The AI-generated patient portal messages that are supposed to make healthcare communication easier are actually increasing clinicians’ cognitive burden.
Disjointed Responses
Researchers have found that these AI-generated messages often don’t align with what a clinician would respond with, forcing medical staff to edit and correct the output. This not only wastes time, but it also introduces potential errors that could compromise patient care.
The study focused on the use of Large Language Models (LLMs) to generate patient portal messages. LLMs are a type of AI that can understand and generate human-like text. However, the researchers found that these models often produce responses that don’t quite match what a clinician would write in a patient portal message.
What’s Being Done to Improve It?
Fortunately, researchers are already working on adapting LLMs to better align with clinician responses. By fine-tuning these models with clinician-approved examples, they’re able to improve the accuracy of the generated messages and reduce the need for editing.
This is a crucial development, as patient portal messages are becoming increasingly important in modern healthcare. These messages allow patients to communicate directly with their clinicians, which can help reduce wait times and improve patient satisfaction. However, if the messages aren’t accurate or clear, it can actually create more problems.
Reducing Cognitive Burden
By improving the accuracy of AI-generated patient portal messages, clinicians can spend less time correcting and editing their responses. This can help reduce their cognitive burden and free up time for more important tasks, such as patient care.
What this means is that clinicians need to start paying closer attention to the patient portal messages that are being generated by AI. While these tools have the potential to improve communication and efficiency, they still need to be carefully managed and edited to ensure they’re producing accurate and helpful responses.
