ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid Concerns

Artificial intelligence (AI) has emerged as a transformative force across various domains, including healthcare. Its potential to revolutionize medical practice, research, and education is undeniable. This review focuses on ChatGPT, an AI-powered conversational large language model (LLM) developed by OpenAI. ChatGPT, trained on vast text datasets spanning multiple languages, has the ability to generate human-like responses to text input. While its potential applications in healthcare are promising, it's crucial to critically examine the associated concerns and limitations. The review aims to explore the utility of ChatGPT in healthcare education, research, and practice while highlighting the potential drawbacks that require careful consideration and mitigation strategies.

The review systematically analyzed a diverse range of records, encompassing research articles, preprints, editorials, commentaries, and opinion pieces. These sources provided valuable insights into the potential benefits and drawbacks of ChatGPT in healthcare education, research, and practice. The review methodology followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to ensure rigor and transparency.

The systematic review followed the PRISMA guidelines, employing a comprehensive search strategy to identify relevant records published in PubMed/MEDLINE and Google Scholar. The search was conducted on February 16, 2023, using the terms "ChatGPT" and specifying a publication date range from November 30, 2022, to the present. This search strategy yielded a total of 280 records. A meticulous screening process was implemented to ensure the inclusion of only relevant records. This process involved several stages: 1. **Initial Screening:** The initial screening involved reviewing titles and abstracts of all identified records. This step excluded duplicate records (n = 40) and records not written in English (n = 32). 2. **Scope Exclusion:** Records that focused on ChatGPT in a context other than healthcare education, research, or practice were excluded (n = 80). 3. **Source Exclusion:** Records published in non-academic sources (e.g., newspapers, magazines, websites) were excluded (n = 18). 4. **Full-Text Screening:** Following the initial screening, 110 records remained. Full-text screening was conducted to further narrow the scope, resulting in the exclusion of 41 records that fell outside the review's focus. An additional nine records were excluded due to limited access (subscription-based). 5. **Final Inclusion:** A total of 60 records met the inclusion criteria and were selected for detailed analysis. Each included record was meticulously assessed for specific information: 1. **Record Type:** The type of record (preprint, research article, editorial, commentary, etc.) was identified. 2. **Benefits/Applications:** The listed benefits or applications of ChatGPT in healthcare education, research, or practice were documented. 3. **Risks/Concerns:** The identified risks or concerns associated with ChatGPT in healthcare education, research, or practice were noted. 4. **Conclusions/Recommendations:** The main conclusions and recommendations regarding ChatGPT's use in healthcare were summarized. The identified benefits and risks were categorized for a clear understanding of their scope and nature.

The analysis of the 60 included records revealed a significant consensus regarding ChatGPT's potential benefits and limitations in healthcare. Here's a summary of the key findings: **Benefits of ChatGPT in Healthcare:** 1. **Academic Writing:** ChatGPT emerged as a valuable tool for scientific writing, promoting efficiency and versatility in generating high-quality text. It aids in improving language, readability, and translation, ultimately fostering research equity by enabling researchers with varying language skills to communicate their findings effectively. 2. **Scientific Research:** ChatGPT's role in scientific research was significant, particularly in analyzing large datasets, such as electronic health records or genomic data. Its ability to generate computer codes and accelerate literature reviews frees up valuable time for researchers to focus on experimental design, drug discovery, and development. 3. **Healthcare Practice:** ChatGPT holds promise for enhancing healthcare practice through personalized medicine, disease risk and outcome prediction, and efficient workflow optimization. Its potential to improve diagnostics, documentation, and cost-effectiveness in healthcare delivery is noteworthy. 4. **Healthcare Education:** ChatGPT can be a powerful tool in healthcare education. It excels in generating realistic and diverse clinical vignettes, facilitating personalized learning experiences, and providing immediate feedback to students. ChatGPT can also serve as an adjunct to group learning. **Concerns Regarding ChatGPT in Healthcare:** 1. **Ethical Issues:** Ethical concerns surrounding ChatGPT's use are pervasive. Key concerns include potential bias introduced by the training datasets, plagiarism risk, and data privacy and security issues. 2. **Inaccurate Information:** The risk of generating inaccurate or incorrect information remains a significant concern, especially in healthcare settings where incorrect information can lead to detrimental consequences. 3. **Transparency Issues:** ChatGPT's black-box nature, lacking transparency in its content generation process, raises concerns about its reliability and interpretability. 4. **Legal Issues:** Legal issues surrounding ChatGPT's use include copyright infringement, authorship status, and potential medico-legal implications in case of medical errors. 5. **Limited Knowledge:** ChatGPT's knowledge cutoff at 2021 limits its ability to provide up-to-date information, which is essential for healthcare research and practice. 6. **Risk of Misinformation:** There is a risk of ChatGPT generating misinformation and contributing to infodemics, potentially leading to harmful consequences for individuals and public health. 7. **Lack of Originality:** ChatGPT's tendency to produce over-detailed, repetitive content raises concerns about originality and the need for careful supervision. 8. **Cybersecurity Issues:** Cybersecurity concerns related to data governance, health information privacy, and potential cyberattacks are paramount in the context of ChatGPT's integration into healthcare.

The potential impact of ChatGPT on healthcare education, research, and practice is multifaceted and profound. While ChatGPT offers significant advantages, its adoption requires careful consideration of its limitations. This review highlights the need for a balanced approach, acknowledging ChatGPT's potential to revolutionize healthcare while proactively addressing the associated concerns. ChatGPT's potential to enhance scientific writing and streamline research processes is undeniable. However, the risk of generating inaccurate information, potential bias, and ethical considerations necessitates robust safeguards and guidelines. The use of ChatGPT in healthcare research should be transparently disclosed and documented, ensuring accountability for the generated content. In healthcare practice, ChatGPT's potential to improve efficiency, personalized medicine, and patient education is exciting. However, concerns about accuracy, transparency, and legal issues need to be addressed. Healthcare professionals should use ChatGPT as an adjunct tool, complementing their expertise and judgment. The "ChatGPT in the Loop: Humans in Charge" motto emphasizes the importance of human involvement in decision-making and oversight. In healthcare education, ChatGPT has the potential to revolutionize learning methods and assessment tools. Its ability to generate personalized learning experiences and assess student knowledge can be transformative. However, the risk of academic dishonesty and the need to cultivate critical thinking skills necessitate careful integration of ChatGPT into educational practices.

The widespread adoption of LLMs, including ChatGPT, within healthcare is inevitable. The significant potential benefits of ChatGPT, such as enhanced scientific writing, efficient research, streamlined healthcare workflows, and personalized learning experiences, are undeniable. However, the identified concerns, encompassing ethical issues, copyright infringements, transparency, legal aspects, potential bias, plagiarism, inaccuracies, limited knowledge, and infodemic risks, cannot be ignored. To mitigate these risks, a code of ethics and responsible use guidelines for ChatGPT and other LLMs are urgently needed. This review underscores the importance of a science-driven debate about ChatGPT's pros and cons, ensuring that its potential benefits are weighed against the risks of misleading results and research misconduct. The "ChatGPT in the Loop: Humans in Charge" philosophy highlights the crucial role of human expertise and oversight in ensuring safe and ethical integration of AI technologies in healthcare.

The findings of this systematic review should be interpreted in light of certain limitations. The review's reliance on English records may have introduced selection bias. While the review included preprints, these have not been peer-reviewed, potentially affecting the generalizability of the findings. The rapid evolution of ChatGPT applications and the limited search period (concluding on February 16, 2023) may necessitate further studies and reviews to account for new developments. The review was conducted by a single author, which could have influenced the interpretation of the results. Future reviews involving collaborative work may improve the quality and credibility of the findings.