Enhancing nurse productivity and patient safety through generative AI integration

Generative artificial intelligence (AI) is increasingly influencing the healthcare delivery and providing transformative opportunities for the nurses. With increased challenges faced by the healthcare systems, such as growing patient demands, complex care needs, and resource constraints, AI offers promising solutions to improve efficiency, accuracy, and patient outcomes (Reddy et. al, 2020). This presentation demonstrates how generative AI can be leveraged to enhance nurse productivity and patient safety through prescription review simulations. A realistic synthetic dataset was created to simulate doctors’ prescriptions for hypothetical patients typically admitted to a medical-surgical unit in a large acute care hospital in the USA. The dataset included diagnoses or procedures, treatments, drug names, dosages, routes, and frequencies for each diagnosis and procedure. Several intentional errors were introduced into selected records, including incorrect medications, doses, routes, frequencies, and drug interactions to assess error detection capacity of generative AI. This study used two commonly used large language models (LLMs), ChatGPT 4.0 and Gemini, to review prescription records and identify potential medication errors. For each flagged record, it was also asked to provide detailed clinical reasoning, focusing on patient safety, pharmacology, and clinical best practices. The accuracy of generative AI in detecting introduced errors was evaluated in terms of both precision and recall. The study results suggest that using generative AI-powered simulations is an effective way to detect medication errors and support nurses’ learning. This study simulated 20 patients for various diagnosis and treatments. There were total of 62 drugs and supportive treatments. In each replication, four to eight errors were introduced. All errors were correctly identified by both ChatGPT and Gemini except one instance where both generative AI failed to point potential contraindication of medications. Since generative AI could not explicitly call out dangerous combination of three CNS depression, it highlights the importance of careful review from the medical provider and the limitations of AI system in fully replicating expert clinical judgement. This study highlights that having a human in the loop is essential, as generative AI technology is maturing and not yet completely free from hallucinations. With improved efficiency in reviewing prescriptions observed in this study, generative AI has promising implications for nursing education and promoting a safe healthcare environment. Nurses can learn with AI because AI acts as a real time mentor, guide and tool that allows nurses’ critical thinking skills, error reduction, workflow efficiency, clinical decision support, and professional growth. Although integration of AI has huge potential to make a positive impact in healthcare, there are several challenges. Collaboration among the healthcare providers, educators, policy makers, and technology developers are essential to overcome those challenges. While integrating AI, it is important to maintain the core value of nursing such as compassion, patient centered care, advocacy, and ethical practice (Wei, Pan, Liu, Hong, Nong C Zhang, 2025).