Photo By: Navy Medicine
Artificial intelligence is changing healthcare in ways that go far beyond helping doctors. AI is now being used to support medical coding, claims processing, billing, revenue cycle management and other healthcare decisions.
That shift raises an important question. Are the policies and oversight systems created for traditional healthcare ready for a world where the electronic health record is becoming a decision-making engine?
For years, the EHR was mainly viewed as a digital record. It stored information about a patient’s symptoms, diagnoses, medications, test results and treatment.
Today, it is becoming much more than that.
AI can review clinical notes, suggest billing codes, identify missing documentation, flag claims that may be denied and help healthcare organizations decide which cases need additional attention.
In many situations, a person still makes the final decision. A coder accepts or rejects an AI recommendation. A claims employee reviews a flagged case. A clinician decides whether to follow a recommendation.
But the fact that a person makes the final decision does not mean AI is not shaping it.
When a recommendation becomes a decision
Consider medical coding.
An AI system reviews a physician’s documentation and recommends a particular code. The coder sees that recommendation and accepts it.
Technically, the coder made the decision.
But what happens when the same process takes place thousands of times? If employees usually accept the recommendation, the AI system may have significant influence over the final outcome.
The same issue can arise with claims. An AI system might predict that a claim is likely to be denied. The healthcare organization then sends the claim for additional review or requests more documentation.
The AI did not technically deny the claim. But it changed what happened to it.
That distinction could become increasingly important as AI becomes part of everyday healthcare operations.
The problem with accountability
Healthcare already has rules covering privacy, billing, fraud, medical devices and patient care.
Many of those rules, however, were created around a relatively simple model. A person or organization makes a decision, and that person or organization can be held responsible.
AI makes that process more complicated.
If an AI system recommends an incorrect billing code, who is responsible? The coder who accepted it? The healthcare organization that deployed the system? The company that built the AI? Or the people who approved its use?
The same questions can arise when AI influences claims, prior authorization or other processes that can affect patients and providers.
These systems may not make decisions on their own, but they can influence which decisions get made and how quickly they happen.
The EHR is becoming an ecosystem
The EHR is increasingly connected to many parts of healthcare, including clinical documentation, coding, billing, scheduling, claims and authorization.
That creates opportunities to reduce administrative work and help healthcare organizations manage large amounts of information.
Jim Maguire, co-founder of Government Market Strategies, points to the potential benefits of using AI in these areas.
“If oversight frameworks ensure that claim edit rules and decision logic are applied consistently and transparently, AI could become one of the most important levers for reducing administrative costs and improving efficiency across the entire healthcare industry.”
His point highlights an important part of the debate. The question is not simply whether AI should be used in healthcare. It is also whether organizations can put the right oversight around these systems so they are applied consistently and transparently.
AI could help reduce administrative work, but those benefits depend on knowing how the systems are making recommendations and having safeguards when those recommendations are wrong.
Healthcare may need a new approach
The answer may not be creating a new rule for every AI application.
Instead, healthcare organizations and regulators may need to focus on how much influence an AI system has and how much harm could result if it is wrong.
A system that helps write a routine message should not require the same level of oversight as a system that can influence reimbursement or access to care.
Organizations should be able to answer basic questions about important AI systems. Who approved the system? What decisions can it influence? What information does it use? How is it tested? How is it monitored? Can a person override it? What happens when it makes a mistake?
Those questions may become just as important as whether the technology is accurate.
The bigger question
Healthcare has spent decades building systems to document what happened to patients.
Now it may need better systems for documenting how important decisions were made.
As AI becomes more deeply embedded in the EHR, the line between decision support and decision-making will become harder to define.
The key question may no longer be whether AI made the decision.
It may be how much AI influenced the decision and who is responsible for that influence.
If the EHR is becoming a decision-making engine, healthcare policy will need to evolve with it.
The goal should not necessarily be to keep AI out of these systems. It may be to make sure that when AI influences healthcare decisions, the rules are clear, the logic is transparent and people remain accountable for the outcomes.
