The Medical Record: How Artificial Intelligence (AI) Impacts Coding

For most of modern healthcare, the medical record was passive. It waited.
It stored information. It was interpreted later by coders, clinical
documentation integrity (CDI) professionals, auditors, and sometimes
attorneys.

The record reflected care, but it did not respond to it.
That reality has changed.

With the rise of generative tools such as ChatGPT and the recent
introduction of healthcare-focused models from ChatGPT Health and
Anthropic Claude, the record no longer sits quietly in the background. It
reads. It analyzes. It highlights gaps. And increasingly, it challenges what
we thought was sufficient.

This shift is not about novelty or speed. It is about visibility.
Artificial intelligence does not create new documentation problems. It
exposes the ones healthcare has tolerated for years. Vague diagnoses.
Missing acuity. Clinical relationships that were assumed but never clearly
stated.

What once passed as “good enough” is now easy to spot. Once something
is visible, it becomes measurable. Once it is measurable, it becomes
accountable.

For coding professionals, this moment is often misunderstood as a threat. It
is not. It is a repositioning.

Coding has always lived at the intersection of clinical narrative and data
structure. These tools now operate in that same space, but without context
or judgment.

They can identify inconsistencies and suggest specificity, but they cannot
determine clinical truth. That responsibility remains human. In fact, it
becomes more important.

The role of the coder is shifting from data extraction to data interpretation
with oversight. Judgment is no longer optional. It is the safeguard.

Clinical Documentation Integrity is undergoing an even more profound
evolution. CDI was never meant to be about queries alone. At its core, it
has always been about accuracy and clinical truth. What has changed is
the audience.

Documentation is no longer read only by humans. It is increasingly
consumed by systems that influence risk adjustment, quality metrics, and
future decision support. Every unresolved ambiguity becomes a data point.
Every clarification becomes instruction.

CDI professionals are no longer just mediators between physicians and
coders. They are shaping how clinical logic is understood at scale. The
question is no longer simply whether documentation supports a diagnosis.
It is what a system will conclude when it processes that documentation
thousands of times.

This reality elevates the importance of leadership understanding.
One of the most significant risks in healthcare today is not resistance to
artificial intelligence. It is superficial adoption without comprehension.
Leaders approve tools, review dashboards, and sign off on
implementations without understanding how conclusions are reached.

That is a governance issue.

Oversight requires more than awareness. It requires literacy. Leaders must
understand how outputs are influenced by inputs, how bias enters systems,
and why confidence does not equal accuracy. Without that foundation,
accountability is impossible.

This leads to a skill that is quickly becoming essential across roles:
prompting.

Prompting is not a technical trick or a novelty skill. It is structured thinking
expressed clearly. In healthcare terms, it is no different from writing a
precise order or a clear clinical note. The way a question is asked
determines the usefulness of the response.

Poorly designed prompts can oversimplify conditions, ignore exclusions, or
reinforce assumptions. Well-designed prompts can surface relationships
that are otherwise missed. Learning to interact with these systems
thoughtfully is becoming part of professional competence.

This is not about replacing clinical reasoning. It is about recognizing that
systems will reflect whatever reasoning we provide. If documentation is
imprecise, conclusions will be imprecise. If logic is incomplete, outcomes
will follow.

The medical record is no longer just a historical record. It is becoming an
active participant in how care is evaluated and understood.
The question facing healthcare is not whether these tools will be used.
They already are.

The real question is whether the people shaping documentation, integrity,
and leadership are prepared to meet the level of precision that the learning
record now demands.