融合临床笔记的Transformer模型提升疾病轨迹预测准确率
Patient Trajectory Prediction: Integrating Clinical Notes with Transformers
- 用Transformer整合病历文本与结构化数据
- 在MIMIC-IV上预测准确率优于仅用结构数据的模型
- 适合关注临床文本挖掘与医疗预测的研究者
从电子健康记录（EHR）中预测疾病轨迹是一项复杂任务，主要挑战包括数据非平稳性、医疗编码粒度高以及多模态数据融合。EHR包含结构化数据（如诊断编码）和非结构化数据（如临床笔记），后者常被忽略却蕴含关键信息。现有模型多基于结构化数据，难以捕捉完整医疗背景，导致信息丢失。本文提出一种将非结构化临床笔记融入基于Transformer的深度学习模型的方法，丰富患者病史表征，从而提升诊断预测准确性。在MIMIC-IV数据集上的实验表明，该方法优于仅依赖结构化数据的传统模型。
原文摘要 · Abstract (English)
Predicting disease trajectories from electronic health records (EHRs) is a complex task due to major challenges such as data non-stationarity, high granularity of medical codes, and integration of multimodal data. EHRs contain both structured data, such as diagnostic codes, and unstructured data, such as clinical notes, which hold essential information often overlooked. Current models, primarily based on structured data, struggle to capture the complete medical context of patients, resulting in a loss of valuable information. To address this issue, we propose an approach that integrates unstructured clinical notes into transformer-based deep learning models for sequential disease prediction. This integration enriches the representation of patients' medical histories, thereby improving the accuracy of diagnosis predictions. Experiments on MIMIC-IV datasets demonstrate that the proposed approach outperforms traditional models relying solely on structured data.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。