Talk with ChatGPT About the Outbreak of Mpox in 2022: Reflections and Suggestions from AI Dimensions

The sudden emergence of mpox (monkeypox) in 2022, outside its typical endemic regions, prompted significant public health concern and highlighted the need for innovative approaches to disease surveillance and response. The World Health Organization (WHO) declared the outbreak a Public Health Emergency of International Concern (PHEIC) on July 23, 2022. By March 17, 2023, over 86,000 confirmed cases and 112 deaths had been reported globally. Previous research, including bibliometric analyses by the authors, revealed that mpox research had historically been neglected compared to other orthopoxviruses. The current surge in mpox-related publications presents an information overload challenge. This study utilizes ChatGPT, a cutting-edge generative artificial intelligence (AI) model, to address this information challenge and explore its potential in understanding and managing infectious disease outbreaks. ChatGPT's ability to process and generate text based on its training data offers a unique opportunity to analyze large datasets and provide insightful summaries and suggestions. This study aims to assess ChatGPT's capabilities in providing reflections and suggestions about the 2022 mpox outbreak from an AI perspective, enriching our knowledge and exploring the potential for human-AI collaboration in future epidemic or pandemic responses. The study focuses on leveraging AI to overcome information overload and facilitate quicker, more efficient analysis and response to disease outbreaks.

The authors cite existing literature highlighting the global research trends on mpox and orthopoxviruses, noting historical neglect of mpox research. They reference studies demonstrating the increasing number of publications on mpox epidemiology, surveillance, treatment, prevention, and vaccines. They also mention prior research on ChatGPT's capabilities in various fields, including its performance on the USMLE and its applications in medical research. This prior work establishes the context for the current study by showing both the need for new tools to process mpox information and the potential capabilities of ChatGPT.

The study employed a qualitative approach involving a series of interactions between the authors and ChatGPT. On March 17, 2023, the authors posed four specific questions to ChatGPT regarding the 2022 mpox outbreak. These questions focused on: 1) the potential causes of the outbreak's sudden emergence and spread; 2) analysis of future trends in confirmed cases; 3) broader implications for the future; and 4) five novel, unpublished systematic review ideas related to mpox. The responses generated by ChatGPT are presented and analyzed within the study. The authors then assessed the quality and relevance of ChatGPT's responses, critically evaluating its strengths and limitations. The study also involved a post-hoc verification of the suggested systematic review topics using the PubMed database. The authors compared the suggested research ideas against existing published work to evaluate the novelty of ChatGPT's suggestions. The assessment of ChatGPT's responses incorporates both qualitative analysis of the responses' coherence, accuracy, and insightfulness and quantitative analysis of the novelty of research ideas suggested by the model.

ChatGPT provided responses to all four questions, offering several potential explanations for the mpox outbreak's emergence (environmental factors, human behavior, pathogen evolution). Regarding future trends, ChatGPT acknowledged the inherent uncertainty but highlighted the importance of control measures, surveillance, and human behavior. The model identified numerous significant implications for public health, economic factors, global health security, vaccine development, and socio-cultural impacts. Perhaps most remarkably, ChatGPT successfully suggested five novel systematic review topics which, upon further investigation via PubMed, proved to be important areas deserving of further exploration. One example highlighted by the authors is the suggestion for a systematic review of mpox transmission dynamics and risk factors in animal-human transmission, noting the still-controversial mode of mpox transmission, encompassing sexual transmission, saliva transmission and airborne transmission. While ChatGPT demonstrated strengths in summarizing information, generating novel ideas, and offering a variety of perspectives, it also showed inherent limitations. The model itself cautioned against its ability to definitively answer causal questions or predict future trends, emphasizing that its responses were based on the information available in its training data. The authors found that the responses were quite general and did not analyze differences between studies critically. Furthermore, the study noted the risk of ChatGPT generating inaccurate results due to the presence of inaccurate information about mpox online, referring to a prior study demonstrating ChatGPT sometimes provided dangerous antimicrobial advice.

The study's findings highlight the potential of ChatGPT as an adjuvant tool in medical research, particularly for information gathering, idea generation, and knowledge summarization. The ability of ChatGPT to conceive novel systematic review topics, verified through a PubMed search, demonstrates its utility in identifying research gaps and prioritizing future research. However, the study emphasizes the critical need for human oversight and verification to mitigate the risk of inaccuracies. The potential for ChatGPT to generate misleading or dangerous information underscores the ethical considerations involved in integrating AI into medical practice. The limitations of the model, as highlighted by the model itself, emphasize the importance of human expertise in interpreting and critically evaluating the information provided by AI tools. Future research should focus on further refining AI models to improve accuracy, consistency, and contextual awareness, mitigating potential biases and inaccuracies.

This study demonstrates the potential of ChatGPT as a valuable tool for assisting medical researchers, particularly in exploring complex topics such as the 2022 mpox outbreak. While offering significant advantages in terms of information processing and idea generation, the study highlights the ethical concerns and limitations related to AI's use in medical research. Human expertise remains crucial for critical evaluation and validation of AI-generated information. Future research should focus on developing guidelines and protocols for responsible AI utilization in scientific research and improving the accuracy and reliability of AI models in the medical field, particularly in areas involving potential clinical implications.

The study's primary limitation is the inherent limitations of ChatGPT itself. The model is unable to definitively answer causal questions or predict future trends with certainty. The accuracy of ChatGPT's responses relies on the quality and accuracy of the data it was trained on, and potential inaccuracies in online information could lead to flawed outputs. Another limitation is the small scale of the interaction between the researchers and ChatGPT, which might not fully represent the capabilities of the model across a wider range of questions and contexts. The study’s reliance on PubMed for verification of the novel research suggestions might overlook relevant research published in other databases or sources.