In a recent review article in npj Digital MedicineIn a systematic review, researchers examined the ethical implications of using Large Language Models (LLMs) in healthcare.
Their conclusions show that while LLMs offer significant benefits such as improved data analysis and decision support, ongoing ethical concerns about fairness, bias, transparency, and privacy underscore the need for clearly defined ethical guidelines and human oversight in their application.
Study: The ethics of ChatGPT in medicine and healthcare: a systematic review of Large Language Models (LLMs)Photo credit: Summit Art Creations/Shutterstock.com
background
LLMs have attracted a lot of interest due to their advanced capabilities in artificial intelligence (AI), which has become evident since OpenAI released ChatGPT in 2022.
This technology has spread rapidly across various sectors, including medicine and healthcare, and shows promising applications in clinical decision-making, diagnosis and patient communication.
But alongside the potential benefits, concerns have also arisen about the ethical implications. Previous research has highlighted risks such as the spread of inaccurate medical information, privacy breaches from handling sensitive patient data, and the perpetuation of biases based on gender, culture or race.
Despite these concerns, there is a notable gap in comprehensive studies that systematically address the ethical challenges of integrating LLMs into healthcare. The existing literature focuses on specific cases rather than providing a holistic overview.
Methods
Since strict ethical standards and regulations are required in the healthcare sector, it is imperative that the existing gaps in this area are closed.
In this systematic review, researchers mapped the ethical landscape surrounding the role of LLMs in healthcare to identify potential benefits and harms to inform future discussions, strategies, and policies on the ethical regulation of LLM use.
The researchers designed a review protocol on practical applications and ethical aspects, which was registered in the International Prospective Register of Systematic Reviews. Ethics approval was not required.
They searched relevant publication databases and preprint servers to collect data, considering preprints because they are widely used in technology fields and have potential relevance that is not yet indexed in databases.
Inclusion criteria were based on intervention, setting, and outcomes. There were no restrictions on the type of publication, but papers that dealt exclusively with medical education or scientific writing were excluded.
After an initial screening of titles and abstracts, data were extracted and coded using a structured form. The quality assessment focused descriptively on procedural quality criteria to distinguish peer-reviewed materials and critically assessed the results in terms of validity and completeness when reporting.
Results
The study analyzed 53 articles to examine the ethical implications and applications of LLMs in healthcare. Four main themes emerged from the review: clinical applications, patient support applications, support for healthcare professionals, and public health perspectives.
In clinical applications, LLMs have potential to assist in initial diagnosis and triage of patients by using predictive analytics to identify health risks and provide treatment recommendations.
However, there are concerns about their accuracy and the possibility of bias in their decision-making processes. This bias could lead to incorrect diagnoses or treatment recommendations and highlights the need for careful monitoring by healthcare professionals.
Patient support applications focus on LLMs that help individuals access medical information, manage symptoms, and navigate healthcare systems.
While LLMs can improve health literacy and communication across language barriers, privacy and the reliability of medical advice generated by these models remain important ethical concerns.
LLMs support healthcare professionals and are designed to automate administrative tasks, consolidate patient interactions, and facilitate medical research.
While this automation could increase efficiency, there are concerns about the impact on professional skills, the integrity of research findings, and the potential for bias in automated data analysis.
From a public health perspective, LLMs offer the opportunity to monitor disease outbreaks, improve access to health information, and strengthen public health communication.
However, the study also points to risks, such as the spread of misinformation and the concentration of AI power in the hands of a few companies, potentially exacerbating health inequalities and undermining public health efforts.
Overall, while LLMs represent promising advances in healthcare, their ethical use requires careful balancing of bias, privacy concerns, and the need for human oversight to mitigate potential harm and ensure equitable access and patient safety.
Conclusions
The researchers found that LLMs like ChatGPT are being widely studied in healthcare because of their potential to improve efficiency and patient care by quickly analyzing large data sets and providing personalized information.
However, ethical concerns remain, including bias, transparency issues and the generation of misleading information, so-called hallucinations, which can have serious consequences in the clinical setting.
The study is in line with broader research on AI ethics and highlights the complexity and risks of using AI in healthcare.
The strengths of this study include a comprehensive literature review and a structured categorization of LLM applications and ethical issues.
Limitations include the evolving nature of ethical review in this field, the reliance on preprint sources, and the predominance of perspectives from North America and Europe.
Future research should focus on defining robust ethical guidelines, improving algorithm transparency, and ensuring equitable use of LLMs in the global health context.
