Our findings highlight that ChatGPT-4.0o displays potential for classification of urine particles, especially when allowed to generate ranked predictions and incorporate contextual parameters, but its reliance on human-provided descriptions of visual data and current inability to autonomously process images underscore the need for hybrid diagnostic approaches, encompassing human supervision.
Pighi, Laura; Negrini, Davide; Tosi, Marco; Lippi, Giuseppe,
Performance assessment of ChatGPT in classification of urine particles«Scandinavian Journal of Clinical and Laboratory Investigation»
, vol. 85
, n. 3
, 2025
, pp. 223-227