Long-Term Survival of Patients With Glioblastoma of the Pineal Gland: A ChatGPT-Assisted, Updated Case of a Multimodal Treatment Strategy Resulting in Extremely Long Overall Survival at a Site With Historically Poor Outcomes

Abstract

We present an updated case report of a patient with glioblastoma isolated to the pineal gland with an overall survival greater than five years and no progression of focal central nervous system (CNS) deficits since initial presentation. The patient underwent radiotherapy up to 60 Gy with concurrent and adjuvant temozolomide with the use of non-standard treatment volumes that included the ventricular system. The utilization of ventricular irradiation as well as the addition of bevacizumab at disease recurrence may have encouraged this unusually long survival by preventing/delaying leptomeningeal spread. We also present an updated review of the literature, which shows a median survival of six months, reinforcing the patients atypical disease trajectory. Finally, we utilize OpenAI's language model ChatGPT to aid in synthesizing this manuscript. In doing so, we demonstrate that ChatGPT is apt at creating concise summaries of relevant literature and topic subjects, however its output is often repetitive with similar sentence/paragraph structure, less than ideal grammar and poor syntax requiring editing. Thus, in its current iteration, ChatGPT is a helpful aid that cuts down on the time spent in data acquisition and processing but is not a replacement for human input in the creation of quality medical literature.

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

“…The generative nature of the LLM algorithms will likely fabricate a fake reference to substantiate false claims [31], a process that has been referred to as “hallucinations” [59]. Additionally, such hallucinations can be communicated via persuasive prose [42], making it more likely to mislead patients. For example, Jo et al mentioned that LLMs (specifically CareCall based on NAVER AI in this paper) may make ambitious or impractical promises to patients, which may add extra burden to therapists or cause a trust crisis [2].…”

“…Another contributing factor to inaccuracy is the outdated knowledge base used to train LLMs [21,25,30,41]. ChatGPT based on GPT3.5 was pre-trained by using data collected until 2021 and does not support Internet connection [49], making it unable to perform appropriately on questions regarding events that happened after 2021 [42].…”

“…Second, ChatGPT struggled to interpret all the coherent laboratory tests [60], generating superficial and incorrect responses. Indeed, ChatGPT could generate overly general answers without citing original references [20,40,42].…”

The Expanding Role of ChatGPT (Chat-Generative Pre-Trained Transformer) in Neurosurgery: A Systematic Review of Literature and Conceptual Framework

“…The comprehensive literature review yielded 22 articles that explored the application of Chat Generative Pre-Trained Transformer (ChatGPT) in healthcare and neurosurgery, shedding light on its potential benefits and limitations (Table 1 ) [ 2 , 5 - 7 , 10 - 31 ]. Some articles highlighted ChatGPT's capability to support medical professionals by providing accurate information and answering queries based on patient data analysis, aiding, therefore, in decision-making processes [ 2 , 11 ]. However, concerns were also raised about potential misinformation and the need for human oversight to ensure patient safety and address ethical considerations [ 2 , 11 ].…”

“…Some articles highlighted ChatGPT's capability to support medical professionals by providing accurate information and answering queries based on patient data analysis, aiding, therefore, in decision-making processes [ 2 , 11 ]. However, concerns were also raised about potential misinformation and the need for human oversight to ensure patient safety and address ethical considerations [ 2 , 11 ].…”