Show simple item record

dc.contributor.advisorUna-May O'Reilly and Erik Hemberg.en_US
dc.contributor.authorRivera, Steven Anthonyen_US
dc.contributor.otherMassachusetts Institute of Technology. Department of Electrical Engineering and Computer Science.en_US
dc.date.accessioned2018-01-12T20:56:51Z
dc.date.available2018-01-12T20:56:51Z
dc.date.copyright2017en_US
dc.date.issued2017en_US
dc.identifier.urihttp://hdl.handle.net/1721.1/113114
dc.descriptionThesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2017.en_US
dc.descriptionThis electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.en_US
dc.descriptionCataloged from student-submitted PDF version of thesis.en_US
dc.descriptionIncludes bibliographical references (pages 101-104).en_US
dc.description.abstractBeatDB is a framework for fast processing and analysis of physiological data, such as arterial blood pressure (ABP) or electrocardiograms (ECG). BeatDB takes such data as input and processes it for machine learning analytics in multiple stages. It offers both beat and onset detection, feature extraction for beats and groups of beats over one or more signal channels and over the time domain, and an extraction step focused on finding condition windows and aggregate features within them. BeatDB has gone through multiple iterations, with its initial version running as a collection of single-use MATLAB and Python scripts run on VM instances in Open- Stack and its second version (known as PhysioMiner) acting as a cohesive and modular cloud system on Amazon Web Services in Java. The goal of this project is primarily to modify BeatDB to support multi-channel waveform data like EEG and accelerometer data and to make the project more flexible to modification by researchers. Major software development tasks included rewriting condition detection to find windows in valid beat groups only, refactoring and writing new code to extract features and prepare training data for multi-channel signals, and fully redesigning and reimplementing BeatDB within Python, focusing on optimization and simplicity based on probable use cases of BeatDB. BeatDB v3 has become more accurate in the datasets it generates, usable for both developer and non-developer users, and efficient in both performance and design than previous iterations, achieving an average AUROC increase of over 4% when comparing specific iterations.en_US
dc.description.statementofresponsibilityby Steven Anthony Rivera.en_US
dc.format.extent104 pagesen_US
dc.language.isoengen_US
dc.publisherMassachusetts Institute of Technologyen_US
dc.rightsMIT theses are protected by copyright. They may be viewed, downloaded, or printed from this source but further reproduction or distribution in any format is prohibited without written permission.en_US
dc.rights.urihttp://dspace.mit.edu/handle/1721.1/7582en_US
dc.subjectElectrical Engineering and Computer Science.en_US
dc.titleBeatDB v3 : a framework for the creation of predictive datasets from physiological signalsen_US
dc.title.alternativeBeatDB version three : a framework for the creation of predictive datasets from physiological signalsen_US
dc.title.alternativeFramework for the creation of predictive datasets from physiological signalsen_US
dc.typeThesisen_US
dc.description.degreeM. Eng.en_US
dc.contributor.departmentMassachusetts Institute of Technology. Department of Electrical Engineering and Computer Science.en_US
dc.contributor.departmentMassachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
dc.identifier.oclc1016449801en_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record