数字孪生模拟新生儿随机对照试验对照组:概念验证可行性与可比性研究
本研究利用加拿大新生儿网络数据库,为已完成MOBYDICK试验(评估母亲补充DHA对<29周母乳喂养婴儿无支气管肺发育不良生存的影响)的对照组患者匹配1或2个数字孪生对照,评估可行性与可比性。255名实际对照新生儿均找到1个匹配,252名找到2个匹配;基线特征除数字孪生对照院外出生率更高、产前皮质类固醇使用率更低外相似,结局除数字孪生对照坏死性小肠结肠炎率更高外相似,组间差异在可接受范围内。
要点
- 利用CNN数据库可为已完成RCT的对照组生成数字孪生对照,匹配可行。
- 数字孪生对照与实际对照基线及结局总体相似,但 NEC 率更高。
- 组间差异指标在可接受范围内,支持进一步评估数字孪生对照。
核心结论
- 本研究证明为已完成随机对照试验的对照组生成数字孪生对照是可行的。
- 数字孪生对照与实际对照在基线和结局上总体相似,但坏死性小肠结肠炎率更高。
研究设计与样本
- 利用加拿大新生儿网络数据库,为MOBYDICK试验对照组的每名患者匹配1或2个数字孪生对照。
- MOBYDICK试验评估母亲补充二十二碳六烯酸对<29周母乳喂养婴儿无支气管肺发育不良生存的影响。
- 通过标准化差异、逐元素平均绝对差异和平均Gower距离评估可行性与可比性。
主要发现
- 255名实际对照新生儿均找到1个数字孪生匹配,252名找到2个匹配。
- 数字孪生对照的院外出生率更高、产前皮质类固醇使用率更低。
- 数字孪生对照的坏死性小肠结肠炎率高于实际对照。
- 逐元素平均绝对差异<0.1,Gower距离<0.3,均在可接受范围内。
关键数值与推荐
- 逐元素平均绝对差异<0.1视为可接受。
- 平均Gower距离<0.3视为可接受。
临床怎么用
- 数字孪生对照可作为历史对照的补充工具,但需注意NEC等结局差异。
- 临床解读数字孪生对照结果时应结合原始试验数据。
局限性
- 每个实际对照匹配的数字孪生数量有限(最多2个)。
- 仅基于一个已完成RCT,需在更多RCT中验证。
- 需要前瞻性同步识别数字孪生对照的进一步评估。
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: To assess the development, feasibility and comparability of an external digital twin control (DTC) arm that resembles the actual control arm of a completed trial utilizing the Canadian Neonatal Network (CNN) database. STUDY DESIGN: We used data from the control arm of a placebo-controlled trial, the MOBYDICK trial, which assessed maternal docosahexaenoic acid supplementation for bronchopulmonary dysplasia-free survival in breastfed infants of <29 weeks' gestation. We identified 1 and 2 DTC matches for each control arm patient of original trial using important inclusion criteria from the CNN database. We assessed feasibility and comparability by calculating standardized difference; the preservation of the correlation structure of groups by reporting element-wise mean absolute difference; and overall group similarity metric by calculating the mean Gower's distance. RESULTS: Among 255 actual control neonates, 1 DTC match was identified for all patients, and 2 matches were identified for 252 neonates. The baseline characteristics between 2 groups were similar except for a higher outborn rate and a lower rate of antenatal corticosteroid use in DTC. For outcomes, DTC were similar to actual controls except that DTC infants had higher rates of NEC. The element-wise mean absolute difference was <0.1 and Gower's distance was <0.3 which were within acceptable range. CONCLUSIONS: This study demonstrated the feasibility of generating DTC for the control arm of a completed RCT. Further evaluations through a greater number of DTC per actual controls; other completed RCTs; and concurrent prospective identification of DTC are needed.
引用
- Shah PS, Marc I, Guillot M, et al. Digital Twins to Mimic the Control Arm of a Neonatal Randomized Controlled Trial: A "Proof-of-Principle" Feasibility and Comparability Study. J Pediatr. 2026. PMID: 42988320. doi:10.1016/j.jpeds.2026.115385
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。