Comparing human versus artificial intelligence for health literacy among patients with neurological disorders (HANDs study)

Chat Generative Pre-Trained Transformer (ChatGPT) is a language model developed by OpenAI that uses deep learning techniques to generate human-like responses to text input [1]. Large Language Models (LLMs) are trained on massive datasets through causal language modeling, predicting the next token in a sequence to learn patterns, grammar, and semantics. By scaling up parameters, data, and computational resources, LLMs achieve impressive text comprehension and generation capabilities. Fine-tuning with task-specific data and human feedback further aligns these models for practical applications. Publicly released in November 2022, ChatGPT supports a variety of natural language processing tasks, including question answering, text completion, and translation. Its ability to generate coherent, contextually appropriate responses has made it a popular tool for chatbots, virtual assistants, and other conversational artificial intelligence (AI) applications. The recently released ChatGPT version 4 (GPT-4) demonstrated significant advancements in natural language synthesis over earlier language models [2]. Since 2022, ChatGPT has also become a cultural phenomenon in the scientific community, spurring new interests and concerns regarding AI’s societal role [3]. An increasing number of publications discussed ChatGPT’s utility in scientific writing and programming. AI has been growing in many medical domains, including image analysis, drug interaction detection, the identification of high-risk patients, and clinical note coding [4–6]. Few studies explore ChatGPT’s role in neurological disorders, namely in the prediction of dementia and epilepsy [7, 8]. Health literacy has received growing attention over the past two decades for its critical role in improving individual and public health and maintaining healthcare systems’ sustainability. Many countries have prioritized health literacy in their policies and practices, and the WHO recommends health literacy as an instrument for achieving several key targets listed in the Sustainable Development Goals [9]. AI has been applied in academic research and in inference tasks across the broader economy with demonstrable success, but less so for the core functions of public health, namely protecting and promoting the health of populations [10]. With its ability to generate human-like text based on large amounts of data, ChatGPT could support individuals and communities in making informed decisions about their health. Our study aims to evaluate the preferences of neurologists and patients regarding responses to common clinical questions using a specific AI tool. Specifically, we assess ChatGPT’s effectiveness in conveying information to individuals with neurological disorders.