Browsing by Author "Li, Kaigang, committee member"

Now showing 1 - 7 of 7

Open Access
A questionnaire integration system based on question classification and short text semantic textual similarity
(Colorado State University. Libraries, 2018) Qiu, Yu, author; Pallickara, Sangmi Lee, advisor; Pallickara, Shrideep, committee member; Li, Kaigang, committee member
Semantic integration from heterogeneous sources involves a series of NLP tasks. Existing re- search has focused mainly on measuring two paired sentences. However, to find possible identical texts between two datasets, the sentences are not paired. To avoid pair-wise comparison, this thesis proposed a semantic similarity measuring system equipped with a precategorization module. It applies a hybrid question classification module, which subdivides all texts to coarse categories. The sentences are then paired from these subcategories. The core task is to detect identical texts between two sentences, which relates to the semantic textual similarity task in the NLP field. We built a short text semantic textual similarity measuring module. It combined conventional NLP techniques, including both semantic and syntactic features, with a Recurrent Convolutional Neural Network to accomplish an ensemble model. We also conducted a set of empirical evaluations. The results show that our system possesses a degree of generalization ability, and it performs well on heterogeneous sources.
Open Access
Adaptive spatiotemporal data integration using distributed query relaxation over heterogeneous observational datasets
(Colorado State University. Libraries, 2018) Mitra, Saptashwa, author; Pallickara, Sangmi Lee, advisor; Pallickara, Shrideep, committee member; Li, Kaigang, committee member
Combining data from disparate sources enhances the opportunity to explore different aspects of the phenomena under consideration. However, there are several challenges in doing so effectively that include inter alia, the heterogeneity in data representation and format, collection patterns, and integration of foreign data attributes in a ready-to-use condition. In this study, we propose a scalable query-oriented data integration framework that provides estimations for spatiotemporally aligned data points. We have designed Confluence, a distributed data integration framework that dynamically generates accurate interpolations for the targeted spatiotemporal scopes along with an estimate of the uncertainty involved with such estimation. Confluence orchestrates computations to evaluate spatial and temporal query joins and to interpolate values. Our methodology facilitates distributed query evaluations with a dynamic relaxation of query constraints. Query evaluations are locality-aware and we leverage model-based dynamic parameter selection to provide accurate estimation for data points. We have included empirical benchmarks that profile the suitability of our approach in terms of accuracy, latency, and throughput at scale.
Open Access
Assessment of physical health, physical performance, and physical activity in exercise promotion intervention for older adults
(Colorado State University. Libraries, 2020) Forsyth, Garrett, author; Diehl, Manfred, advisor; Luong, Gloria, committee member; Li, Kaigang, committee member
This study examined the effects of an intervention program, known as AgingPlus, on indicators of physical health (i.e. systolic and diastolic blood pressure), physical performance (i.e. left- and right-hand grip strength), and physical activity levels (i.e. total steps walked, total kilocalorie expenditure, and total distance). The sample included 120 older-adult participants who were relatively healthy and community residing. The study used a randomized pretest-posttest control group design. Findings showed that NVOA and self-efficacy beliefs did not mediate the association between the intervention and the outcome variables. We found that participants in the treatment and the control group improved their physical health, physical performance, and physical activity from the baseline assessment to the Week 8 follow-up. Additionally, the results showed that only participants in the treatment condition significantly decreased their systolic and diastolic blood pressure and significantly improved their left- and right-hand grip strength over an eight-week interval. These findings suggest that targeting NVOA and self-efficacy beliefs may be an effective strategy to optimize adults' healthy aging.
Open Access
Determinants of driving performance following stroke
(Colorado State University. Libraries, 2022) Pollet, Aviva Katherine, author; Lodha, Neha, advisor; Li, Kaigang, committee member; Schmid, Arlene A., committee member
Overall introduction: Individuals with stroke experience motor and cognitive deficits both of which can impact driving performance. Using two separate studies, we evaluated the influence of motor and cognitive factors on driving performance in stroke survivors. In the first study, we evaluated how driving impairments in stroke survivors is influenced by the use of either the paretic or non-paretic leg for pedal control. Methods 1: Twenty-two individuals with chronic stroke were recruited in two groups depending on their lower-limb choice for pedal control 1) paretic leg drivers, individuals using their paretic leg to control the car pedals (N = 11, 68.4 ± 7.8 years) and 2) non-paretic leg drivers, individuals using their non-paretic leg to control the car pedals (N = 11, 61.1 ± 13.7 years). Both groups performed a car following task in a driving simulator. The task required participants to follow a lead car by controlling the gas pedal accurately and respond to brake lights by pressing the brake pedal as fast as possible. We quantified gas pedal error using root mean square error (RMSE). We measured brake response time as the time from the onset of the brake lights of the lead car to the application of the brake pedal. We also dissociated the brake response time into pre-motor and motor response times. We used the Driving Habits Questionnaire (DHQ) to measure self-reported on-road driving behavior. Additionally, using surface electromyography (EMG), we analyzed neuromuscular activation using burst duration and amplitude, and coordination using overlap and coactivation of the tibialis anterior (TA) and medial gastrocnemius (MG) during the braking portion of the car following task. Results 1: The paretic leg drivers showed greater gas pedal RMSE than the non-paretic leg drivers (p ≤ 0.01). The paretic leg drivers had a slower brake response time than the non-paretic leg drivers (p < 0.05). Premotor response time was not different between the two groups (p = 0.71), however, the paretic leg drivers had a significantly slower motor response time relative to the non-paretic leg drivers (p < 0.05). The paretic leg drivers had lower DHQ scores than the non-paretic leg drivers (p ≤ 0.01). DHQ and brake response time were negatively correlated (r = - 0.42, p ≤ 0.05). Additionally, paretic leg drivers showed longer TA EMG burst duration (p <0.05) and more TA-MG overlap (p <0.05). TA EMG burst duration was positively correlated to brake response time (r = 0.51, p < 0.05) and motor response time (r = 0.61, p < 0.05). TA-MG overlap was positively correlated to brake response time (r = 0.76, p = 0.001). In the second study, we evaluated how cognitive load influenced driving impairments in stroke survivors. Methods 2: Ten individuals with chronic stroke participated in the current study (N = 10, 65.6 ± 14.9 years). The participants performed simulated driving without (single-task) and with (dual-task) a cognitive load. The single-task driving required participants to drive along a rural road and brake as quickly as possible when an unexpected hazard, such as wildlife crossing into the driving lane, was encountered. The dual-task driving required participants to drive in the same driving scenario while performing a secondary cognitive task. The cognitive task involved mental arithmetic to induce higher cognitive load while driving. Specifically, participants were asked to subtract 4 and add 3 to a random number and do so repeatedly until the end of the driving task. We measured lane departures as the number of times the edge of the participant's vehicle left the designated driving lane. We measured speed compliance as the percent of total time the individual was within +/- 5 MPH of the speed limit between events. Additionally, we measured brake response time as the time from the appearance of the hazard stimulus to the application of the brake pedal. Results 2: Individuals with stroke show more lane departures throughout the entire drive during dual-task driving than single-task driving (p < 0.05). Additionally, individuals with stroke show worse speed compliance during dual-task driving than single-task driving (p < 0.05). There was no difference in brake response time between the single-task and dual-task driving (p = 0.18). Overall conclusion: Driving performance in stroke survivors is influenced by limb selection for pedal control and cognitive load. The current studies demonstrate the need to assess and train motor and cognitive deficits that contribute to driving performance in individuals with stroke. Motor deficits in pedal control and brake response time contribute to unsafe driving in individuals with stroke. Cognitive deficits in lane departures and speed compliance in driving with cognitive load also contribute to unsafe driving in individuals with stroke. To address these deficits, stroke driving rehabilitation programs should focus on driving leg and cognitive environment of driving.
Open Access
The effects of footwear cushioning on walking performance in females with multiple sclerosis
(Colorado State University. Libraries, 2018) Monaghan, Andrew S., author; Fling, Brett W., advisor; Li, Kaigang, committee member; Stephens, Jaclyn, committee member
Multiple sclerosis is a chronic and progressive neurodegenerative disease which incurs a multitude of walking impairments. Protective strategies targeted at maintaining postural stability during walking include increasing stance and double support time with reciprocal decreases in swing and single support time, however these adaptions inadvertently increase fall risk. The midsole construct of footwear has demonstrated the ability to mediate these deficits in running but has not been explored in a neurologic population with known fall risk. The purpose of this study was to investigate the effects of two different midsole conditions on the spatiotemporal parameters of gait in females with multiple sclerosis (MS). Gait testing was conducted while 18 females with MS performed two-minute walk tests in 1) a high-cushion and 2) a standard-cushion midsole shoe. Spatiotemporal gait parameters were assessed using wireless inertial sensors. Participants spent less time in double support and stance phase with concomitantly more time in single support and swing phase in the high-cushion midsole shoe as compared to the standard-cushion. The high-cushion shoe may decrease fall risk by improving gait parameters associated with increased risk of falls.
Open Access
The prevalence and clustering of cardiovascular risk factors in college students
(Colorado State University. Libraries, 2017) DeYoung, Wendy A., author; Kuk, Linda, advisor; Gloeckner, Gene, advisor; Li, Kaigang, committee member; Mallette, Dawn, committee member
Cardiovascular disease (CVD) has been the leading cause of death in the United States for adult men and women for the last 80 years and is a major cause of disability. Additionally, CVD is the second leading cause of death in young adults ages 18 to 29. This chronic disease is typically associated with adults; however, recently CVD has been identified in the younger population as well. The literature on CVD risk factors and college students is very limited. College campuses serve as an ideal setting to examine risk factors for CVD among young adults. College life can lead to multiple changes in lifestyle including changes in activity patterns, dietary intake, sleep patterns, weight fluctuations, alcohol consumption, tobacco use, and drug use. Collectively, the impact of these behaviors sets the stage for the development of multiple risk factors associated with CVD. Therefore, the purpose of this investigation was to identify the prevalence and clustering of CVD risk factors with undergraduate students' age 18 – 25 years old enrolled at Colorado State University (CSU), during the spring semester, 2017. A non-experimental, cross-sectional research design was used to identify the prevalence and clustering of CVD risk factors in the sample. Multiple screenings were centrally located on campus for student convenience. The screening included informed consent, health history questionnaire, resting blood pressure, lipid analysis, and health and wellness questionnaire. A total of 180 students were recruited for the study. The average age was 21.40 years with a range of 18 – 25 year. Over half, 62.18 percent were female, 53.75 percent were seniors, and 81.88 percent were White. Although the study was open to the entire university, 78.62 percent were from the department of Health and Exercise Science. Students from 23 different academic departments were represented in the sample. A total of 706 CVD risk factors were identified including; 208 for nicotine use, 238 with family history of CVD, 42 for high LDLs, 32 for elevated SBP, 24 for elevated DBP, 22 for inactivity, 21 for elevated triglycerides, 20 for elevated total cholesterol, 20 for elevated blood glucose, 19 for low HDLs in males, 15 for low HDLs in females, 39 for BMI ≥ 25 kg/m2), 4 for increase in waist circumference for females, and 2 for an elevated waist circumference in males. The range of CVD risk factors per student was from zero to six. The significance in totality of CVD risk factors in this apparently healthy undergraduate student sample is startling and warrants further examination. Male students showed statistically significant higher glucose, TCHOL/HDL, SBP, and DBP, and were more likely to use cigarettes e-cigarettes, cigars, and smokeless tobacco, anabolic steroids and beer than females. Female students had a statistically significant higher total cholesterol level, HDL, and wine consumption than males. White students had a higher prevalence of hookah and smokeless tobacco, wine, liquor, drinking up to five drinks in one setting, driving after drinking alcohol, and consuming marijuana edibles. Freshmen had a statistically significant lower SBP than sophomores, and seniors. A statistically significant difference was found with seniors consuming more beer than freshman and sophomores. Seniors were also more likely to drive after drinking alcohol than freshman, sophomores, and juniors. Lastly, juniors had a statistically significant higher consumption of marijuana edibles than sophomores did. CSU undergraduate students are more likely to rank their general health as "very good" or "excellent", less likely to have a history of elevated blood pressure, more likely to use hookah, and less likely be obese when compared to undergraduate college students across the nation. Multiple correlations were identified and followed up with simultaneous multiple regressions were completed to investigate the best predictors of tobacco use, hookah use, elevated SBP, elevated DBP, BMI, and elevated total cholesterol. K-means cluster analysis provided a visual display of various groupings for family history of CVD, blood lipids and general health, blood pressure, tobacco and marijuana use, alcohol use, and general health tobacco and alcohol use combined, and drug use. Data were standardized to Z-scores for comparison. The Z-scores greater than three included cigarettes, e-cigarettes, hookah, cigars, smokeless tobacco, cocaine, methamphetamines, and other illegal drugs. Collectively, these results indicate a significant prevalence of CVD risk factors and high alcohol and drug use among the CSU student sample. It is apparent that this undergraduate college student sample may be more at risk for developing subsequent CVD than previously thought and should be screened for CVD beginning at age 20 as recommended by health and medical experts.
Open Access
Towards interactive analytics over voluminous spatiotemporal data using a distributed, in-memory framework
(Colorado State University. Libraries, 2023) Mitra, Saptashwa, author; Pallickara, Sangmi Lee advisor; Pallickara, Shrideep, committee member; Ortega, Francisco, committee member; Li, Kaigang, committee member
The proliferation of heterogeneous data sources, driven by advancements in sensor networks, simulations, and observational devices, has reached unprecedented levels. This surge in data generation and the demand for proper storage has been met with extensive research and development in distributed storage systems, facilitating the scalable housing of these voluminous datasets while enabling analytical processes. Nonetheless, the extraction of meaningful insights from these datasets, especially in the context of low-latency/ interactive analytics, poses a formidable challenge. This arises from the persistent gap between the processing capacity of distributed systems and their ever-expanding storage capabilities. Moreover, the interactive querying of these datasets is hindered by disk I/O, redundant network communications, recurrent hotspots, transient surges of user interest over limited geospatial regions, particularly in systems that concurrently serve multiple users. In environments where interactive querying is paramount, such as visualization systems, addressing these challenges becomes imperative. This dissertation delves into the intricacies of enabling interactive analytics over large-scale spatiotemporal datasets. My research efforts are centered around the conceptualization and implementation of a scalable storage, indexing, and caching framework tailored specifically for spatiotemporal data access. The research aims to create frameworks to facilitate fast query analytics over diverse data-types ranging from point, vector, and raster datasets. The frameworks implemented are characterized by its lightweight nature, residence primarily in memory, and their capacity to support model-driven extraction of insights from raw data or dynamic reconstruction of compressed/ partial in-memory data fragments with an acceptable level of accuracy. This approach effectively helps reduce the memory footprint of cached data objects and also mitigates the need for frequent client-server communications. Furthermore, we investigate the potential of leveraging various transfer learning techniques to improve the turn-around times of our memory-resident deep learning models, given the voluminous nature of our datasets, while maintaining good overall accuracy over its entire spatiotemporal domain. Additionally, our research explores the extraction of insights from high-dimensional datasets, such as satellite imagery, within this framework. The dissertation is also accompanied by empirical evaluations of our frameworks as well as the future directions and anticipated contributions in the domain of interactive analytics over large-scale spatiotemporal datasets, acknowledging the evolving landscape of data analytics where analytics frameworks increasingly rely on compute-intensive machine learning models.