Out-of-Control Detection in the Disease Surveillance System: A Novel Approach Using Principal Component Analysis and Machine Learning with Radial Basis Function Neural Networks

Improperly monitored disease surveillance system produce unreliable system status, leading to public health and safety risks. The disease surveillance system collects, analyzes and interprets disease data from several sources to implement the disease spread prevention controls. Detecting an in-control system status for an actually out-of-control system will prevent the timely implementation of the corrective actions. Traditional multivariate control charts require higher average run length to detect the out-of-control behavior and are thus known to be ineffective to detect shifts with increasing number of system variables. Besides combining with the traditional multivariate control charts, neural networks for classification in process monitoring employing the back-propagation algorithm require extensive computations and may not provide correct system status classifications. Using the foodborne and waterborne disease variables, this research introduces a novel capability to reliably detect the out-of-control state in the disease surveillance system, combining the principal component analysis with the radial basis function neural network. By extracting the most important information in the data and through the neural network experimentations to reliably detect the out-of-control state, this research enables the disease surveillance system manager’s timely implementation of the corrective actions, thus avoiding the health and safety risks to the public.Keywords: Principal Component Analysis, Radial Basis Function, Neural Network, Machine Learning, Disease Surveillance Reliability, Public Health and Safety

Author

• Krishnaswamy, Sriram

Language

• en

Date created

• 2018

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Systems Engineering

Degree

• D.Engr.

Advisor

• Etamadi, Amir H

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/mg74qm36k