Search

The recent growth in the availability of biomedical data promises to reshape healthcare by ushering in an era of personalized medicine where data can be used to diagnose and treat patients with pinpoint accuracy. Truly realizing this goal requires building statistical models that individually model patient variations such as age, sex, and genetic makeup, which leads to a combinatorial growth in the number of parameters in a model and noisy estimates. Fortunately, Bayesian hierarchical models, along with recent computational advances, provide a solution to this issue. By naturally embedding the hierarchical structure that many datasets exhibit into the model, these models allow for separate estimates that capture population-level variation and simultaneously avoid noise via regularization to a population mean.In this thesis, we describe novel models and computational methods for Bayesian hierarchical modeling of biomedical data. We begin by describing our contributions to various areas of Bayesian modeling, along with the problems from our applied work that motivated these contributions.Specifically, we describe our disease progression model, which hierarchically models patient disease trajectories. The model, which was motivated by our applied work on Alzheimer's Disease, utilizes I-splines to capture the characteristic monotonic shape of dementia disease trajectories, along with Dirichlet distributions over the coefficients of these I-splines to hierarchically model these trajectories. Next we describe our work on using the Givens Representation of orthogonal matrices to infer models with orthogonal matrix parameters, such as factor models, in a general Bayesian framework. We describe the innovations in our method along with our motivating hierarchical example based on the analysis of protein biomarkers of coagulopathic trauma patients. Next, we describe a mechanistic model of coagulopathy that relates clotting assay data to protein concentrations, effectively providing a fast and convenient way for clinicians to understand key protein markers involved in clotting.Next we transition specifically to Hamiltonian Monte Carlo (HMC) and elucidate the connection between multiscale posterior distributions and the efficiency of HMC. We describe the issue of numerical stability inside HMC and present our implicit HMC algorithm for efficiently sampling non-Gaussian posterior distributions.Lastly, we provide a summary of our contributions along with ideas for future work.

Drug and alcohol neonate simulators were used to highlight the effects of substance misuse on prenatal development within lifespan development modules to 61 Psychology undergraduates and 12 sixth-form Health and Social Care students. A mixed method approach was used considering both knowledge development and perceptions of experiential hands-on learning. Student knowledge of the effects of drugs and alcohol on prenatal development was assessed before and after an interactive session on teratogenesis utilising neonate simulators demonstrating these effects. With this experiential learning approach, the statistical analysis combining both groups showed increased knowledge after the interactive session, and the students perceived this enhanced understanding to be a direct result of hands-on interaction with the neonate simulators. The researchers encourage the use of neonate simulators as good practice that can be effectively incorporated into the psychology and other health-related teaching contexts to facilitate knowledge on teratogenesis.

We introduce an approach based on the Givens representation for posterior inference in statistical models with orthogonal matrix parameters, such as factor models and probabilistic principal component analysis (PPCA). We show how the Givens representation can be used to develop practical methods for transforming densities over the Stiefel manifold into densities over subsets of Euclidean space. We show how to deal with issues arising from the topology of the Stiefel manifold and how to inexpensively compute the change-of-measure terms. We introduce an auxiliary parameter approach that limits the impact of topological issues. We provide both analysis of our methods and numerical examples demonstrating the effectiveness of the approach. We also discuss how our Givens representation can be used to define general classes of distributions over the space of orthogonal matrices. We then give demonstrations on several examples showing how the Givens approach performs in practice in comparison with other methods.

Students starting on their higher education journeys at universities globally face the issue of becoming part of a new social scene which may involve drugs, alcohol and new sexual experiences, a growing issue also faced by older secondary school aged pupils. This project's main objective was to use creative pedagogy to engage Level 4 Psychology students and Sixth-form Academy students to learn about the impact of risk-taking behaviours on foetal development. The students engaged with an interactive Drug Affected Simulator doll and a non-interactive Foetal Alcohol Simulator doll and were tasked to think creatively about how they could be used to educate others. It was predicted that through this hands-on learning with the virtual-baby dolls and higher-order creative thinking, students would develop their metacognitive understanding of the negative impact of drugs and alcohol on pre-natal development.

Relating Disparate Measures of Coagulapathy Using Unorthodox Data: A Hybrid Mechanistic-Statistical Approach, Code and data available at github.com/stan-dev/stancon_talks

This study aimed to build upon previous research by assessing various predictors of academic performance, namely personality, motivation, curiosity and learning approach. In line with the view of von Stumm (2016), the study additionally examined students’ estimates of their performance, and whether these same variables predicted these estimates. Forty, Level 6, undergraduate students were asked at the beginning of the academic year to complete four questionnaires and to estimate their assignment grades for their core units. At the end of the year their Level 6 average percentage was sourced from exam board documents. Multiple regression analysis showed that autonomous regulation (a measure of intrinsic motivation), and need for cognition (a measure of curiosity) both predicted 25% of the variance in academic performance. Need for cognition also predicted 14% of the variance in the students’ average estimated grades, but none of the variables predicted whether the students would under or overestimate their grades. It was concluded that intellectual curiosity is an important variable that should be considered as a key predictor of academic performance.

There is limited research assessing the use of the Foetal Alcohol and Drug Affected virtual-baby dolls (FAS/DAS) in education programmes. This research aimed to address this gap by using the simulator dolls with undergraduates and Year 12 and 13 students to educate them about the teratogenic effects of drugs and alcohol on prenatal development. Students were additionally tasked to think creatively about how the dolls could be used to educate others. It was predicted that through hands-on learning with the virtual-baby dolls and higher-order creative thinking, students would develop their metacognitive understanding of the potential risks to the foetus.

Alzheimer's Disease is one the most debilitating diseases, but how do we diagnose it accurately? Researchers have been trying to answer this question by building generative models to describe how patient biomarkers, such as MRI scans, psychological tests, and lab tests relate over time to the underlying brain deterioration that's present in Alzheimer's Disease. In this notebook we show how we translated these models to the Bayesian framework in Stan and how this allowed for several model improvements that can ultimately improve our understanding of Alzheimer's and help physicians in diagnosis. In particular, we describe how we hierarchically model patient disease trajectories to obtain stable estimates for patients who lack data. We describe how fitting in Stan yields uncertainties on these disease trajectories, and why that is important for weighing the pros and cons of risky treatment. Lastly, we describe a new method for Bayesian modeling of these monotonic disease trajectories in Stan using I-Splines., Code and data available at github.com/stan-dev/stancon_talks

We introduce the Givens Transform, a novel transform between the space of orthonormal matrices and $\mathbb{R}^D$. The Givens Transform allows for the application of any general Bayesian inference algorithm to probabilistic models containing constrained unit-vectors or orthonormal matrix parameters. This includes a variety of matrix factorizations and dimensionality reduction models such as Probabilistic PCA (PPCA), Exponential Family PPCA (BXPCA), and Canonical Correlation Analysis (CCA). While previous Bayesian approaches to these models relied on separate sampling update rules for constrained and unconstrained parameters, the Givens Transform enables the treatment of unit-vectors and orthonormal matrices agnostically as unconstrained parameters. Thus any Bayesian inference algorithm can be used on these models without modification. This opens the door to not just sampling algorithms, but Variational Inference (VI) as well. We illustrate with several examples and supplied code, how the Givens Transform allows end-users to easily build complex models in their favorite Bayesian modeling framework such as Stan, Edward, or PyMC3, a task that was previously intractable due to technical constraints.

© 2015 Pourzanjani et al. Being able to infer one way direct connections in an oscillatory network such as the suprachiastmatic nucleus (SCN) of the mammalian brain using time series data is difficult but crucial to understanding network dynamics. Although techniques have been developed for inferring networks from time series data, there have been no attempts to adapt these techniques to infer directional connections in oscillatory time series, while accurately distinguishing between direct and indirect connections. In this paper an adaptation of Granger Causality is proposed that allows for inference of circadian networks and oscillatory networks in general called Adaptive Frequency Granger Causality (AFGC). Additionally, an extension of this method is proposed to infer networks with large numbers of cells called LASSO AFGC. The method was validated using simulated data from several different networks. For the smaller networks the method was able to identify all one way direct connections without identifying connections that were not present. For larger networks of up to twenty cells the method shows excellent performance in identifying true and false connections; this is quantified by an area-under-the-curve (AUC) 96.88%. We note that this method like other Granger Causality-based methods, is based on the detection of high frequency signals propagating between cell traces. Thus it requires a relatively high sampling rate and a network that can propagate high frequency signals.

We study the association between weight fluctuation and activity tracking in an on-line population of thousands of individuals using digital health trackers (1,749 ≤ N ≤ 14,411, depending on the activity tracker considered) with millions of recorded activities (119,292 ≤ N ≤ 2,221,382) over the years 2013-2015. In a first between-subject analysis, we found a positive association between activity tracking frequency and weight loss. Users who log food with moderate frequency lost an additional 0.63% (CI [0.55, 0.72]; p < .001) of their body weight per month relative to low frequency loggers. Frequent workout loggers lost an additional 0.38% (CI [0.20, 0.56]; p < .001) and frequent weight loggers lost an additional 0.40% (CI [0.33, 0.47]; p < .001) as compared to infrequent loggers. In a subsequent within-subject analysis on a subset of the population (799 ≤ N ≤ 6,052) with sufficient longitudinal data, we used fixed effect models to explore the temporal relationship between a change in tracking adherence and weight change. We found that for the same individual, weight loss is significantly higher during periods of high adherence to tracking vs. periods of low adherence: +2.74% of body weight lost per month (CI [2.68, 2.81]; p < .001) during adherent weight tracking, +1.35% per month (CI [1.26, 1.43]; p < .001) during adherent food tracking, and +0.60% per month (CI [0.44, 0.76]; p < .001) during adherent workout tracking. The findings suggest that adherence to activity tracking can be utilized as a convenient real-time predictor of weight fluctuations, enabling large-scale, personalized intervention strategies.

“Today, the idea of allowing aircraft pilots to operate as they see fit, without any control from the land, is unthinkable. Will the same things apply to ships in twenty to thirty years’ time?”

Numerous factors over the past three decades have contributed to a major decline in the number of seagoing staff in the United Kingdom. These include shipping companies opting for other flags, as well as the profession itself becoming less and less attractive for school leavers. This decrease in numbers is not only significant for shipping companies, but also for the allied services which used to get supported from seafarers coming ashore in search of a new career. It is now recognised that changes in the maritime skills base will affect both the seagoing as well as the shore side of the industry.

The shipping industry is under a great deal of pressure from different sources to improve the safety records, and reduce the number of accidents at sea. Human element has been identified and accept...

Nonlinear time history analysis is being increasingly utilized in the seismic design and evaluation of building structures. The general intent of such an analysis is to better understand the behavior of a building subjected to a given seismic motion. However, the methodology and particularly the modeling of the structural element properties have a significant impact on the results of this type of analysis. This paper evaluates the analytical models and the incorporated assumptions in two commonly used nonlinear programs: DRAIN-2DX 1·10 and IDARC 4·0. An instrumented reinforced concrete frame building is analyzed using both programs, and the analytical results are compared with the recorded data. A parametric study follows the program comparison to develop an understanding of the impact of various parameters involved in such an analysis. The study indicates that varying degrees of accuracy are possible with nonlinear time history analysis, depending on the program used and the assumptions incorporated. This paper also evaluates the sensitivity of the analytical results to the parameters involved. Copyright © 1999 John Wiley & Sons, Ltd.

Safety of ship handling is an important issue and should be considered seriously at early design stages. In this work a comprehensive four quadrant mathematical model is introduced. The model is used to simulate ship handling behaviour in open and confined waters. Human handling technique was modelled by a set of operating procedure. In this study special attention is paid to look at the ship handling behaviour rather than the human operator. Examples of normal handling performance, engine failure and partial tug failure mode are given.

Because early administration of intravenous fat emulsions (IVFEs) has been linked to infectious complications in trauma patients, we began withholding IVFE for the first seven to ten days of parenteral nutrition (PN) in all surgical intensive care unit (SICU) patients. Prior to this, IVFE had been infused from the start of PN.To evaluate the influence of delaying IVFE on infectious complications in SICU patients.Retrospective review from October 2006 to June 2009 of SICU patients before and after a change in IVFE practice patterns in a 44-bed SICU at an academic medical center. Adult patients who received PN for more than six days were included. Patients receiving PN with IVFE prior to SICU admission or being given other lipid emulsion therapy were excluded. The data collected included demographics, transfusion requirements, nutritional assessments, and laboratory and microbiology results. The infectious complications studied were pneumonia, urinary tract infections (UTIs), blood stream infections (BSIs), and catheter-related blood stream infections (CRBSIs).Sixty-four patients received IVFE; 30 at initiation of PN and 34 starting after seven to ten days. The two groups had similar demographics, severity of illness, transfusion requirements, and duration of PN. Infectious complications occurred in 65.6% of patients (63.3% having immediate IVFE vs. 67.6% having delayed IVFE; p = 0.79). Seventeen patients developed BSI or CRBSI while receiving PN (26.7% immediate IVFE vs. 26.5% delayed IVFE; p 0.99). The mortality rates were 63.3% and 55.9%, respectively (p = 0.63).Withholding IVFE therapy during the first seven to ten days of PN did not influence infectious complications or the mortality rate in SICU patients. The benefits of delaying IVFE therefore may not be generalizable to all critically ill patients.

Windsurfers appeared in the leisure market during the past two decades and have now been established as one of the most popular water sports. Their design has been dominated by the qualitative feedback from the users of such craft. They have been able to achieve a speed record of about 40 knots at the perfect sailing conditions. It is believed that a more scientific approach in design, using CFD, will result in a craft capable of breaking this speed record.

The mathematical modeling of ship behavior is still far from a systematic procedure. Considerable further research is needed to evolve simple techniques that can be used with confidence in preliminary ship design or as diagnostic tools to investigate problems in the behavior of an existing ship. To that end, this paper is intended as a contribution to the development of a relatively simple yet effective method for estimating the hydrodynamic forces experienced by the hull in a translational motion. The work is based on a Force Mathematical Model introduced by McCallum.

Alcohol consumption and drug abuse during pregnancy can have numerous adverse health consequences for the developing foetus, including foetal alcohol syndrome (FAS) and teratogenic effects, increased morbidity and mortality and a high incidence of birth defects that can have long-term consequences. Therefore, it is important to assess ways of increasing knowledge and understanding of these effects. There is limited research assessing the use of the Foetal Alcohol and Drug Affected virtual-baby dolls in education programmes to highlight the detrimental effects. However, using simulators as health promotion and sociotechnical change is increasing due to the growing reality that smoking, alcohol and illicit drug use in women of reproductive age is increasing and the continued use of drugs during pregnancy is common (Whittaker, 2003). The current research aimed to build upon previous research by Dr Hussain and Dr Pourzanjani to assess the use of simulator dolls to educate students about the teratogenic effects of drugs and alcohol on prenatal development. The current study compared four groups of students: college students (Year 12 and 13) and undergraduates from Psychology, Midwifery and Social work. A mixed-methods approach firstly compared knowledge prior to the teaching session with knowledge afterwards. Qualitative data was then sought to determine the students' perceptions of how the simulator dolls had aided their understanding in order to provide more depth to the quantitative findings. It was predicted that interaction with the simulator dolls and higher-order creative thinking about their use to educate others, as well as how this might be beneficial within their discipline area, would help students develop their metacognitive understanding of the potential risks to the foetus and to enhance their professional practice.