arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.02340cs.LGcs.AIcs.CL
收藏

MUFASA:电子健康记录的多模态融合体系结构搜索

MUFASA: Multimodal Fusion Architecture Search for Electronic Health Records

Google Research(谷歌研究院)

Zhen Xu, David R. So, Andrew M. Dai

摘要:

将深度学习应用于电子健康记录(EHR)的一个重要挑战是其多模态结构的复杂性。EHR通常包含结构化(代码)和非结构化(自由文本)数据的混合物,这些数据具有稀疏和不规则的纵向特征-所有这些都是医生在做出决策时使用的。在深度学习领域,确定不同的模态表示应该如何融合在一起是一个困难的问题,这通常是通过手工建模和直觉来解决的。在这项工作中,我们扩展了最先进的神经架构搜索(NAS)的方法,并提出了多模态融合架构SeArch(MUFASA),同时搜索跨多模态融合策略和模态特定的架构第一次。我们的经验表明,我们的MUFASA方法优于公共EHR数据与可比的计算成本建立单峰NAS。此外,MUFASA还生产性能优于Transformer和Evolved Transformer的架构。与CCS诊断代码预测的这些基线相比,我们发现的模型将前5名的召回率从0.88提高到0.91,并展示了推广到其他EHR任务的能力。通过深入研究我们的顶级架构,我们提供了经验证据,证明MUFASA的改进来自于其为每种数据模态定制建模并找到有效融合策略的能力。

英文摘要:

One important challenge of applying deep learning to electronic health records (EHR) is the complexity of their multimodal structure. EHR usually contains a mixture of structured (codes) and unstructured (free-text) data with sparse and irregular longitudinal features -- all of which doctors utilize when making decisions. In the deep learning regime, determining how different modality representations should be fused together is a difficult problem, which is often addressed by handcrafted modeling and intuition. In this work, we extend state-of-the-art neural architecture search (NAS) methods and propose MUltimodal Fusion Architecture SeArch (MUFASA) to simultaneously search across multimodal fusion strategies and modality-specific architectures for the first time. We demonstrate empirically that our MUFASA method outperforms established unimodal NAS on public EHR data with comparable computation costs. In addition, MUFASA produces architectures that outperform Transformer and Evolved Transformer. Compared with these baselines on CCS diagnosis code prediction, our discovered models improve top-5 recall from 0.88 to 0.91 and demonstrate the ability to generalize to other EHR tasks. Studying our top architecture in depth, we provide empirical evidence that MUFASA's improvements are derived from its ability to both customize modeling for each data modality and find effective fusion strategies.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑