Effective and rapid decision-making from randomized controlled trials (RCTs) requires unbiased and precise treatment effect inferences. Two strategies to address this requirement are to adjust for covariates that are highly correlated with the outcome, and to leverage historical control information via Bayes' theorem. We propose a new Bayesian prognostic covariate adjustment methodology, referred to as Bayesian PROCOVA, that combines these two strategies. Covariate adjustment is based on generative artificial intelligence (AI) algorithms that construct a digital twin generator (DTG) for RCT participants. The DTG is trained on historical control data and yields a digital twin (DT) probability distribution for each participant's control outcome. The expectation of the DT distribution defines the single covariate for adjustment. Historical control information are leveraged via an additive mixture prior with two components: an informative prior probability distribution specified based on historical control data, and a non-informative prior distribution. The weight parameter in the mixture has a prior distribution as well, so that the entire additive mixture prior distribution is completely pre-specifiable and does not involve any information from the RCT. We establish an efficient Gibbs algorithm for sampling from the posterior distribution, and derive closed-form expressions for the posterior mean and variance of the treatment effect conditional on the weight parameter, of Bayesian PROCOVA. We evaluate the bias control and variance reduction of Bayesian PROCOVA compared to frequentist prognostic covariate adjustment (PROCOVA) via simulation studies that encompass different types of discrepancies between the historical control and RCT data. Ultimately, Bayesian PROCOVA can yield informative treatment effect inferences with fewer control participants, accelerating effective decision-making.

翻译：随机对照试验（RCT）中有效且快速的决策需要无偏且精确的治疗效应推断。实现这一需求的两项策略是：调整与结局高度相关的协变量，以及通过贝叶斯定理利用历史对照组信息。我们提出一种新的贝叶斯预后协变量调整方法，称为贝叶斯PROCOVA，该方法将上述两种策略相结合。协变量调整基于生成式人工智能（AI）算法，该算法为RCT受试者构建数字孪生生成器（DTG）。DTG利用历史对照数据进行训练，并为每位受试者的对照结局生成数字孪生（DT）概率分布。DT分布的期望值定义了用于调整的单一协变量。历史对照信息通过加性混合先验加以利用，该先验包含两个分量：基于历史对照数据指定的信息性先验概率分布，以及非信息性先验分布。混合模型中的权重参数也具有先验分布，因此整个加性混合先验分布可完全预先指定，且不涉及RCT的任何信息。我们建立了一种高效的吉布斯算法用于从后验分布中抽样，并推导出贝叶斯PROCOVA在给定权重参数条件下治疗效应后验均值和方差的闭式表达式。通过涵盖历史对照与RCT数据之间不同类型差异的模拟研究，我们评估了贝叶斯PROCOVA相较于频率学派预后协变量调整（PROCOVA）的偏倚控制与方差缩减效果。最终，贝叶斯PROCOVA能够在减少对照组受试者的条件下得到信息丰富的治疗效应推断，从而加速有效决策。