Are You Standing on Thin Ice When Using AI for Healthcare

Are You Standing on Thin Ice with Your AI?

During the winter lakes and ponds can freeze over, giving an opportunity for a lot of activities not otherwise available. A wise man knows not to go out onto the ice until it is thick enough to bear weight. If you go out on thin ice, then you are lucky if you just got soaked.

What is Standard of Care?

When a clinician provides care to a patient, he or she has the duty to provide the best possible care. Actions must be reasonable, prudent and adherent to standards generally accepted by fellow clinicians. Any other actions may represent negligence, making the clinician liable for damages if the patient suffers harm.

Typically, there is no scientifically defined standard of care. Rather experts are asked about what a reasonable clinician would do in a situation, looking for a consensus. Reaching consensus on what is the best care can take time. Changes in consensus tend to occur slowly, taking months or years for significant change.

How does this apply to AI?

There are several ways to use AI in healthcare. Some uses are quite reasonable, working with the strengths and capabilities of the technology. However, it is easy to get carried away by the possibilities and to push the envelope. People always want to be the first to achieve something new.

However, there are certain things that should be done with caution:

• Prescriptive orders.
• Black box operations.
• Full automation with no clinician review.
• Any action carrying high risks for the patient.

If these or similar actions are taken, then there is a risk of violating a standard of care.

Why care?

Unfortunately, many people assume that AI is guilty if a problem involved AI. This belief is strengthened if you cannot produce clear justification for what actually happened. Good luck doing that with a black box.

• Who will have your back if you get caught up in a malpractice suit involving AI?
• AI platforms? All have disclaimers saying blame belongs to the user.
• Malpractice insurance? They may say the actions were outside of coverage.
• Payers? They are always eager to find ways to dodge payments.
• Regulators? They are glad to find examples to pillory.

If you are an early adopter and are accused of malpractice, then you will find yourself alone and on thin ice. Your only hope is that you have taken appropriate precautions to protect yourself.

How can I stand on solid ground?

Business process modeling offers many advantages.

• It is generally accepted standards-based solution used by many industries and countries.
• It can provide clear documentation of what was done, when, and why.
• It can show the evidence that was used for each decision.
• Clinicians can provide oversight at key points through the use of attended tasks.
• AI services can be constrained to operate where they are most effective with little chance to hallucinate.

Proper modelling can provide the solid ground that you need when using the latest technology.

Take Home Message

AI tools are great and powerful, but they can blindside you if you rush in unprepared. On the other hand, if you hold off using these tools, then you risk falling behind in the use of technology. Business process modeling can be the ace that you need to manage risk.

If you want to discuss these topics or see AI integrated with business models. come see us at HIMSS 2026 in Las Vegas (Booth 10018-31 in the AI Pavilion).