An official website of the United States government
Showing 132 Results
Showing 131-132 of 132
Publications
TITLE:
Automated Histologic Grading from Free-text Pathology Reports Using Graph-of-Words Features and Machine Learning , IEEE EMBS International Conference on Biomedical and Health Informatics : 369-372 , 2017
PUBLICATION DATE: 02-16-2017
ABSTRACT: Traditional n-gram feature representation of free-text documents often fails to capture word ordering and semantics, thus compromising text comprehension. Graph-of-words, a new text representation approach based on graph analytics, is a superior method overcoming the limitations by modeling word co-occurrence. In this study, the authors presented a novel application of graph-of-words text description for automated extraction of histologic grade from unstructured pathology reports. Using 10-fold cross-validation tests, the proposed approach resulted in substantially higher macro and micro-F1 scores with undirected graph-of-words features, compared to traditional bi-gram text features. This feasibility study demonstrated that graph-of-words is a highly efficient method of text comprehension for information extraction from free-text clinical documents.
PROJECT: MOSSAIC
Go To Publications
TITLE:
Multi-task Deep Neural Networks for Automated Extraction of Primary Site and Laterality Information from Cancer Pathology Reports , Advances in Big Data : 195-204 , 2016
PUBLICATION DATE: 10-08-2016
ABSTRACT: Automated annotation of free-text cancer pathology reports is a critical challenge for cancer registries and the national cancer surveillance program. In this paper, the authors investigated deep neural networks (DNNs) for automated extraction of the primary cancer site and its laterality, two fundamental targets of cancer reporting. Their experiments showed that single-task DNNs are capable of extracting information with higher precision and recall than traditional classification methods for the more challenging target. Furthermore, a multi-task learning DNN resulted in further performance improvement. This preliminary study indicates the strong potential for multi-task deep neural networks to extract cancer-relevant information from free-text pathology reports.
Go To Publications