How To Use AI Responsibly For Medical Advice

More people are turning to AI for questions they have – including questions about their health. About 1/3 of U.S. adults have asked AI for advice about their health information, according to a recent poll from KFF.

At Rochester Regional Health, we’re seeing more patients turn to artificial intelligence (AI) to better understand their health, whether it’s learning about a diagnosis, reviewing test results, or preparing questions for a visit. While AI can be a helpful tool, it is important to use it thoughtfully.

Everett Weiss, MD, co-chair of our AI Center of Excellence and Medical Director of Health Informatics, shares how to use AI safely and effectively as part of your care.

How patients are using AI for their health

Dr. Weiss, who is also a pediatrician at the special care nurseries at both Rochester General Hospital and Unity Hospital, hears often from patients who use AI to look up more information about health conditions.

Patients are using AI today as an extension of what they used to do with a simple Internet search. Instead of exploring answers by clicking through links and reviewing multiple websites, AI summarizes information into a single response that would otherwise take more time to find.

Some patients use AI to:

• create a list of potential diagnoses based on symptoms they have
• understand a diagnosis given by a provider
• explore treatment recommendations
• prepare questions to ask ahead of a visit
• better understand their lab or test results

“When an AI system presents you with an answer to a question, you will want to learn whether the AI system used accurate, relevant, trusted medical sources to summarize for the consumer to read,” Dr. Weiss said. “AI systems will often present to you a link to the source it referenced for the answer it provided you. If there is no link, or the referenced article is not from a trusted medical source, the AI answer should give you some pause.”

Taking a thoughtful approach with AI

Taking a thoughtful, questioning approach to AI can help you make better decisions about your health.

Checking where an AI tool is getting its information is a great practice – especially because AI systems are designed to give an answer, not to say, ‘I don’t know’ when it can’t find an accurate answer.

To get a helpful result, Dr. Weiss encourages patients to remember two things.

1. AI systems summarize existing articles and websites. Sometimes those articles and websites contain misinformation.
2. While AI is getting better, it can still hallucinate conclusions or invent sources of information that are incorrect.

“When an AI system can't find an answer, it is more inclined to make something up than to tell you it has no answer at all,” Dr. Weiss said. “If you understand how AI works, when it's presenting information back to you, you can ask critical questions about its response.”

When to use caution with AI

AI tools can be useful, but there are times when it’s important to rely on your healthcare provider instead of AI.

Protect your personal health information: Experts caution patients from oversharing and putting in very specific details or information related to their personal health. Those details can be identified and traced back to a specific individual, and the information shared with the AI company may not be protected in the same way as your medical record.

Don’t rely on AI for a definitive diagnosis: When symptoms could be attributed to a few different health conditions, it’s best to talk with your primary care provider or a more specialized healthcare provider. Rely on your actual healthcare provider for a firm diagnosis.

Talk with your provider before starting medications: Medical decisions often depend on unique factors for every patient. Starting, stopping, or changing medications based soley on AI generated advice can be dangerous. AI tools do not have access to your full health history or current medications. Talk to your provider to ensure a treatment plan or medication is safe and appropriate for you.

“Your healthcare provider can walk through the probabilities and likelihood and some of the reasons why he or she may be considering or not considering some of those diagnoses given your particular situation,” Dr. Weiss said.

Is it helpful to providers when patients use AI to answer questions?

The answer: it depends on how the patient approaches their AI conversation.

If a patient uses an AI chatbot to become more informed about their diagnosis or ask it to come up with questions where they want to engage in a conversation with their provider, it can be very helpful.

“When patients use artificial intelligence to help them to learn what questions to ask and how to better engage their provider in a conversation, it can be very helpful to the patient better understand their diagnosis,” Dr. Weiss said.

“These patients tend to be more adherent to treatment regimens and medications because they better understand their diagnosis and treatment plan.”

How providers are using AI

Patients are not the only ones using AI in their day-to-day life.

Many healthcare providers at Rochester Regional Health and across the country are incorporating this into their daily work. More than 80 percent of physicians who were asked by the American Medical Association about using AI at work responded that they use it for:

• summarizing medical research
• creating an initial draft of visit, discharge, and care plans notes
• preparing initial responses to patients in private communications
• generating chart summaries

“As everyone’s comfort level grows, the instances of providers who are using AI in their daily routines is going to grow exponentially,” Dr. Weiss said.

Understanding AI and your health privacy

It is important to understand how your information is handled when using AI tools.

When you enter a question into an AI system or chatbot, that information is being shared with the vendor or company that created the AI system.

For example, if you ask an AI chatbot, “I have a sore throat and fever. Should I go to the doctor?”, that information about a potential illness is linked to your:

• name
• email address
• date and time of usage
• other provided health information

This information, and certainly more sensitive information depending on what you ask of the AI system, can be stored by the company to improve its systems and data models. While these companies may not share your information with third parties, it is not always protected under the same strict privacy laws and security standards used by healthcare systems.

“Your information is not being protected in the same way that a health system is going to protect your health information with regards to privacy and security laws,” Dr. Weiss said.

“When a health system publishes AI systems for patients to use, there are agreements in place with AI vendors to ensure that any health information being put in is protected.”

Using AI responsibility as part of your care

AI can be a helpful tool for learning, exploring questions, and preparing for medical visits. However, it should be used with caution and works best when used alongside your healthcare team, not in place of it.

At Rochester Regional Health, we are here to help you navigate your health with confidence. If you have any questions or concerns, your provider can help you understand the information and make the best decisions for your care.