Zong Nansu, Chowdhury Shaika, Zhou Shibo, Rajaganapathy Sivaraman, Yu Yue, Wang Liewei, Dai Qiying, Li Pengyang, Liu Xiaoke, Bielinski Suzette J, Chen Jun, Chen Yongbin, Cerhan James R
Department of Artificial Intelligence and Informatics Research, Mayo Clinic, Rochester, MN, USA.
Department of Quantitative Health Sciences, Mayo Clinic, Rochester, MN, USA.
NPJ Digit Med. 2025 May 24;8(1):306. doi: 10.1038/s41746-025-01705-z.
The complexities inherent in EHR data create discrepancies between real-world evidence and RCTs, posing substantial challenges in determining whether a treatment is likely to have a beneficial impact compared to the standard of care in RCTs. The objective of this study is to enhance the prediction of efficacy direction for repurposed drugs tested in RCTs for heart failure (HF). To achieve this, we propose an efficacy direction prediction framework that integrates drug-target predictions with EHR-based Emulation Trials (ET) to derive surrogate endpoints for prediction using HF prognostic markers. Our validation of the proposed novel drug-target prediction model against the BETA benchmark demonstrates superior performance, surpassing existing baseline algorithms. Furthermore, an evaluation of our framework in identifying 17 repurposed drugs-derived from 266 phase 3 HF RCTs-using data from 59,000 patients at the Mayo Clinic highlights its remarkable predictive accuracy.
电子健康记录（EHR）数据固有的复杂性导致真实世界证据与随机对照试验（RCT）之间存在差异，这给确定一种治疗方法与RCT中的标准治疗相比是否可能产生有益影响带来了重大挑战。本研究的目的是提高对在心力衰竭（HF）的RCT中测试的重新利用药物的疗效方向预测。为实现这一目标，我们提出了一种疗效方向预测框架，该框架将药物靶点预测与基于EHR的模拟试验（ET）相结合，以使用HF预后标志物得出用于预测的替代终点。我们针对BETA基准对所提出的新型药物靶点预测模型进行的验证显示出卓越的性能，超过了现有的基线算法。此外，我们使用梅奥诊所59000名患者的数据，对我们的框架在识别源自266项3期HF RCT的17种重新利用药物方面进行的评估突出了其显著的预测准确性。
