Title:  Design and Validity of Automated Impedance Cardiogram Analysis Algorithms for Clinical Application and Interpretation using Artificial Intelligence Framework

Committee: 

Dr. Clifford, Advisor

Dr. Inan, Co-Advisor     

Dr. Sarioglu, Chair

Dr. Shah

Abstract: The objective of the proposed research is to investigate the utility of the impedance cardiogram (ICG) signal to perform post-traumatic stress disorder (PTSD) physiology analysis. The sensitivity of the ICG signal to the artifacts induced by respiration, speaking, and movement, coupled with inter-subject/intra-subject morphological variability results in an inaccurate analysis. First, an automated noise-removal algorithm will be designed to provide a robust framework for the removal of noisy beats and accurate estimation of ICG-based parameters for fast, reproducible, and accurate physiologic analyses. Second, the physiological validation of the proposed algorithm will be performed in the PTSD application in the presence of speech-related artifacts to evaluate its potential clinical applications in psychophysiology. Third, a machine-learning-based automated algorithm will be designed to detect the opening of the aortic valve via the ICG signal, which will increase the accuracy and reliability of ICG-based analysis without the need for expert visual scoring. Fourth, an artificial intelligence framework will be used to find relevant important features for PTSD classification, which can assist in the comprehension of the PTSD stress physiology and help investigate possible future therapies. An ICG open-source toolbox will also be developed, which will act as a benchmark for other studies, help accelerate the field, and aid reproducibility. Although the clinical focus of this research is PTSD stress physiology, however, the designed algorithms can also be applied to any ICG-based analysis.