Bayesian algorithms with applications

Advances in algorithms for approximate sampling from a multivariable target function have led to solutions to challenging statistical inference problems that would otherwise not be considered by the applied scientist. Such sampling algorithms are particularly relevant to Bayesian statistics, since the target function is the posterior distribution of the unobservables given the observables. In this thesis we develop, adapt and apply Bayesian algorithms, whilst addressing substantive applied problems in biology and medicine as well as other applications. For an increasing number of high-impact research problems, the primary models of interest are often sufficiently complex that the likelihood function is computationally intractable. Rather than discard these models in favour of inferior alternatives, a class of Bayesian "likelihoodfree" techniques (often termed approximate Bayesian computation (ABC)) has emerged in the last few years, which avoids direct likelihood computation through repeated sampling of data from the model and comparing observed and simulated summary statistics. In Part I of this thesis we utilise sequential Monte Carlo (SMC) methodology to develop new algorithms for ABC that are more efficient in terms of the number of model simulations required and are almost black-box since very little algorithmic tuning is required. In addition, we address the issue of deriving appropriate summary statistics to use within ABC via a goodness-of-fit statistic and indirect inference. Another important problem in statistics is the design of experiments. That is, how one should select the values of the controllable variables in order to achieve some design goal. The presences of parameter and/or model uncertainty are computational obstacles when designing experiments but can lead to inefficient designs if not accounted for correctly. The Bayesian framework accommodates such uncertainties in a coherent way. If the amount of uncertainty is substantial, it can be of interest to perform adaptive designs in order to accrue information to make better decisions about future design points. This is of particular interest if the data can be collected sequentially. In a sense, the current posterior distribution becomes the new prior distribution for the next design decision. Part II of this thesis creates new algorithms for Bayesian sequential design to accommodate parameter and model uncertainty using SMC. The algorithms are substantially faster than previous approaches allowing the simulation properties of various design utilities to be investigated in a more timely manner. Furthermore the approach offers convenient estimation of Bayesian utilities and other quantities that are particularly relevant in the presence of model uncertainty. Finally, Part III of this thesis tackles a substantive medical problem. A neurological disorder known as motor neuron disease (MND) progressively causes motor neurons to no longer have the ability to innervate the muscle fibres, causing the muscles to eventually waste away. When this occurs the motor unit effectively ‘dies’. There is no cure for MND, and fatality often results from a lack of muscle strength to breathe. The prognosis for many forms of MND (particularly amyotrophic lateral sclerosis (ALS)) is particularly poor, with patients usually only surviving a small number of years after the initial onset of disease. Measuring the progress of diseases of the motor units, such as ALS, is a challenge for clinical neurologists. Motor unit number estimation (MUNE) is an attempt to directly assess underlying motor unit loss rather than indirect techniques such as muscle strength assessment, which generally is unable to detect progressions due to the body’s natural attempts at compensation. Part III of this thesis builds upon a previous Bayesian technique, which develops a sophisticated statistical model that takes into account physiological information about motor unit activation and various sources of uncertainties. More specifically, we develop a more reliable MUNE method by applying marginalisation over latent variables in order to improve the performance of a previously developed reversible jump Markov chain Monte Carlo sampler. We make other subtle changes to the model and algorithm to improve the robustness of the approach.

Self-determination theory as applied to the design of a software learning system using whole-body controls

Whole-body computer control interfaces present new opportunities to engage children with games for learning. Stomp is a suite of educational games that use such a technology, allowing young children to use their whole body to interact with a digital environment projected on the floor. To maximise the effectiveness of this technology, tenets of self-determination theory (SDT) are applied to the design of Stomp experiences. By meeting user needs for competence, autonomy, and relatedness our aim is to increase children's engagement with the Stomp learning platform. Analysis of Stomp's design suggests that these tenets are met. Observations from a case study of Stomp being used by young children show that they were highly engaged and motivated by Stomp. This analysis demonstrates that continued application of SDT to Stomp will further enhance user engagement. It also is suggested that SDT, when applied more widely to other whole-body multi-user interfaces, could instil similar positive effects.

Modeling as a bridge between real world problems and school mathematics

This article argues for an interdisciplinary approach to mathematical problem solving at the elementary school, one that draws upon the engineering domain. A modeling approach, using engineering model eliciting activities, might provide a rich source of meaningful situations that capitalize on and extend students’ existing mathematical learning. The study reports on the developments of 48 twelve-year old students who worked on the Bridge Design activity. Results revealed that young students, even before formal instruction, have the capacity to deal with complex interdisciplinary problems. A number of students created quite appropriate models by developing the necessary mathematical constructs to solve the problem. Students’ difficulties in mathematizing the problem, and in revising and documenting their models are presented and analysed, followed by a discussion on the appropriateness of a modeling approach as a means for introducing complex problems to elementary school students.

Finite element shoulder models

Shoulder joint is a complex integration of soft and hard tissues. It plays an important role in performing daily activities and can be considered as a perfect compromise between mobility and stability. However, shoulder is vulnerable to complications such as dislocations and osteoarthritis. Finite element (FE) models have been developed to understand shoulder injury mechanisms, implications of disease on shoulder complex and in assessing the quality of shoulder implants. Further, although few, Finite element shoulder models have also been utilized to answer important clinical questions such as the difference between a normal and osteoarthritic shoulder joint. However, due to the absence of experimental validation, it is questionable whether the constitutive models applied in these FE models are adequate to represent mechanical behaviors of shoulder elements (Cartilages, Ligaments, Muscles etc), therefore the confidence of using current models in answering clinically relevant question. The main objective of this review is to critically evaluate the existing FE shoulder models that have been used to investigate clinical problems. Due concern is given to check the adequacy of representative constitutive models of shoulder elements in drawing clinically relevant conclusion. Suggestions have been given to improve the existing shoulder models by inclusion of adequate constitutive models for shoulder elements to confidently answer clinically relevant questions.

A hybrid smoothed finite element method (H-SFEM) to solid mechanics problems

In this paper, a hybrid smoothed finite element method (H-SFEM) is developed for solid mechanics problems by combining techniques of finite element method (FEM) and Node-based smoothed finite element method (NS-FEM) using a triangular mesh. A parameter is equipped into H-SFEM, and the strain field is further assumed to be the weighted average between compatible stains from FEM and smoothed strains from NS-FEM. We prove theoretically that the strain energy obtained from the H-SFEM solution lies in between those from the compatible FEM solution and the NS-FEM solution, which guarantees the convergence of H-SFEM. Intensive numerical studies are conducted to verify these theoretical results and show that (1) the upper and lower bound solutions can always be obtained by adjusting ; (2) there exists a preferable at which the H-SFEM can produce the ultrasonic accurate solution.

On robust face recognition via sparse encoding : the good, the bad, and the ugly

In the field of face recognition, Sparse Representation (SR) has received considerable attention during the past few years. Most of the relevant literature focuses on holistic descriptors in closed-set identification applications. The underlying assumption in SR-based methods is that each class in the gallery has sufficient samples and the query lies on the subspace spanned by the gallery of the same class. Unfortunately, such assumption is easily violated in the more challenging face verification scenario, where an algorithm is required to determine if two faces (where one or both have not been seen before) belong to the same person. In this paper, we first discuss why previous attempts with SR might not be applicable to verification problems. We then propose an alternative approach to face verification via SR. Specifically, we propose to use explicit SR encoding on local image patches rather than the entire face. The obtained sparse signals are pooled via averaging to form multiple region descriptors, which are then concatenated to form an overall face descriptor. Due to the deliberate loss spatial relations within each region (caused by averaging), the resulting descriptor is robust to misalignment & various image deformations. Within the proposed framework, we evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder Neural Network (SANN), and an implicit probabilistic technique based on Gaussian Mixture Models. Thorough experiments on AR, FERET, exYaleB, BANCA and ChokePoint datasets show that the proposed local SR approach obtains considerably better and more robust performance than several previous state-of-the-art holistic SR methods, in both verification and closed-set identification problems. The experiments also show that l1-minimisation based encoding has a considerably higher computational than the other techniques, but leads to higher recognition rates.

Update on complex nonpharmacological interventions for breathlessness

Purpose of review: The study provides a review of current evidence about the role of complex nonpharmacological strategies in managing the multidimensional components of the breathlessness experience for individuals with life-limiting conditions. Recent findings: Evidence continues to demonstrate the significant impact of breathlessness on patients’ quality of life, day-to-day activity, and physical and psychosocial functioning. Recent evidence also confirms that patients draw on a number of self-initiated actions to cope with breathlessness, although many do not use strategies that are supported by a growing body of evidence from randomized controlled trials. Current literature supports the use of multicomponent, nonpharmacological interventions comprising strategies to improve breathing efficiency and reducing psychological distress to manage breathlessness. However trials of these approaches have mostly been conducted among patients with chronic obstructive pulmonary disease (COPD) or lung cancer, and few studies have investigated the benefits of nonpharmacological for patients in later stages of disease. Further investigation of interventions is required across a broader range of chronic life-limiting conditions. Addressing breathlessness and its co-occurring symptoms (symptom clusters) is also an area for future enquiry. Summary: The experience of breathlessness and strategies adopted by patients to manage the experience highlight the importance of multidimensional approaches to improve outcomes for patients with life-limiting conditions. There is good evidence to support the role of multicomponent, nonpharmacological interventions in reducing breathlessness for patients with COPD and lung cancer, although further studies are required to understand the particular clinical contexts in which such interventions are appropriate.

Effect of cold water immersion after exercise in the heat on muscle function, body temperatures, and vessel diameter

Cold water immersion (CWI) is a popular recovery modality, but actual physiological responses to CWI after exercise in the heat have not been well documented. The purpose of this study was to examine effects of 20-min CWI (14 degrees C) on neuromuscular function, rectal (T(re)) and skin temperature (T(sk)), and femoral venous diameter after exercise in the heat. Ten well-trained male cyclists completed two bouts of exercise consisting of 90-min cycling at a constant power output (216+/-12W) followed by a 16.1km time trial (TT) in the heat (32 degrees C). Twenty-five minutes post-TT, participants were assigned to either CWI or control (CON) recovery conditions in a counterbalanced order. T(re) and T(sk) were recorded continuously, and maximal voluntary isometric contraction torque of the knee extensors (MVIC), MVIC with superimposed electrical stimulation (SMVIC), and femoral venous diameters were measured prior to exercise, 0, 45, and 90min post-TT. T(re) was significantly lower in CWI beginning 50min post-TT compared with CON, and T(sk) was significantly lower in CWI beginning 25min post-TT compared with CON. Decreases in MVIC, and SMVIC torque after the TT were significantly greater for CWI compared with CON; differences persisted 90min post-TT. Femoral vein diameter was approximately 9% smaller for CWI compared with CON at 45min post-TT. These results suggest that CWI decreases T(re), but has a negative effect on neuromuscular function.

Preliminary evaluation of a primary care intervention for cry-fuss behaviours in the first 3-4 months of life (‘The Possums Approach’) : effects on cry-fuss behaviours and maternal mood

Problem crying in the first few months of life is both common and complex, arising out of multiple interacting and co-evolving factors. Parents whose babies cry and fuss a lot receive conflicting advice as they seek help from multiple health providers and emergency departments, and may be admitted into tertiary residential services. Conflicting advice is costly, and arises out of discipline-specific interpretations of evidence. An integrated, interdisciplinary primary care intervention (‘The Possums Approach’) for cry-fuss problems in the first months of life was developed from available peer-reviewed evidence. This study reports on preliminary evaluation of delivery of the intervention. A total of 20 mothers who had crying babies under 16 weeks of age (average age 6.15 weeks) completed questionnaires, including the Crying Patterns Questionnaire and the Edinburgh Postnatal Depression Scale, before and 3-4 weeks after their first consultation with trained primary care practitioners. Preliminary evaluation is promising. The Crying Patterns Questionnaire showed a significant decrease in crying and fussing duration, by 1 h in the evening (P = 0.001) and 30 min at night (P = 0.009). The median total amount of crying and fussing in a 24-h period was reduced from 6.12 to 3 h. The Edinburgh Postnatal Depression Scale showed a significant improvement in depressive symptoms, with the median score decreasing from 11 to 6 (P = 0.005). These findings are corroborated by an analysis of results for the subset of 16 participants whose babies were under 12 weeks of age (average age 4.71 weeks). These preliminary results demonstrate significantly decreased infant crying in the evening and during the night and improved maternal mood, validating an innovative interdisciplinary clinical intervention for cry-fuss problems in the first few months of life. This intervention, delivered by trained health professionals, has the potential to mitigate the costly problem of health professionals giving discipline-specific and conflicting advice post-birth.

Non-intrusive continuous user behaviour analysis using computerized systems

There are different ways to authenticate humans, which is an essential prerequisite for access control. The authentication process can be subdivided into three categories that rely on something someone i) knows (e.g. password), and/or ii) has (e.g. smart card), and/or iii) is (biometric features). Besides classical attacks on password solutions and the risk that identity-related objects can be stolen, traditional biometric solutions have their own disadvantages such as the requirement of expensive devices, risk of stolen bio-templates etc. Moreover, existing approaches provide the authentication process usually performed only once initially. Non-intrusive and continuous monitoring of user activities emerges as promising solution in hardening authentication process: iii-2) how so. behaves. In recent years various keystroke dynamic behavior-based approaches were published that are able to authenticate humans based on their typing behavior. The majority focuses on so-called static text approaches, where users are requested to type a previously defined text. Relatively few techniques are based on free text approaches that allow a transparent monitoring of user activities and provide continuous verification. Unfortunately only few solutions are deployable in application environments under realistic conditions. Unsolved problems are for instance scalability problems, high response times and error rates. The aim of this work is the development of behavioral-based verification solutions. Our main requirement is to deploy these solutions under realistic conditions within existing environments in order to enable a transparent and free text based continuous verification of active users with low error rates and response times.

Solving inherent problems of anomaly detection by cooperation

Anomaly detection compensates shortcomings of signature-based detection such as protecting against Zero-Day exploits. However, Anomaly Detection can be resource-intensive and is plagued by a high false-positive rate. In this work, we address these problems by presenting a Cooperative Intrusion Detection approach for the AIS, the Artificial Immune System, as an example for an anomaly detection approach. In particular we show, how the cooperative approach reduces the false-positive rate of the detection and how the overall detection process can be organized to account for the resource constraints of the participating devices. Evaluations are carried out with the novel network simulation environment NeSSi as well as formally with an extension to the epidemic spread model SIR

Problems and awareness of support services among University students

Although Australian universities have allocated significant resources toward the development of student support services, administrators have little systematic information about the problems undergraduate university students experience or students' knowledge about available support services. The author surveyed 441 students in an urban, nonresidential university to examine the prevalence of difficulties associated with learning, sexual harassment, discrimination, emotional distress, health problems, course and career concerns, financial difficulties, and difficulties with lecturers; he also assessed students' knowledge of support services in each of these areas. Course concerns were the most common problem, followed by emotional distress, worry about career choices, financial difficulties, and problems with lecturers. More than half of the students were unaware of the support services available to them to address a range of concerns from sexual harassment and discrimination to emotional distress. Approximately 20% of the students reported having used university counseling or career services. Implications for targeting specific areas for outreach programs are discussed.

Markers of disease severity are associated with malnutrition in Parkinson's disease

Objective In Parkinson's disease (PD), commonly reported risk factors for malnutrition in other populations commonly occur. Few studies have explored which of these factors are of particular importance in malnutrition in PD. The aim was to identify the determinants of nutritional status in people with Parkinson's disease (PWP). Methods Community-dwelling PWP (>18 years) were recruited (n = 125; 73M/52F; Mdn 70 years). Self-report assessments included Beck's Depression Inventory (BDI), Spielberger Trait Anxiety Inventory (STAI), Scales for Outcomes in Parkinson's disease – Autonomic (SCOPA-AUT), Modified Constipation Assessment Scale (MCAS) and Freezing of Gait Questionnaire (FOG-Q). Information about age, PD duration, medications, co-morbid conditions and living situation was obtained. Addenbrooke's Cognitive Examination (ACE-R), Unified Parkinson's Disease Rating Scale (UPDRS) II and UPDRS III were performed. Nutritional status was assessed using the Subjective Global Assessment (SGA) as part of the scored Patient-Generated Subjective Global Assessment (PG-SGA). Results Nineteen (15%) were malnourished (SGA-B). Median PG-SGA score was 3. More of the malnourished were elderly (84% vs. 71%) and had more severe disease (H&Y: 21% vs. 5%). UPDRS II and UPDRS III scores and levodopa equivalent daily dose (LEDD)/body weight(mg/kg) were significantly higher in the malnourished (Mdn 18 vs. 15; 20 vs. 15; 10.1 vs. 7.6 respectively). Regression analyses revealed older age at diagnosis, higher LEDD/body weight (mg/kg), greater UPDRS III score, lower STAI score and higher BDI score as significant predictors of malnutrition (SGA-B). Living alone and higher BDI and UPDRS III scores were significant predictors of a higher log-adjusted PG-SGA score. Conclusions In this sample of PWP, the rate of malnutrition was higher than that previously reported in the general community. Nutrition screening should occur regularly in those with more severe disease and depression. Community support should be provided to PWP living alone. Dopaminergic medication should be reviewed with body weight changes.