Department of Neurology, Boston University School of Medicine, 85 E. Concord St., Suite 1116, Boston, MA, 02118, USA.
Saïd Business School, University of Oxford, Oxford, UK.
Neurocrit Care. 2022 Aug;37(Suppl 2):291-302. doi: 10.1007/s12028-022-01513-3. Epub 2022 May 9.
Abstraction of critical data from unstructured radiologic reports using natural language processing (NLP) is a powerful tool to automate the detection of important clinical features and enhance research efforts. We present a set of NLP approaches to identify critical findings in patients with acute ischemic stroke from radiology reports of computed tomography (CT) and magnetic resonance imaging (MRI).
We trained machine learning classifiers to identify categorical outcomes of edema, midline shift (MLS), hemorrhagic transformation, and parenchymal hematoma, as well as rule-based systems (RBS) to identify intraventricular hemorrhage (IVH) and continuous MLS measurements within CT/MRI reports. Using a derivation cohort of 2289 reports from 550 individuals with acute middle cerebral artery territory ischemic strokes, we externally validated our models on reports from a separate institution as well as from patients with ischemic strokes in any vascular territory.
In all data sets, a deep neural network with pretrained biomedical word embeddings (BioClinicalBERT) achieved the highest discrimination performance for binary prediction of edema (area under precision recall curve [AUPRC] > 0.94), MLS (AUPRC > 0.98), hemorrhagic conversion (AUPRC > 0.89), and parenchymal hematoma (AUPRC > 0.76). BioClinicalBERT outperformed lasso regression (p < 0.001) for all outcomes except parenchymal hematoma (p = 0.755). Tailored RBS for IVH and continuous MLS outperformed BioClinicalBERT (p < 0.001) and linear regression, respectively (p < 0.001).
Our study demonstrates robust performance and external validity of a core NLP tool kit for identifying both categorical and continuous outcomes of ischemic stroke from unstructured radiographic text data. Medically tailored NLP methods have multiple important big data applications, including scalable electronic phenotyping, augmentation of clinical risk prediction models, and facilitation of automatic alert systems in the hospital setting.
使用自然语言处理（NLP）从非结构化放射报告中提取关键数据是一种强大的工具，可以自动检测重要的临床特征并增强研究工作。我们提出了一系列 NLP 方法，以从计算机断层扫描（CT）和磁共振成像（MRI）的放射学报告中识别急性缺血性脑卒中患者的关键发现。
我们训练机器学习分类器来识别水肿、中线移位（MLS）、出血转化和实质血肿的分类结果，以及基于规则的系统（RBS）来识别 CT/MRI 报告中的脑室内出血（IVH）和连续 MLS 测量值。使用来自 550 名急性大脑中动脉区域缺血性脑卒中患者的 2289 份报告的推导队列，我们在来自另一家机构的报告以及任何血管区域的缺血性脑卒中患者的报告上对我们的模型进行了外部验证。
在所有数据集上，具有预训练生物医学词嵌入的深度神经网络（BioClinicalBERT）在水肿（精度召回曲线下面积[AUPRC]＞0.94）、MLS（AUPRC＞0.98）、出血转化（AUPRC＞0.89）和实质血肿（AUPRC＞0.76）的二进制预测方面实现了最高的区分性能。BioClinicalBERT 在除实质血肿外的所有结果上均优于套索回归（p＜0.001）（p＝0.755）。针对 IVH 和连续 MLS 的定制 RBS 分别优于 BioClinicalBERT（p＜0.001）和线性回归（p＜0.001）。
我们的研究表明，一种核心 NLP 工具包在从非结构化放射学文本数据中识别缺血性脑卒中的分类和连续结果方面具有强大的性能和外部有效性。针对医学的 NLP 方法具有多个重要的大数据应用，包括可扩展的电子表型、临床风险预测模型的增强以及在医院环境中促进自动警报系统。
