| dc.description.abstract | PTSD is a psychiatric condition experienced by individuals after exposure to a traumatic event such as war. PTSD is a major public health concern in the United States since it is among one of the most prevalent mental disorders. Over 6% of the U.S. population suffer from this condition at any given time. PTSD has serious consequences including (but not limited to) depression and anxiety which will lead to avoidance, intrusive thoughts and hyperarousal. Hyperarousal symptoms include hypervigilance, feelings of irritability, and an exaggerated startle response following a startling event.
PTSD mostly has been assessed using subjective methods such as surveys and questionnaires. Although these methods are promising for PTSD diagnosis, they lack the capability of detecting the onset of symptoms (e.g., hyperarousal). Capturing hyperarousal events is specifically crucial because individuals may experience the most intense moments of their lives during these events when they are not with their clinicians. Therefore, there is a vital need to monitor hyperarousal events and provide timely feedback for individuals.
In this research I tried to address this gap by 1. statistically understanding hyperarousal events, 2. detecting them using machine learning algorithms, and 3. creating an actual tool that individuals who have PTSD can use to monitor their events. To do so, I used heart rate since heart rate is the main physiological indicator of PTSD. In chapter 2, I created a framework that can be used to analyze heart rate in response to PTSD. In chapter 3 I used the framework to investigate specific heart rate patterns associated with hyperarousal events. In chapter 4, I used these patterns along with a few other heart rate and body acceleration features to develop a machine learning algorithm that can detect hyperarousal events in real time. Finally, in chapter 5, I validated the developed algorithm in naturalistic settings to investigate the real world application of such algorithms. Altogether, this research presents a tool that can predict hyperarousal events in real time and has real-world validity. | |
