Optimized integrated software package for classification of small airway dysfunction using genetic algorithms
Abstract
Due to the similarity in the symptoms of respiratory disorders, an accurate diagnosis could be a difficult task. Regarding asthma as the main focus of this research, early detection of small airway impairment could represent a better opportunity for timely diagnosis. An integrated software package (ISP) based on the impulse oscillometry technique, has been developed to serve as an important tool for clinicians during classification of asthma severities on children [1]. The current version of the ISP was tested with data obtained during several years by Dr. Homer Nazeran and Dr. Michael Goldman from 110 Hispanic children [2], providing an accuracy of 88% during the classification process. The main goal of this research project was to optimize the ISP incrementing accuracy, adaptability, efficiency as well as compatibility with the newest computer operating systems. Looking for new techniques to accomplish these goals, genetic algorithms were incorporated into the software structure as well as other refinements. The design of this integrated software package was done in modules. This structure gives the opportunity to test the software at a particular stage, working independently with any module outside the software to then be able to reintegrate it into the software package. The ISP is composed of two modules, the first module plays a very important role generating and validating the patient data by matching the simulated and real data. The data obtained from the first module is the input of the second module which performs the classification. This research project focused on developing an upgraded version the first module looking forward to improve the ISP overall performance. Genetic Algorithms (GAs) were implemented into the first module during the parameter estimation process. The software package was tested after the implementation of GAs into the structure to analyze any improvement in accuracy, adaptability and efficiency. The results confirmed an improvement of the overall software performance in terms of accuracy and efficiency. Improvements in adaptability to new patient data which could lead into a future expansion of the software to other respiratory system diseases. Finally, the new ISP was tested on the 110 patient's data giving an accuracy increase of 7% on the overall performance. This thesis summarizes the research work, the techniques and process during the upgrade and optimization of the ISP.
Subject Area
Biomedical engineering|Electrical engineering
Recommended Citation
Mata, Elias Estrada, "Optimized integrated software package for classification of small airway dysfunction using genetic algorithms" (2014). ETD Collection for University of Texas, El Paso. AAI1557757.
https://scholarworks.utep.edu/dissertations/AAI1557757