Risk factors for infections in people with multiple sclerosis (Diploma thesis):
Multiple sclerosis (MS) is an inflammatory, demyelinating disease of the central nervous system with a high burden of disease due to its onset in young adulthood and possible functional impairment in the course of the disease. The risk for infections is increased with MS resulting not only in a decreased quality of life, but also potentially in an increased disease morbidity. Thus, early identification of high risk individuals for infection is particularly important. This study investigated 19 covariates regarding their association with the frequency of infection. 389 patients with MS or clinically isolated syndrome from Innsbruck and Helsinki were surveyed retrospectively with a questionnaire about their amount of upper respiratory tract and urinary tract infections within the past two years. For variable selection, a stepwise negative binomial regression model was trained on 203 patients from Innsbruck and tested on 87 from Innsbruck and 76 from Helsinki. The most important variables to predict the frequency of infection were female sex (IRR = 2.51), progressive disease course (IRR = 1.50), chronic lung disease (IRR = 1.86), high leukocytes (IRR = 1.31), and bladder disturbance (IRR = 1.58). Predictive accuracy is similar in both test samples. The main limitations are the retrospective design and the time frame during the SARS-CoV-2 pandemic in which people were obligated to social distancing and to use face covering masks. A future study on a larger sample should focus on the different substances used for treatment as well as protective factors. Overall this study paved the way for a preliminary risk score to assess any patient's individual infection risk, providing valuable insights for clinical decision making.
Keywords: multiple sclerosis; risk score; upper respiratory tract infections; urinary tract infections; prediction model
Development of computer-based tasks for patients with unilateral right-hemispheric neglect A study to examine the degree of automation through velocity and acceleration on healthy participants (Master's thesis):
Unilateral right-hemispheric neglect is one of the most common neuropsychological diseases as a condition after a stroke. It is characterized by the negligence of one half of perception without consciously noticing it, with the possible additional involvement of motor neglect. There are still contradictory opinions about whether it is caused solely by a shift of attention or if motor deficiencies – even within the ipsilesional space – are always involved. Computer-based assessments offer further insights about this question as they provide dynamic measurements such as velocity and acceleration every few milliseconds and not only the response time of the task itself. An efficient measurement to test for potential deficiencies is the degree of automation. This study develops novel tablet-based tasks to diagnose neglect patients, which are evaluated in respects of automation and other important parameters like peak velocity, peak acceleration, and number of inversions in velocity (NIV) on healthy participants to identify potential influencing factors. The velocity and acceleration profile of every person should be normally distributed. It is expected that the dependent variables peak velocity, peak acceleration, and NIV are influenced by the extent of education and frequency of writing, but not by gender and age when conducting a task of wave drawing. While drawing horizontal lines, peak velocity is expected to differ between line direction but neither between age groups nor gender. The aim of this study is to develop novel computer-based tasks and to establish reference values for future clinical trials. 51 subjects conducted 23 tasks with the dominant right hand and 19 tasks with the left hand on a graphical Wacom tablet. For the task of wave drawing, every person’s velocity and acceleration profile is investigated regarding normality with the Shapiro-Wilks test, a probability density function, and a Q-Q plot. Furthermore, a multivariate linear regression is conducted to investigate the influence of gender, age, years of education, and writing frequency on the dependent variables peak velocity, peak acceleration and NIV. The evaluation of horizontal line drawing is examined with a repeated measures ANOVA to test differences in line direction. The velocity and acceleration profiles are not sufficiently normally distributed for individuals and skewed to the left over all subjects with a long tail to the right. Neither of the independent variables have any influence on velocity, acceleration or NIV, whereas the ANOVA shows a significant difference in line direction between leftward and rightward lines. The deviations from a Gaussian distribution might be best explained by the frequently described skewness of response times, as velocity is a function of response time. The tablet-based tasks should be further modified after clinical trials to make sure there is a high sensitivity even with this small number of tasks. Since there are no notable influencing factors, it can be used for neglect patients independent of their demographics for future clinical trials. This study paves the way for clinical use, where it might potentially increase diagnostic sensitivity considerably. Furthermore, it might provide insights about its cause as well as further differentiation between motor and non-motor symptoms.
Keywords: visuo-spatial neglect; computer-based drawing analysis; automation of drawing; neglect assessment
Varianz der Power pseudo-exakter oder konditionaler Tests von Annahmen des Rasch Modells - Vergleich des Rasch Samplers mit dem Exact Sampler (Bachelor's thesis):
Draxler & Zessin (2015) have proposed a class of pseudo-exact or conditional tests for power calculation of assumptions of the Rasch model. Sampling algorithms are required to simulate the data required for power calculation. Verhelst (2008) has designed a relatively fast algorithm called the Rasch Sampler, which approximates the true distribution using Markov Chain Monte Carlo procedures. Miller & Harrison (2013) have developed an algorithm called the Exact Sampler, which can count the exact distribution and draw from it. The accuracy of the two samplers is compared by examining potential influences of sample size, DIF-parameters and item difficulty on the accuracy of the power calculation. Furthermore, the burn-in phase and the step parameter are checked as influencing factors on the Rasch Sampler. The accuracy of the samplers does not differ meaningfully. The power increases with higher sample size. Also the power increases with larger positive and negative model deviations. With moderate item difficulty, the power for positive and negative DIF parameters is almost equal. If an easy item deviates from the model, the power is greater if the deviation is positive than if the item is negative. With a difficult item, a contrasting trend can be observed with the difference that the range of the power values is relevantly higher. Neither the burn-in phase nor the step parameter has any influence on the accuracy of the Rasch Sampler. Due to more efficient calculation the Rasch Sampler should be used in any case. The results concerning the behaviour of the power under variation of different parameters correspond to the observations of Draxler & Zessin (2015).
Keywords: Rasch model, power, pseudo-exact tests, conditional tests, Rasch sampler, exact sampler