Online Learning Based System Sample Clauses

Online Learning Based System. Our goal is to provide a generic information extraction and classification framework that is adaptable to diverse clinical reports, enables a dynamic interaction between the human and machine, and produces highly accurate results with minimal human effort. We have developed a system, IDEAL-X (Information and Data Extraction using Adaptive Online Learning), to support adaptive information extraction from diverse clinical reports with heterogeneous structures and vocabularies. A demo video can be found in YouTube [8]. IDEAL-X takes an online machine learning approach that integrates machine learning with interactive human intervention, and combines it with customizable vocabularies. The data extraction and classification engines can automatically predict answers to annotate or classify reports, gradually learn from human feedback, and iteratively improve its accuracy. The online learning algorithms [9-11] make predictions one report at a time, and utilize human feedback to update its prediction model immediately. Compared to traditional batch training based algorithm, which requires training with large volumes of carefully annotated data before deploying a system, online learning based algorithms can avoid costly expense of preparing training data and also render the possibility of updating the deployed system in a dynamically changing data environment. The online machine learning based approach is enabled through an intuitive interface and workflow. The system provides an interactive interface for users to view and annotate reports, and to verify answers generated by the data extraction or classification engine. When a user makes manual annotations or revises system-generated answers on a report, IDEAL-X learns important linguistic features and patterns of the values to be extracted in subsequent reports. The process is repeated, and the knowledge accrued by the system helps it to improve the extraction or classification accuracy. In turn, this reduces the user’s effort over time. Besides online machine learning, IDEAL-X allows for customizable controlled vocabularies to support data extraction from clinical reports, where a vocabulary enumerates the possible values that can be extracted for a given attribute. (The X in IDEAL-X represents the controlled vocabulary plug-in.) The use of online machine learning and controlled vocabularies is not mutually exclusive; they are complementary, which provide the user with a variety of modes for working with IDEAL-X.