Quality of ChatGPT-Generated Responses to Common Patient Questions About Peripheral Nerve Stimulation: A...

Area of Science:

• Neurology
• Pain Management
• Artificial Intelligence in Healthcare

Background:

• Peripheral nerve stimulation (PNS) is a growing treatment for neuropathic pain.
• Patients frequently seek online information regarding PNS procedures and recovery.
• Large language models (LLMs) like ChatGPT offer potential for scalable patient education.

Purpose of the Study:

• To evaluate the reliability, accuracy, and comprehensibility of ChatGPT-5.0 responses to common patient questions about PNS.
• To assess the performance of an LLM in providing patient education for a specific interventional pain procedure.

Main Methods:

• A cross-sectional study involving 21 standardized PNS-related questions.
• Expert consensus was used to formulate questions covering pre-implantation, implantation, and post-implantation phases.
• Board-certified interventional pain specialists and a nurse educator rated responses using validated scales for reliability, accuracy, and comprehensibility.

Main Results:

• ChatGPT-5.0 responses received generally high ratings for reliability (mean 4.7/6), accuracy (mean 2.6/3), and comprehensibility (mean 2.8/3).
• Questions about PNS mechanisms, expectations, and postoperative care were rated highest.
• Responses to implantation-specific questions requiring procedural nuance showed more variability.

Conclusions:

• ChatGPT-5.0 demonstrates capability in generating generally reliable, accurate, and understandable patient information for PNS, especially for non-procedural aspects.
• Clinician oversight is crucial for verifying information related to the procedural nuances of PNS implantation.
• This study establishes a benchmark for LLM performance in patient education for interventional pain management, highlighting the need for continuous evaluation.