Background: ChatGPT (Contextual Hierarchical Attention Graphic Based Programming tool) has garnered enormous attention of healthcare practitioners, researchers and academicians. This systematic review aims to provide a critical evidence-based evaluation of the current state of ChatGPT including its applications, merits, limitations and future prospects within this domain.
Materials and Methods: A systematic review was conducted following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Comprehensive searches were performed in PubMed via Medline, Science Direct, Cochrane Library, Cochrane Central Register of Controlled Trials (CENTRAL), LILACS, SciELO, ClinicalTrials.gov and Opengrey.eu without any time or language barriers.
Results: Of the 712 studies screened, 44 met the inclusion criteria. Quantitative studies (n=26) were evaluated for methodological rigour, results, coherence and relevance using the AXIS (Appraisal tool for cross-sectional studies) tool while qualitative studies (n=18) were assessed using the GRADE-CERQual (Confidence in the Evidence from Reviews of Qualitative Research) approach.
Conclusions: The findings indicate a rapidly evolving body of low to moderate level evidence regarding applications of ChatGPT in healthcare including medical practice, education, scientific research, scientific writing, clinical decision making and virtual consultation. Limitations such as hallucination, plagiarism, lack of contextual understanding and ethical concerns related to bias, data privacy, scientific misconduct, human autonomy and authorship were evident, highlighting need for further research in generative AI, coupled with development of policy guidelines and robust ethical and legal frameworks.
Keywords ChatGPT, Healthcare, Systematic review, Artificial Intelligence, Natural language processing, Large language model
