Bilateral Vocal Fold Paralysis in a Patient With Neurosarcoidosis: A ChatGPT-Driven Case Report Describing an Unusual Presentation

Abstract

This ChatGPT-driven case report describes a unique presentation of neurosarcoidosis. The patient, a 58year-old female, initially presented with hoarseness and was found to have bilateral jugular foramen tumors and thoracic lymphadenopathy. Imaging revealed significant enlargement and thickening of the vagus nerve and a separate mass of the cervical sympathetic trunk. The patient was referred for an ultrasound-guided biopsy of the abnormal neck masses to establish a pathologic diagnosis. The patient subsequently underwent neck dissection for exposure of the vagus nerve and isolation of the great vessels in preparation for a transmastoid approach to the skull base. The presence of multifocal tumors prompted the need for a biopsy, which ultimately revealed sarcoid granulomas in the nervous system. The patient was diagnosed with neurosarcoidosis. This case highlights the potential for sarcoidosis to affect the nervous system, with multiple cranial nerve involvement, seizures, and cognitive impairment. It also emphasizes the need for a combination of clinical, radiological, and pathological findings for an accurate diagnosis of neurosarcoidosis. Additionally, this case highlights the utility of natural language processing (NLP), as the entire case report was written using ChatGPT. This report serves as a comparison of the quality of case reports generated by humans versus NLP algorithms. The original case report can be found in the references.

A Systematic Review of ChatGPT and Other Conversational Large Language Models in Healthcare

“…At the same time, there are four major themes of concern when using these models in practice, including reliability, bias, privacy, and public acceptance. Specifically, the most popular application (30 out of 65 papers) for LLMs was for medical knowledge inquiries, with the second most popular (25) being summarization, followed by prediction (22), and then administration (9). At the same time, most of the papers expressed concerns about reliability (55), followed by bias ( 16), then privacy ( 6), and finally public acceptance (4).…”

“…ChatGPT has been shown to be effective in summarizing medical documents for a diverse set of applications [13,14], including tasks such as adapting clinical guidelines for diagnosis, treatment, and disease management [15], summarizing medical notes [16][17][18], assisting in writing medical case reports [19][20][21][22][23][24], and generating and translating radiological reports [18,25]. Notably, efforts have been made to integrate ChatGPT-4 1 with "Link Reader" for automating medical text synthesis [26], which boosted model performance in providing answers according to clinical guidelines [26].…”

A Systematic Review of ChatGPT and Other Conversational Large Language Models in Healthcare (Preprint)

“…At the same time, there are four major themes of concern when using these models in practice, including reliability, bias, privacy, and public acceptance. Specifically, the most popular application (30 out of 65 papers) for LLMs was for medical knowledge inquiries, with the second most popular (25) being summarization, followed by prediction (22), and then administration (9). At the same time, most of the papers expressed concerns about reliability (55), followed by bias ( 16), then privacy (6), and finally public acceptance (4).…”

“…ChatGPT has been shown to be effective in summarizing medical documents for a diverse set of applications [13,14], including tasks such as adapting clinical guidelines for diagnosis, treatment, and disease management [15], summarizing medical notes [16][17][18], assisting in writing medical case reports [19][20][21][22][23][24], and generating and translating radiological reports [18,25]. Notably, efforts have been made to integrate ChatGPT-4 1 with "Link Reader" for automating medical text synthesis [26], which boosted model performance in providing answers according to clinical guidelines [26].…”