TitleUtilizing real-time location data for performance monitoring in healthcare systems
NameDemir, Cenk (author), Boucher, Thomas (chair), Elsayed, Elsayed (internal member), Luxhoj, James (internal member), Rutgers University, Graduate School - New Brunswick,
SubjectIndustrial and Systems Engineering,
Medical care--Evaluation--Mathematical models,
Medical care--Quality control--Mathematical models,
DescriptionAlthough both public and private healthcare expenditures have been very high in the US when compared to other wealthy and industrialized nations; the quality and reach of services provided have constantly been in dispute. For that reason, improving the efficiency of healthcare services has emerged to be an important goal. However, the unpredictable and complex nature of the healthcare environments makes this goal difficult to achieve. Recently industrial engineering methods started being applied to improve hospital efficiencies. In addition, the utilization of technological advancements such as RFID based real time location systems (RTLS) in the healthcare sector provides an additional opportunity to apply industrial engineering methods to healthcare.
In this thesis, a data transformation and analysis framework is developed to be employed as part of a RTLS schema in a hospital unit. This framework consists of a software agent that is capable of monitoring, analyzing and predicting the performance of a process from RTLS data. The software agent performs its task by means of several statistical methods customized for specific purposes. It can identify steady state behavior, changes in the mean and transient states such as learning curves. Its main purpose is to detect the effects of modifications on the system and forecast the future performance level. A simulation model is built to produce tracking data of entities in a hypothetical hospital unit to be fed into the software agent via a database structure. This completes the framework and enables the testing of the developed agent using realistic data.
RTLSs typically suffer from accuracy issues which result in imperfect tracking data. This makes performance monitoring very infeasible or even impossible. To overcome this issue, a data cleaning algorithm is developed which can be fully integrated with the developed performance agent. The algorithm utilizes a Bayesian approach in a sliding window analysis. Testing of the data cleaning algorithm is performed for various scenarios.
NoteIncludes bibliographical references (p. 60-61)
Noteby Cenk Demir
CollectionGraduate School - New Brunswick Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work