Topic Modeling on Triage Notes With Semiorthogonal Nonnegative Matrix Factorization
Topic Modeling on Triage Notes With Semiorthogonal Nonnegative Matrix Factorization
复制标题
DOI：
10.1080/01621459.2020.1862667
复制
发表时间：
2020-12
影响因子：
3.7
通讯作者：
Yutong Li;Ruoqing Zhu;A. Qu;Han Ye;Zhankun Sun
中科院分区：
文献类型：
--
作者：
Yutong Li;Ruoqing Zhu;A. Qu;Han Ye;Zhankun Sun
摘要 急诊科（ED）拥挤是一个普遍的健康问题，影响医院管理效率和患者护理质量。当患者申请病床的请求被推迟到医生做出入院决定时，急诊室拥挤的情况经常发生。在本案例研究中，我们构建了一个分类器，使用阿尔伯塔医疗中心提供的分诊过程中收集的手动输入的护士笔记来预测患者的处置情况。这些预测有可能被纳入早期床位协调和快速通道流策略中，以缓解急诊室的过度拥挤和等待时间。然而，这些分类笔记涉及高维、噪声和稀疏的文本数据，这使得模型拟合和解释变得困难。为了解决这个问题，我们为连续和二元预测器提出了一种新颖的半正交非负矩阵分解，以降低维度并导出单词主题。然后，分类注释可以被解释为正交基主题向量的非减线性组合。我们的真实数据分析表明，分诊记录包含强有力的预测信息，可对患者对某些医疗投诉（例如意识改变或中风）的处置进行分类。此外，我们还表明，通过我们的方法生成的文档主题向量可以用作特征，以进一步将不同医疗投诉的分类准确度提高 1%，例如，对于有中风症状的患者，分类准确度为 74.3%–75.3%。这种改进可能对某些患者产生临床影响，特别是当医院患者规模很大时。此外，由于正交公式，生成的词主题向量在每个主题下提供了双聚类解释，这有利于医院更好地了解患者就诊的症状和原因。本文的补充材料可在线获取。
Abstract Emergency department (ED) crowding is a universal health issue that affects the efficiency of hospital management and patient care quality. ED crowding frequently occurs when a request for a ward-bed for a patient is delayed until a doctor makes an admission decision. In this case study, we build a classifier to predict the disposition of patients using manually typed nurse notes collected during triage as provided by the Alberta Medical Center. These predictions can potentially be incorporated to early bed coordination and fast track streaming strategies to alleviate overcrowding and waiting times in the ED. However, these triage notes involve high dimensional, noisy, and sparse text data, which make model-fitting and interpretation difficult. To address this issue, we propose a novel semiorthogonal nonnegative matrix factorization for both continuous and binary predictors to reduce the dimensionality and derive word topics. The triage notes can then be interpreted as a non-subtractive linear combination of orthogonal basis topic vectors. Our real data analysis shows that the triage notes contain strong predictive information toward classifying the disposition of patients for certain medical complaints, such as altered consciousness or stroke. Additionally, we show that the document-topic vectors generated by our method can be used as features to further improve classification accuracy by up to 1% across different medical complaints, for example, 74.3%–75.3% accuracy for patients with stroke symptoms. This improvement could be clinically impactful for certain patients, especially when the scale of hospital patients is large. Furthermore, the generated word-topic vectors provide a bi-clustering interpretation under each topic due to the orthogonal formulation, which can be beneficial for hospitals in better understanding the symptoms and reasons behind patients’ visits. Supplementary materials for this article are available online.