Chebly Alain, Veronese Lauren, Chevret Edith
Center Jacques Loiselet for Medical Genetics and Genomics (CGGM), Faculty of Medicine, Saint Joseph University of Beirut (USJ), Beirut, Lebanon.
Université Clermont Auvergne, Unité de Recherche 7453 (CHELTER), Clermont-Ferrand, France.
Mol Cytogenet. 2025 Jul 18;18(1):16. doi: 10.1186/s13039-025-00717-4.
This study evaluates the accuracy of ChatGPT in generating chromosomal representations (formulas) based on ISCN rules in clinical cytogenetics. While ChatGPT generated correct answers for simple cases, it frequently failed in complex cases. These findings highlight the limitations of current AI tools in accurate clinical cytogenetic reporting, reinforcing the vital role of skilled clinical cytogeneticists for accurate interpretation and reporting.
本研究评估了ChatGPT在根据临床细胞遗传学中的ISCN规则生成染色体表征(公式)方面的准确性。虽然ChatGPT在简单病例中能给出正确答案,但在复杂病例中常常出错。这些发现凸显了当前人工智能工具在准确的临床细胞遗传学报告方面的局限性,强化了熟练的临床细胞遗传学家在准确解读和报告中的关键作用。