将人工智能整合于甲状腺结节管理:临床结局与成本效果分析
本研究回顾性分析2022年5月至2024年11月间接受甲状腺手术的176例患者,比较美国放射学会甲状腺影像报告和数据系统(ACR TI-RADS)下专家与新手判读及KOIOS人工智能(AI)风险分层的诊断表现,并计算增量成本效果比(ICER)。KOIOS的诊断表现与专家相近、优于新手,ICER为-€8085.56,提示其可能节省成本。
要点
- KOIOS与专家判读的AUC相近(0.794对0.784,P=.754),并优于新手(0.619,P<.001)。
- 在第三方支付方视角、1年时间范围内,ICER为-€8085.56。
- 作者认为KOIOS是具成本效果的替代方案,尤其可减少良性甲状腺结节的过度诊断与过度治疗。
核心结论
- KOIOS人工智能是具成本效果的替代方案,尤其可减少良性甲状腺结节的过度诊断与过度治疗。
- 作者指出需要前瞻性、真实世界研究验证这些发现并探索长期影响。
研究设计与样本
- 对2022年5月至2024年11月间接受甲状腺手术的176例患者进行回顾性分析。
- 超声图像由专家与新手操作者独立按美国放射学会甲状腺影像报告和数据系统(ACR TI-RADS)判读,KOIOS提供人工智能适配的风险分层。
- 进行了敏感度、特异度与受试者工作特征(ROC)曲线分析。
- 增量成本效果比(ICER)基于最佳诊疗干预(细针穿刺活检与甲状腺手术)的数量定义。
- 进行了确定性与概率敏感性分析以评估模型稳健性。
主要发现
- KOIOS人工智能与专家操作者的诊断表现相近(曲线下面积AUC:0.794,95%CI 0.718-0.871 对 0.784,95%CI 0.706-0.861;P=.754)。
- KOIOS人工智能显著优于新手操作者(AUC:0.619,95%CI 0.526-0.711;P<.001)。
- ICER分析估计每增加一次最佳诊疗决策的成本为-€8085.56。
- 确定性敏感性分析显示手术成本是变异的主要驱动因素,概率敏感性分析一致支持KOIOS为最优策略。
关键数值与推荐
- KOIOS人工智能AUC为0.794(95%CI 0.718-0.871),专家操作者AUC为0.784(95%CI 0.706-0.861),两组差异P=.754。
- 新手操作者AUC为0.619(95%CI 0.526-0.711),与KOIOS比较P<.001。
- 在第三方支付方视角、1年时间范围内,每增加一次最佳诊疗决策的ICER为-€8085.56。
临床怎么用
- 原文未据此给出临床推荐;作者认为KOIOS人工智能是具成本效果的替代方案,尤其可减少良性甲状腺结节的过度诊断与过度治疗。
- 作者建议开展前瞻性、真实世界研究验证这些发现并探索长期影响。
局限性
- 原文为回顾性分析。
- 原文仅纳入176例接受甲状腺手术的患者。
- 作者指出需要前瞻性、真实世界研究验证这些发现并探索长期影响。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVE: The increasing incidence of thyroid nodules (TNs) raises concerns about overdiagnosis and overtreatment. This study evaluates the clinical and economic impact of KOIOS, a Food and Drug Administration-approved artificial intelligence (AI) tool for the management of TN. METHODS: A retrospective analysis was conducted on 176 patients who underwent thyroid surgery between May 2022 and November 2024. Ultrasound images were evaluated independently by an expert and novice operators using the American College of Radiology Thyroid Imaging Reporting and Data System, while KOIOS provided AI-adapted risk stratification. Sensitivity, specificity, and receiver operating (ROC) curve analysis were performed. The incremental cost-effectiveness ratio (ICER) was defined based on the number of optimal care interventions (fine-needle aspiration biopsies and thyroid surgeries). Both deterministic and probabilistic sensitivity analyses were conducted to evaluate model robustness. RESULTS: KOIOS AI demonstrated similar diagnostic performance to the expert operator [area under the curve (AUC): 0.794, 95% confidence interval (CI): 0.718-0.871 vs 0.784, 95% CI: 0.706-0.861; P = .754] and significantly outperformed the novice operator (AUC: 0.619, 95% CI: 0.526-0.711; P < .001). ICER analysis estimated the cost per additional optimal care decision at -€8085.56, indicating KOIOS as a dominant and cost-saving strategy when considering a third-party payer perspective over a 1-year horizon. Deterministic sensitivity analysis identified surgical costs as the main drivers of variability, while probabilistic analysis consistently favored KOIOS as the optimal strategy. CONCLUSION: KOIOS AI is a cost-effective alternative, particularly in reducing overdiagnosis and overtreatment for benign TNs. Prospective, real-life studies are needed to validate these findings and explore long-term implications.
引用
- Bodoque-Cubas J, Fernández-Sáez J, Martínez-Hervás S, et al. Integrating Artificial Intelligence in Thyroid Nodule Management: Clinical Outcomes and Cost-effectiveness Analysis. J Clin Endocrinol Metab. 2026. PMID: 40650885. doi:10.1210/clinem/dgaf399
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。