Department of Artificial Intelligence & Informatics, Mayo Clinic, Rochester, MN, United States.
Department of Quantitative Health Sciences, Mayo Clinic, Rochester, MN, United States.
J Biomed Inform. 2022 Nov;135:104202. doi: 10.1016/j.jbi.2022.104202. Epub 2022 Sep 24.
Post-surgical complications (PSCs) have been an increasing concern for hospitals in light of Medicare penalties for 30-day readmissions. PSCs have become a target for quality improvement and benchmark for the healthcare system. Over half (60 %) of the deep or organ space surgical site infections are discovered after discharge, leading to a readmission. There has thus been a push to develop risk prediction models for targeted preventive interventions for PSCs.
We experiment several Gated Recurrent Unit with Decay (GRU-D) based deep learning architectures with various feature sampling schemes in predicting the risk of colorectal PSCs and compare with atemporal logistic regression models (logit).
We used electronic health record (EHR) data of 3,535 colorectal surgical patients involved in the national surgical quality improvement program (NSQIP) between 2006 and 2018. Single layer, stacked layer, and multimodal GRU-D models with sigmoid activation were used to develop risk prediction models. Area Under the Receiver Operating Characteristic curve (AUROC) was calculated by comparing predicted probability of developing complications versus truly observed PSCs (NSQIP adjudicated) within 30 days after surgery. We set up cross-validation and an independent held-out dataset for testing model performance consistency.
The primary contribution of our study is the formulation of a novel real-time PSC risk prediction task using GRU-D with demonstrated clinical utility. GRU-D outperforms the logit model in predicting wound and organ space infection and shows improved performance as additional data points become available. Logit model outperforms GRU-D before surgery for superficial infection and bleeding. For the same sampling scheme, there is no obvious advantage of complex architectures (stacked, multimodal) over single layer GRU-D. Obtaining more data points closer to the occurrence of PSCs is more important than using a more frequent sampling scheme in training GRU-D models. The fourth predicted risk quartile by single layer GRU-D contains 63 %, 59 %, and 66 % organ space infection cases, at 4 h before, 72 h after, and 168 h after surgery, respectively, suggesting its potential application as a bedside risk assessment tool.
鉴于医疗保险对 30 天再入院的处罚，术后并发症（PSCs）一直是医院越来越关注的问题。PSCs 已成为质量改进的目标和医疗保健系统的基准。超过一半（60%）的深部或器官间隙手术部位感染是在出院后发现的，导致再入院。因此，人们一直在努力开发针对 PSCs 的靶向预防干预措施的风险预测模型。
我们在预测结直肠 PSCs 风险时，尝试了几种具有不同特征采样方案的门控循环单元与衰减（GRU-D）深度学习架构，并与无时间逻辑回归模型（logit）进行了比较。
我们使用了 2006 年至 2018 年间参与国家手术质量改进计划（NSQIP）的 3535 例结直肠手术患者的电子健康记录（EHR）数据。使用具有 Sigmoid 激活功能的单层、堆叠层和多模态 GRU-D 模型来开发风险预测模型。通过比较术后 30 天内预测的并发症发生概率与真正观察到的 PSCs（NSQIP 裁定），计算接收者操作特征曲线下面积（AUROC）。我们设置了交叉验证和独立的保留数据集来测试模型性能的一致性。
我们的研究的主要贡献是提出了一种使用 GRU-D 对实时 PSC 风险进行预测的新任务，具有临床应用价值。GRU-D 在预测伤口和器官间隙感染方面优于 logit 模型，并且随着可用数据点的增加，性能得到提高。在手术前，logit 模型在预测浅表感染和出血方面优于 GRU-D。对于相同的采样方案，复杂架构（堆叠、多模态）并没有比单层 GRU-D 具有明显优势。在训练 GRU-D 模型时，获得更接近 PSCs 发生的更多数据点比使用更频繁的采样方案更重要。单层 GRU-D 的第四预测风险四分位数在手术前 4 小时、术后 72 小时和术后 168 小时分别包含 63%、59%和 66%的器官间隙感染病例，这表明其作为床边风险评估工具的潜在应用。
