Why healthcare’s next breakthrough may not be doing more work faster—but preventing unnecessary work from happening in the first place.
“These burdens can be reduced or eliminated through
thoughtful interventions at all levels of the health care system.” — Centers
for Medicare & Medicaid Services (CMS), Optimizing Care Delivery: A
Framework for Improving the Health Care Experience.
Six-year-old Selma Davis’s transplant journey offers a
powerful reminder of what healthcare is ultimately about: people. But while the
industry races to automate everything from documentation to billing, a more
uncomfortable question remains: How much of the work we are getting better at
managing should never have existed in the first place? For physicians and
clinic owners, the answer could reshape how we think about AI, administrative
burden, and the future of medical billing.
The word worth paying attention to is eliminated.
Not merely automated. Not transferred to another department.
Not buried beneath a more attractive dashboard.
Eliminated.
That distinction matters more than ever as healthcare
embraces artificial intelligence, automation, and increasingly sophisticated
software.
We are becoming very good at doing things faster. But are we
getting equally good at questioning whether those things need to be done at
all?
Consider what happens when a claim contains incorrect
insurance information.
The claim is rejected. A biller investigates. Someone
contacts the front desk. A staff member checks the patient record. The
information is corrected. The claim is resubmitted. Another employee monitors
the outcome.
Now introduce AI.
The system detects the rejection faster, categorizes the
reason, drafts a message, prioritizes the follow-up, and produces a beautiful
report.
Impressive.
But if the original information could have been captured
accurately and validated earlier, how much of that downstream activity was
necessary?
We may have made the recovery process more efficient without
preventing the problem.
In other words, we have built a faster ambulance for an
accident we might have prevented.
And healthcare is not short of ambulances, digital or
otherwise.
My contrarian view: the next major opportunity in
healthcare technology may not be automating more work. It may be eliminating
the work that should never have existed.
Seven months waiting for a heart
A story from Nashville, Tennessee, makes the human stakes of
healthcare innovation impossible to ignore.
Six-year-old Selma Davis spent more than seven months
waiting for a donor heart at Monroe Carell Jr. Children’s Hospital at
Vanderbilt.
For a child, seven months is an enormous stretch of life.
For a family waiting for a transplant, every day carries uncertainty that most
of us can scarcely imagine.
Selma’s father, Jessie Davis, carried her toward
surgery. Her mother, Rachel Elder Davis, accompanied her. Her
grandmother, Donna Elder, known as “DeeDee,” and her grandfather, Charles
Elder, were part of the family supporting her through the ordeal.
Her medical team included pediatric heart transplant
surgical director Dr. Ziv Beckerman, congenital heart surgeon Dr.
Garrett Coyan, transplant surgeon Dr. Aaron Williams, and pediatric
heart failure and transplant cardiologist Dr. Rachel Harris. One of
Selma’s favorite nurses, Kaylee McGuire, helped comfort her. The
hospital’s therapy dog, Squid, offered another kind of companionship.
According to reporting by The Tennessean, Selma’s
transplant involved a donor heart recovered after circulatory death and a
preservation approach developed at Vanderbilt that could help expand the pool
of hearts available for transplantation.
This is innovation with a deeply human purpose: expanding
possibilities for patients who need a second chance at life.
Selma’s story is not evidence that medical billing
inefficiencies affected her treatment. There is no basis for making that claim.
Her transplant journey and the administrative problems faced by physician-owned
practices are different subjects.
But the story prompts a broader question about healthcare.
When we talk about innovation, are we measuring the
sophistication of the technology—or the meaningful problem it helps solve?
That is a question worth asking whether we are discussing
transplant medicine, clinical workflows, or medical billing.
A breakthrough should be judged by what it makes possible,
what burden it removes, and whether it improves an outcome that matters.
Technology is the means. Human benefit is the point.
Healthcare has an automation addiction
The current technology conversation often follows a
predictable pattern.
A problem appears. We introduce software.
The software adds steps. We introduce integration.
The integration fails. We assign someone to reconcile the
data.
The reconciliation produces another exception. We create a
dashboard.
The dashboard becomes overwhelming. We introduce AI to
summarize the dashboard.
At some point, it is fair to ask whether we are improving
healthcare or simply creating increasingly sophisticated ways to manage its
complexity.
If an unnecessary process can be automated, we may have
created a more efficient unnecessary process.
That is not transformation. That is administrative friction
with better branding.
To be fair, automation can be tremendously valuable. It can
reduce repetitive work, identify exceptions, improve access to information, and
help staff focus on tasks requiring human judgment.
The mistake is assuming that automation automatically
produces improvement.
Imagine a clinic that repeatedly contacts patients to obtain
information missing from their records.
A new system sends reminders automatically. Staff no longer
have to make every call manually.
That may save time.
But suppose the information was missing because the
registration process was confusing, the required fields were unclear, or the
information captured during scheduling was never transferred into the billing
workflow.
The reminders may be useful, but they do not resolve the
source of the problem.
The clinic has automated the chase.
What if the better solution is to prevent the missing
information in the first place?
The answer will depend on the circumstances. Some
information cannot be obtained in advance. Some patients need assistance. Some
processes involve legitimate uncertainty.
But those realities are reasons to design better systems,
not reasons to stop asking questions.
AI should help healthcare eliminate unnecessary work—not
simply help it become more productive at repeating it.
The hidden economics of “just one more step”
Small and medium-sized physician-owned clinics understand
the cumulative cost of administrative friction.
One missing field may seem insignificant. One insurance
discrepancy may take only a few minutes to correct. One unclear handoff may
require a quick message.
But multiply those events across a busy practice, and the
work accumulates.
A front-desk employee corrects patient information. A
medical assistant tracks down missing details. A clinician clarifies
documentation. A biller investigates a rejected claim. A practice manager tries
to understand why payment is delayed.
Everyone is working.
Yet the underlying problem may be a small piece of
information that was not captured correctly at the beginning.
This is how administrative burden grows: not necessarily
through one enormous failure, but through hundreds of recurring exceptions that
become part of the daily routine.
Eventually, people stop questioning the process.
“That’s how we’ve always done it.”
“The system requires it.”
“We need someone to double-check.”
“We tried fixing it before.”
Sometimes these explanations are legitimate. Healthcare has
real compliance obligations, clinical safeguards, payer requirements, and
operational constraints.
But sometimes a recurring manual check is a warning sign.
It may indicate that information is not reliable, that
systems cannot share it properly, or that no one can tell whether a task has
actually been completed.
A recurring workaround may be evidence of a recurring
process failure.
And when organizations respond by adding another check
without investigating the cause, they can end up paying several people to
compensate for a weakness nobody has taken ownership of fixing.
That is not always a staffing problem.
Sometimes it is a design problem.
Medical billing begins before the claim
Medical billing is often treated as a downstream function.
The patient receives care. The encounter is documented. The
claim is created. The billing team takes over.
When the claim is denied or payment is delayed, the billing
department is expected to investigate and resolve the issue.
But the person who encounters a problem is not necessarily
the person—or process—that created it.
A claim can be affected by missing information, inaccurate
demographics, coverage discrepancies, incomplete documentation, authorization
issues, coding questions, or a breakdown in the handoff between departments.
Each issue has a different cause and requires a different
response.
Yet the billing team often inherits the consequences after
the original opportunity to prevent the problem has passed.
This is why I believe medical billing is a data-quality
problem as much as it is a workflow problem.
A billing system can process information efficiently. It
cannot reliably compensate for information that is absent, inaccurate, or
inaccessible.
An AI tool can summarize a clinical note. It cannot make an
undocumented fact true.
A dashboard can flag an authorization request. It cannot
turn that request into an approval.
An automated claim system can identify a discrepancy. It
cannot always determine the correct resolution without additional information
or human judgment.
Garbage in, beautifully summarized, is still garbage.
The answer is not to abandon billing tools. Practices need
effective systems to submit claims, track payments, manage denials, and resolve
exceptions.
The opportunity is to improve the information and processes
feeding those systems.
Was the relevant information captured accurately? Was it
available to the people who needed it? Was the handoff clear? Could the problem
have been detected earlier? Is the practice paying staff to repair the same
category of mistake repeatedly?
Those are not merely billing questions.
They are questions about how a practice operates.
Stop measuring activity as if it were achievement
Healthcare technology often celebrates activity.
Claims processed. Tasks automated. Messages generated.
Exceptions identified. Minutes saved.
These metrics can be useful. But they do not automatically
demonstrate that a practice is performing better.
A system that identifies twice as many exceptions may be
more capable—or it may be exposing a process that generates too many
exceptions.
A tool that sends more reminders may increase activity
without reducing delays.
A dashboard that reports more detailed denial data may
improve visibility without changing the reasons claims are denied.
The important question is what happens next.
Are fewer errors recurring? Is the information more
complete? Are staff members spending less time reconstructing events? Are
claims more likely to be accepted on the first submission? Are unresolved
issues being addressed earlier?
A claim submitted is not necessarily a claim paid.
An authorization request is not an approval.
A task marked complete is not proof that the intended
outcome occurred.
Activity is not the same as completion.
The distinction should shape how practices evaluate
technology.
Rather than asking only how many tasks a system performs,
ask how many unnecessary tasks it prevents, how much avoidable rework it
reduces, and whether it improves the final result.
Not every improvement will produce immediate financial
savings. Some will improve reliability, reduce risk, or free staff capacity for
other work.
Those benefits matter, but they should be measured honestly
rather than buried beneath a vendor’s impressive productivity statistics.
Before buying more AI, ask three uncomfortable questions
AI can be part of the solution, but it should not be the
starting assumption.
Before adopting another tool, physicians and clinic owners
should ask:
1. Should this task exist at all?
Is the task required for patient safety, accurate billing,
compliance, or another legitimate purpose? Or does it exist because the process
is unreliable?
2. If the task is necessary, why is it so difficult to
complete correctly?
Is the information missing? Is the system disconnected? Is
responsibility unclear? Are requirements poorly communicated? Is there a
genuine constraint the practice cannot control?
3. What would make the task unnecessary—or prevent it
from recurring?
Could the practice improve data capture, validate
information earlier, clarify a handoff, or make status information accessible?
Would automation help after the process is redesigned?
These questions sound simple. Their implications are not.
They challenge the assumption that every existing task
deserves to survive simply because it has become familiar.
A process may be necessary. It may also be poorly designed.
A task may be unavoidable. The repeated corrections
surrounding it may not be.
And sometimes the most effective solution is not a new
technology product. It is a clearer process that everyone can follow reliably.
The best tool is the one that addresses the actual
problem—not necessarily the one with the most impressive demonstration.
A practical framework for eliminating avoidable rework
The good news is that a practice does not need to redesign
its entire operation to begin.
Start with one recurring problem.
1. Identify the work everyone has learned to tolerate
Ask the people doing the work:
- What
do you have to correct repeatedly?
- Which
information is hardest to find?
- Where
do handoffs break down?
- Which
tasks require repeated follow-up?
- What
causes the same exceptions to return to the work queue?
Listen before selecting a solution. Staff members closest to
a process often understand its weaknesses better than anyone looking at it from
a distance.
2. Trace the problem to its origin
Suppose billers repeatedly contact clinical staff to clarify
documentation.
Do not immediately conclude that the practice needs a better
messaging tool.
Investigate the pattern. Are requests concentrated around
particular encounter types? Are documentation expectations unclear? Is the
information available but inaccessible? Is a payer requirement being
misunderstood?
The visible problem may have several possible causes. Find
the actual one before prescribing a solution.
3. Separate necessary work from avoidable rework
Map the process from its beginning to its final outcome.
Identify which steps are required for clinical, legal,
operational, or payer reasons. Then identify where work is repeated because
information is missing, inconsistent, delayed, or inaccessible.
Do not remove safeguards simply because they add time.
Instead, ask whether they can be made more reliable or
whether the conditions requiring repeated checks can be corrected.
4. Fix the source before adding another layer
If information is inaccurate, improve how it is captured and
validated.
If systems cannot share necessary information, examine
integration and access.
If responsibility is unclear, define ownership.
If tasks depend on memory, make status and next steps
visible.
If a requirement is unavoidable, make compliance with it as
clear and efficient as possible.
Then determine whether automation can reliably support the
redesigned process.
5. Measure the outcome
Establish a baseline before changing the workflow.
Depending on the problem, useful measures might include:
- First-pass
claim acceptance or clean-claim rate.
- Denial
rate and the reasons for denials.
- Frequency
of corrected or resubmitted claims.
- Staff
time spent on repeated data entry and follow-up.
- Time
required to resolve missing-information exceptions.
- Days
in accounts receivable, interpreted alongside payer mix and other relevant
factors.
- The
proportion of tasks completed correctly without intervention.
Compare results over a reasonable period. Account for
changes in volume, staffing, payer behavior, and other relevant factors.
Do not assume that every improvement automatically becomes a
cash saving. Reduced staff time may free capacity rather than reduce payroll.
That capacity still has value, but the business case should
reflect what actually changed.
6. Confirm that the work really disappeared
This is where many improvement efforts fall short.
A new tool is installed. The team likes the interface. The
vendor reports successful automation. Everyone celebrates.
But did the work disappear, or did it move somewhere else?
Perhaps the AI drafts a message, but an employee must still
verify every detail. Perhaps a claim is flagged automatically, but no one owns
the exception. Perhaps duplicate entry disappears from one screen but remains
necessary in another.
A task is not eliminated just because the software no
longer displays it.
Follow the process all the way to the outcome.
If the work has simply moved, the improvement may be smaller
than expected.
The financial case for preventing problems
Avoidable rework has a cost.
Staff spend time finding information, checking records,
communicating across departments, correcting errors, resubmitting claims, and
monitoring unresolved issues.
Those hours have an opportunity cost.
A staff member correcting a preventable mistake cannot spend
that same time helping a patient, resolving a complex claim, or improving the
practice’s operations.
The actual financial impact varies by clinic, specialty,
payer mix, staffing model, and error type. It would be misleading to promise
that eliminating a particular number of administrative steps will automatically
produce a specific return.
A better approach is to calculate the local cost.
Estimate the staff hours consumed by a recurring problem.
Determine the relevant labor cost. Measure how much of the work remains after
an intervention. Separately assess whether the change affects payment timing,
denial recovery, or other financial outcomes.
Avoid double-counting benefits. An hour saved is not
necessarily an hour of direct cash savings if staffing costs remain unchanged.
It may instead be capacity that can be redirected toward
higher-value work.
The strongest business case is grounded in evidence: a
clearly defined problem, a targeted intervention, and a measurable result.
No magic AI multiplier required.
Just a problem worth solving, a thoughtful approach, and a
way to determine whether it worked.
The human cost of unnecessary work
There is another cost that does not always appear on a
dashboard.
Administrative friction changes how people experience their
work.
When employees repeatedly correct the same errors, they can
begin to feel that the system is designed to make them fail. When
responsibility is unclear, people spend time protecting themselves from blame
rather than improving the process. When queues constantly refill, completion
can feel permanently out of reach.
This is not to say that administrative work is the sole
cause of burnout or that technology alone can solve workforce problems.
Healthcare teams face many pressures.
But avoidable rework is worth examining because it consumes
time without necessarily adding value.
It can also reveal an organizational habit: expecting the
people closest to the problem to compensate indefinitely for weaknesses in the
system.
A better question is not simply, “Who made the mistake?”
It is, “What conditions made the mistake likely, and what
would make the correct outcome easier to achieve?”
Accountability matters. Training matters. Human judgment
matters.
But blaming the final person in a chain does not necessarily
fix a problem that began several steps earlier.
Sometimes the best way to support staff is not to give them
another tool to manage the workload.
It is to stop generating so much unnecessary workload.
Where OnnX fits into this conversation
As a physician-entrepreneur and founder of OnnX, I
think about this problem through the lens of medical billing and practice
operations.
OnnX is being developed around a simple premise: for small
and medium-sized physician-owned clinics, better billing should begin with
better information and more reliable processes—not simply more downstream
activity.
That means looking beyond the final claim and asking what
happened before it was created.
Was the relevant information captured accurately? Was it
available to the people who needed it? Did the handoff preserve the context?
Could an avoidable error have been detected earlier? Is the practice repeatedly
paying staff to repair a problem that the process could prevent?
These questions matter because adding another layer of
software is not automatically the right answer.
Sometimes a practice needs automation. Sometimes it needs
integration. Sometimes it needs clearer processes or better data capture.
Sometimes it needs to stop doing something that no longer serves a legitimate
purpose.
The solution should follow the problem.
This is especially important for smaller practices, where
time and resources are limited. A tool that creates more alerts, more
dashboards, and more steps can become another obligation rather than a
meaningful improvement.
The objective is to help practices reduce avoidable rework,
improve reliability, and make billing less dependent on repeated manual
intervention.
That is the philosophy behind OnnX—not the idea that AI
should do everything, but that healthcare should not have to keep doing the
same unnecessary things.
Healthcare needs fewer heroic workarounds
Selma Davis’s story is ultimately about a child, her family,
a donor, and the clinicians who worked to give her a chance at a new heart.
The medical innovation described in the reporting mattered
because it expanded what the transplant team could consider possible. It
addressed a real constraint.
That is a useful standard for healthcare technology.
Start with a meaningful problem. Understand why it exists.
Develop an approach that addresses the cause. Evaluate whether it improves the
outcome.
Then ask whether the solution reduces the burden on the
people who must live with the process every day.
In medical billing, that may mean preventing inaccurate
information from moving downstream. In practice operations, it may mean
designing clearer handoffs. In technology, it may mean choosing not to automate
a task until someone has established that the task is necessary.
We do not need to romanticize inefficiency. We do not need
to assume every old process is bad. And we should not pretend that one
transplant story proves a theory about billing.
We do need to become more disciplined about what we call
progress.
A system that processes more work is not necessarily a
system that works better.
A team that resolves more errors may still be trapped in a
process that generates too many of them.
And a clinic that buys more technology may still be asking
its staff to compensate for information that was never captured properly in the
first place.
Perhaps the next wave of healthcare innovation should come
with a different question on the evaluation form:
What unnecessary work did this eliminate?
If the answer is “none,” that does not automatically make
the technology useless. But it should make us curious about whether we are
solving the right problem.
The future of healthcare will undoubtedly include more AI,
more automation, and more sophisticated tools.
The more important question is whether those tools will help
healthcare teams spend less time managing preventable problems—and more time
doing the work that genuinely matters.
After all, we do not need AI to make every administrative
task faster.
Sometimes we need the wisdom to stop creating the task.
And perhaps the most innovative workflow is the one that
quietly gives a busy clinic something increasingly rare:
One less problem to solve.
A question for physicians and clinic owners
What is one recurring administrative task in your practice
that everyone has accepted as “just part of the job”—even though it may be
preventable?
Is the opportunity to automate the fix, or to eliminate the
cause?
About the author
Daniel Cham, MD, is a physician-entrepreneur and
founder of OnnX, an AI-powered medical billing SaaS initiative focused on small
and medium-sized physician-owned clinics. His work explores how better data
quality and more reliable processes can help practices reduce avoidable
administrative work.
References
- Centers
for Medicare & Medicaid Services (CMS). Optimizing Care
Delivery: A Framework for Improving the Health Care Experience. Official
CMS framework.
- The
Tennessean. Beth Warren, “Doctors worried a storm might thwart their
flight to save a 6-year-old,” October 11, 2026. Read
the report.
- Vanderbilt
University Medical Center. “First-of-its-kind heart transplant gives
Sissy a second chance,” October 7, 2026. Read
the institutional report.
#HealthcareInnovation #MedicalBilling #HealthcareAI
#RevenueCycleManagement #PhysicianEntrepreneur #PracticeManagement
#HealthcareAdministration #DataQuality #OperationalEfficiency #OnnX
