizations
issued a joint statement September 30, 2026 emphasizing that physicians
evaluate patients in context and warning against treating AI as a substitute
for physician expertise and responsibility.
Current interoperability and administrative-burden data
The Office of the National Coordinator for Health Information Technology
reported September 30, 2026 that more than three-quarters of surveyed family
physicians experienced at least one substantial burden involving external
information retrieval, prior authorization or after-hours documentation;
prior-authorization burden increased from 54% in 2024 to 58% in 2026.
Current prior-authorization physician survey
The AMA's 2026 survey reported an average of 40 prior authorizations per
physician per week and approximately 13 hours of physician and staff time per
week devoted to them, alongside reported delays, burnout and patient
consequences.
Disclaimer
This article is for general educational and informational
purposes only. It does not constitute medical, legal, coding, reimbursement,
compliance, regulatory or financial advice.
Healthcare organizations should obtain appropriate
professional advice for their specific circumstances.
References
- American
Medical Association. “Statement from leading physician organizations
on the role of augmented intelligence in medicine.” September 30, 2026.
American Medical Association - Office
of the National Coordinator for Health Information Technology (ONC). “Less
Pajama Time, More Patient Time: How Better Interoperability Can Reduce
Physician Burden.” September 30, 2026.
ONC — HealthIT.gov - American
Medical Association. “AMA survey: Prior authorization reform pledge
falls short for physicians.” 2026.
AMA Prior Authorization Survey
#Healthcare #HealthcareAI #MedicalBilling #HealthTech
#HealthcareInnovation #PhysicianLeadership #RevenueCycle #MedicalPractice
#HealthcareData #PhysicianEntrepreneur #ClinicalWorkflow #PracticeManagement
#PatientExperience #IndependentPractice #HealthcareTechnology
