Dissertation Defense Announcements

Individuals with chronic ankle instability (CAI) often demonstrate abnormal biomechanical patterns which may be associated with the onset of posttraumatic osteoarthritis (PTOA). There is a lack of interventions capable of changing gait impairments and targeting the sensorimotor dysfunction in this population. The use of external focus of attention biofeedback during walking has shown to be capable of improving biomechanics in real-time, however, the retention of these changes needs to be further explored prior to implementing into clinical practice. Further, technological advancements have shown ultrasonography of the talar articular cartilage is a plausible mechanism to monitor joint health, yet the associations of ultrasound-based images and biomechanical patterns in patients with CAI remains unknown. This dissertation focused on three main research questions: 1) Does incorporating an auditory biofeedback device during common rehabilitation exercises improve biomechanics during a single session, 2) Is there a relationship between abnormal walking biomechanics in patients with CAI and ultrasound-based measures of talar articular cartilage health, 3) Does incorporating an auditory biofeedback device into a 2-week gait training program improve biomechanics and ankle joint health in patients with CAI compared to a control condition. First, we identified auditory biofeedback is effective at improving biomechanics during functional exercises in addition to its improvements in gait. Next, we found significant correlations between abnormal walking biomechanics, talar cartilage thickness, and echo intensity. Finally, our 2-week gait training intervention has shown to be effective in improving walking gait in patients with CAI immediately and up to 1 week following the intervention. The findings of this dissertation have found the relationship between biomechanical gait patterns and ankle joint health which will guide clinical practice in identifying modifiable factors to potentially mitigate the onset of PTOA after an ankle injury. This dissertation has also identified a clinically applicable mechanism to improve the abnormal biomechanics that are related to cartilage joint health.

Romantic relationships have an impact on both partners health and well-being, however, much remains unknown regarding how health behaviors are shaped in the early stages of dating. The developmental time of emerging adulthood targets an age group where romantic relationships and alcohol use commonly intersect and may contribute to lifelong patterns of use. This study utilized novel research methodology of combining new couple’s text messages during the early stages of dating (<6 months) with survey data as well as advanced statistical methods (i.e., Actor Partner Interdependence Model) to examine how emerging adult couples text messaging about alcohol early in their romantic relationships impacted each partner intra- and interpersonally in terms of alcohol use and attitudes. Additionally, using the moderated Actor-Partner Interdependence Model this study examined how relationship factors (i.e., relationship power and relationship satisfaction) linked to health outcomes and engagement in risk behavior moderated these relationships. Results indicated that text messages between partners about alcohol were significantly positively correlated with alcohol use, but not attitudes, and in a dyadic context text messages predicted one’s own frequency of alcohol use but not their partners. Moderation analyses were underpowered due to a small sample size. These findings indicate that communication about alcohol via text messages may play an important role in frequency of alcohol use among early dating couples, particularly on one’s own behavior. Research and clinical implications of this work are also discussed.

Convolutional Neural Network (CNN) models have become the mainstream method in Artificial Intelligence (AI) areas for computer vision tasks like image classification and image segmentation. Deep CNNs contain a large volume of convolution calculations. Training a large CNN may take days or even weeks, which is time-consuming and costly. When we need multiple runs to search for the optimal CNN hypermeter settings, it would take a couple of months with limited GPUs, which is not acceptable and hinders the development of CNNs. It is essential to train CNN faster.

We have proposed a novel Conditional Reduction (CR) module to compress a single 1×1 convolution layer. Then we have developed a novel three-layer Conditional block (C-block) to compress the CNN bottleneck or inverted bottlenecks. At last we have developed a novel Conditional Network (CRnet) based on the CR module and C-block. We have tested the CRnet on two image classification datasets: CIFAR-10 and CIFAR-100, with multiple network expansion ratios and compression ratios. The experiments verify our methods’ correctness with attention to the importance of the input-output pattern when selecting a compression strategy. The experiments show that our proposed CRnet better balances the model complexity and accuracy compared to the state-of-the-art group convolution and Ghost Module compression.

We have proposed a flat reduced random sampling training strategy and a bottleneck reduced random sampling strategy. We have proposed a three-stage training method based on the bottleneck reduced random sampling. Furthermore, we have proved the data visibility of a sample in the whole training process and the theoretical reduced time by four theorems and two corollaries. We have tested the two sampling methods on three image classification datasets: CIFAR-10, CIFAR-100 and ImageNet. The experiments show that our proposed two sampling strategies effectively reduce a significant training time percentage at a very small accuracy loss.

Inadequate emotion regulation may underlie the development of psychopathology as well as worsened physical health, particularly in the context of stress. Cognitive reappraisal is typically considered an adaptive strategy to manage negative emotions. However, the extent to which reappraisal is beneficial may hinge upon contextual and individual differences. Specifically, it is unclear how the ability to reappraise effectively (i.e., reappraisal ability) and exposure to stressful life events moderates the association between habitual reappraisal and health. Using a series of questionnaires and an experimental task designed to measure the ability to effectively down-regulate sad emotions using reappraisal, this dissertation examines the interactive effects of habitual reappraisal, reappraisal ability, and exposure to stressful life events on depressive and anxiety symptoms as well as self-reported physical health. Results indicate that habitual reappraisal may protect against elevated depressive symptoms and worsened self-reported physical health for people exposed to stressful life events, and that reappraising often appears to be particularly important when people are less effective in their attempts. These findings provide novel contributions to the field of emotion regulation and health by clarifying that exposure to stressful life events is an important moderator in the association between reappraisal and health and by elucidating the important roles of both habitual reappraisal and reappraisal ability.