Leveraging Artificial Intelligence Chatbot to Generate Cochlear Implant Insurance Appeal Letter

Objectives Cochlear implant (CI) denial letters are a time-consuming and repetitive task that could benefit from artificial intelligence (AI)-powered large language model (LLM) chatbots. This study assesses ChatGPT’s accuracy in creating insurance appeal letters compliant with the American Cochlear Implant (ACI) Alliance guidelines for CI necessity. Methods ChatGPT was prompted to generate zero-shot appeal letters for asymmetric hearing loss or single-sided deafness (AHL/SSD) CI denials. Thirty prompt variations were created, and responses were scored for guideline concordance by three CI providers. Accuracy and relevance of citations were assessed. One output error was excluded, yielding 29 unique responses and 87 total scores. Results Almost all (96.6%) ChatGPT-generated responses included multiple benefits of CI for AHL/SSD, with 79.3% partially aligning with the ACI Alliance guidelines. However, 51.7% contained hallucinated benefits. Of the 96 citations, only six (6.3%) accurately referenced peerreviewed sources; the rest were hallucinated (57.3%), erroneous (24%), or irrelevant (10%). Conclusion ChatGPT inconsistently produced accurate CI insurance appeal letters, often mixing correct guideline-based content with erroneous or misinterpreted information. CI providers should exercise caution and verify AI-generated materials before clinical or advocacy use.