Doctors are now using artificial intelligence to communicate with their patients in a more friendly and efficient manner. AI tools like ChatGPT by OpenAI have been helping doctors by answering messages and taking notes during exams. Thousands of doctors around the world are already using similar AI products based on large language models, with one company boasting that its tool works in 14 different languages.
Advocates of AI in healthcare say that these tools save doctors time, prevent burnout, and improve patient care. However, the use of AI in doctor-patient communication also raises important questions about trust, transparency, privacy, and the future of human connection in healthcare.
Here is a look at how new AI tools are impacting patients:
Is my doctor using AI?
In recent years, medical devices powered by machine learning have been performing tasks such as reading mammograms, diagnosing eye diseases, and detecting heart problems. What sets generative AI apart is its ability to respond to complex instructions by predicting language.
In the near future, your doctor might use an AI-powered smartphone app to record your check-up, which will then be transcribed into notes for easy reference. These tools can also help doctors bill insurance companies more accurately, ensuring that no details are overlooked.
Prior to using AI tools, your doctor should seek your consent. You may also notice new language in the forms you sign at the doctor’s office, indicating the use of AI technology.
Other AI tools may assist in drafting messages to patients, but patients may not always be aware of this. Some health systems encourage disclosure of AI use, while others leave it up to the discretion of individual healthcare providers.
Doctors and nurses are required to review and approve AI-generated messages before sending them to patients. In some cases, messages contain a sentence indicating that they were generated by AI, but this disclosure can be deleted by the doctor.
Patient Tom Detner from Denver received an AI-generated message from his doctor that he found remarkably authentic. However, the message ended with a disclosure that it was automatically generated and edited by his doctor, which Detner appreciated for its transparency.
Will AI make mistakes?
While large language models can sometimes misinterpret input or fabricate inaccurate responses, AI tools typically have built-in guardrails to prevent such errors from reaching patients or being included in electronic health records.
Dr. Alistair Erskine from Emory Healthcare in Georgia emphasizes the importance of ensuring accurate documentation in clinical notes and uses a product from Abridge to facilitate this process. The tool reviews doctor-patient conversations across multiple language models to eliminate inaccuracies and hallucinations.
Abridge CEO Dr. Shiv Rao notes that doctors play a crucial role in reviewing AI-generated notes for accuracy, as they can listen to specific segments of the patient visit to verify information. Despite occasional errors, AI tools are designed to enhance, not replace, the doctor’s expertise.
However, in a case in Buffalo, New York, an AI tool misunderstood a conversation between a doctor and a teenage cancer patient, resulting in a misleading note. Dr. Lauren Bruckner highlights the importance of ensuring that AI tools accurately reflect the conversation.
What about the human touch?
AI tools can be programmed to convey empathy and friendliness in their messages. While these tools can be beneficial, they must be used carefully to avoid misunderstandings or inaccuracies.
Dr. C.T. Lin from UC Health in Colorado acknowledges the mixed effectiveness of using a Microsoft AI tool to draft messages for patients. The tool can be helpful with routine advice but may require additional teaching for new medical information to ensure accuracy.
On the positive side, AI tools allow doctors to focus on their patients during appointments rather than being tied to their computers, as the tools record the exam details. Doctors are learning to verbalize their actions during exams for the benefit of AI tools and patient understanding.
Dr. Robert Bart from UPMC in Pittsburgh explains how talking through the exam helps both the AI tool and patients understand the medical process better. This approach enhances the doctor-patient interaction and ensures that information is accurately documented.
What about privacy?
US law requires healthcare systems to ensure the protection of patient data when using AI tools. Doctors interviewed for this article express confidence in the data security of these products and emphasize that patient information will not be sold.
While sharing information with AI tools helps improve their functionality, it also poses a risk of potential healthcare data breaches. Dr. Lance Owens from the University of Michigan Health-West believes that patient data is secure and protected when using Microsoft AI tools to document patient exams.
Doctors recognize the importance of maintaining patient privacy and trust while leveraging AI tools to enhance patient care and streamline communication.
