Mucahide Gokcen Gokalp, Turkan Calıskan, Berna Cafer Karalar
When evaluated by specialist nurses, ChatGPT-4 was found to provide generally accurate information; however, it exhibited shortcomings in the nursing job description and remained limited in patient-specific care.
BACKGROUND: The aim of this study was to compare ChatGPT-4-generated responses with expert nurse opinions regarding nursing care after lumbar disk surgery and to assess the adequacy of these responses in supporting nursing knowledge and practice.
METHODS: This descriptive and comparative study used a 10-question form developed by the researchers from a literature review on nursing care after lumbar disk surgery. Ten nurses rated the accuracy of the responses generated by ChatGPT using a 5-point Likert scale (5: strongly agree, 1: strongly disagree). The maximum possible score on the rating scale was 50, and the minimum was 10.
RESULTS: Specialist nurses rated ChatGPT-4 responses highly in terms of pain management strategies (4.0 ± 1.41), prevention of bowel and bladder complications (4.0 ± 1.41), and monitoring of drains and dressings (4.5 ± 0.70). However, the expert panel reported that ChatGPT-4 provided partially inappropriate or incomplete information regarding discharge education (3.0 ± 1.41) and patient mobilization timing (3.0 ± 1.41).
CONCLUSION: When evaluated by specialist nurses, ChatGPT-4 was found to provide generally accurate information; however, it exhibited shortcomings in the nursing job description and remained limited in patient-specific care.