Personalised medicine is replacing the one-drug-fits-all approach with many prognostic models incorporating biomarkers available for risk stratifying patients. Evidence has been emerging that the effects of biomarkers change over time and therefore violate the assumption of proportional hazards when performing Cox regression. Analysis using the Cox model when the assumptions are invalid can result in misleading conclusions. This thesis reviews existing approaches for the analysis of non-proportional effects with respect to survival data. A number of well-developed approaches were identified but to date their uptake in practice has been limited. There is a need for more widespread use of flexible modelling to move away from standard analysis using a Cox model when the assumption of proportional hazards is violated. Two novel approaches were applied to investigate the impact of follow-up duration on two residual risk models, IHC4 and Mammostrat, for predicting risk in early breast cancers using two studies with different lengths of follow up; the Edinburgh Breast Conservation Series (BCS) and the Tamoxifen versus Exemestane Adjuvant Multinational (TEAM) trial. Similar results were observed between the two approaches that were considered, the multivariable fractional polynomial time (MFPT) approach and Royston-Parmer flexible parametric models, with their respective advantages and disadvantages being discussed. The analyses identified a strong time-varying effect of IHC4 score with the prognostic effect of IHC4 score on time-to distant recurrence decreasing with increasing follow-up time. Mammostrat score identified a group of patients with an increased risk of distant recurrence over full follow-up in the TEAM and Edinburgh BCS cohorts. The results suggest a combined IHC4 and Mammostrat risk score could provide information on the risk of recurrence and warrants further study.

The electromyogram (EMG) signal is desired to be used as a control signal for applications such as multifunction prostheses, wheelchair navigation, gait generation, grasping control, virtual keyboards, and gesture-based interfaces [25]. Several research studies have attempted to relate the electromyogram (EMG) activity of the forearm muscles to the mechanical activity of the wrist, hand and/or fingers [41], [42], [43]. A primary interest is for EMG control of powered upper-limb prostheses and rehabilitation orthotics. Existing commercial EMG-controlled devices are limited to rudimentary control capabilities of either discrete states (e.g. hand close/open), or one degree of freedom proportional control [4], [36]. Classification schemes for discriminating between hand/wrist functions and individual finger movements have demonstrated accuracy up to 95% [38], [39], [29]. These methods may provide for increased amputee function, though continuous control of movement is not generally achieved. This thesis considered proportional control via EMG-based estimation of finger forces with the goal of identifying whether multiple degrees of freedom of proportional control information are available from the surface EMG of the forearm. Electromyogram (EMG) activity from the extensor and flexor muscles of the forearm was sensed with bipolar surface electrodes and related to the force produced at the four fingertips during constant-posture, slowly force-varying contractions from 20 healthy subjects. The contractions ranged between 30% maximum voluntary contractions (MVC) extension and 30% MVC flexion. EMG amplitude sampling rate, least squares regularization, linear vs. nonlinear models and number of electrodes used in the system identification were studied. Results are supportive that multiple degrees of freedom of proportional control information are available from the surface EMG of the forearm, at least in healthy subjects. An EMG amplitude sampling frequency of 4.096 Hz was found to produce models which allowed for good EMG amplitude estimates. Least squares regularization with a pseudo-inverse tolerance of 0.055 resulted in significant improvement in modeling results, with an average error of 4.69% MVC-6.59% MVC (maximum voluntary contraction). Increasing polynomial order did not significantly improve modeling results. Results from smaller electrode arrays remained fairly good with as few as six electrodes, with the average %MVC error ranging from 5.13%-7.01% across the four fingers. This study also identified challenges in the current experimental study design and subsequent system identification when EMG-force modeling is performed with four fingers simultaneously. Methods to compensate for these issues have been proposed in this thesis.

This dissertation focuses on developing novel model selection techniques, the process by which a statistician selects one of a number of competing models of varying dimensions, under an array of different statistical assumptions on observed data. Traditionally, two main reasons have been advocated by researchers for performing model selection strategies over classical maximum likelihood estimates (MLEs). The first reason is prediction accuracy, where by shrinking or setting to zero some model parameters, one sacrifices the unbiasedness of MLEs for a reduced variance, which in turn leads to an overall improvement in predictive performance. The second reason relates to interpretability of the selected models in the presence of a large number of predictors, where in order to obtain a parsimonious representation exhibiting the relationship between the response and covariates, we are willing to sacrifice some of the smaller details brought in by spurious predictors. In the first part of this work, we revisit the family of variable selection techniques known as sure independence screening procedures for generalized linear models and the Cox proportional hazards model. After clever combination of some of its most powerful variants, we propose new extensions based on the idea of sample splitting, data-driven thresholding, and combinations thereof. A publicly available package developed in the R statistical software demonstrates considerable improvements in terms of model selection and competitive computational time between our enhanced variable selection procedures and traditional penalized likelihood methods applied directly to the full set of covariates. Next, we develop model selection techniques within the framework of statistical network analysis for two frequent problems arising in the context of stochastic blockmodels: community number selection and change-point detection. In the second part of this work, we propose a composite likelihood based approach for selecting the number of communities in stochastic blockmodels and its variants, with robustness consideration against possible misspecifications in the underlying conditional independence assumptions of the stochastic blockmodel. Several simulation studies, as well as two real data examples, demonstrate the superiority of our composite likelihood approach when compared to the traditional Bayesian Information Criterion or variational Bayes solutions. In the third part of this thesis, we extend our analysis on static network data to the case of dynamic stochastic blockmodels, where our model selection task is the segmentation of a time-varying network into temporal and spatial components by means of a change-point detection hypothesis testing problem. We propose a corresponding test statistic based on the idea of data aggregation across the different temporal layers through kernel-weighted adjacency matrices computed before and after each candidate change-point, and illustrate our approach on synthetic data and the Enron email corpus. The matrix completion problem consists in the recovery of a low-rank data matrix based on a small sampling of its entries. In the final part of this dissertation, we extend prior work on nuclear norm regularization methods for matrix completion by incorporating a continuum of penalty functions between the convex nuclear norm and nonconvex rank functions. We propose an algorithmic framework for computing a family of nonconvex penalized matrix completion problems with warm-starts, and present a systematic study of the resulting spectral thresholding operators. We demonstrate that our proposed nonconvex regularization framework leads to improved model selection properties in terms of finding low-rank solutions with better predictive performance on a wide range of synthetic data and the famous Netflix data recommender system.

Cranial vault thickness (CVT) has been reported at many different osteometric landmarks and features on the vault. Historically, only a few landmarks are used, often bregma, lambda, vertex, and right and left euryon, and frequently comparisons are based only on “thick” versus “thin” to describe the vault overall. What is inherent in this strategy is the use of a few locations to characterize the entire vault. The problem remains that there is little information concerning CVT variation throughout an individual's vault, and the causes of variation within recent Homo sapiens important to investigating thickness variation between species in Homo. This work describes thickness variation over the entire superior vault and compares the sexes, age groups and populations in recent H. sapiens. A proportional grid is applied to the superior vault to measure thickness at 219 sampling points in a geographically diverse sample of recent H. sapiens. Thickness values are analyzed in their two-dimensional spatial relationships to determine patterns of vault thickness. Males were identified to be thicker than females at more lateral locations and along the midsagittal plane, although this finding is not statistically significant. Individuals over the age of 45 years are found to be statistically significantly thicker than individuals younger than 31 years at more lateral locations of the vault. Aboriginal Australians are statistically significantly thicker at more lateral locations of the vault than any other populations, whereas Northern Canada/Greenland individuals were thinner than other populations at these locations. The trend of thicker vaults in the older age group and the Australians is identified across the vault, although is not statistically significant at more locations. Several thickness patterns are identified. The boss thickening pattern is the most common pattern, followed by a midsagittal pattern, a posterior pattern, and an anterior pattern. Some specimens do not demonstrate thickness variation and are coded as undifferentiated. Each pattern is observed alone and in combination with others, signifying that pattern causes are not mutually exclusive. Boss thickening is interpreted as the result of passive bone thickening during normal bone and brain growth during fetal and adolescent development. The midsagittal thickness pattern coincides with inferred strain along the sagittal suture from nuchal muscle engagement during mastication. Previous researchers have proposed adaptive explanations for thickness variation, such as protection from interpersonal violence; the patterns of cranial vault thickness reported here point to normal growth and development of the brain as a driving force, a relationship that could drive thickness variation in other Homo species. Comparing thickness at bregma, and the frontal and parietal eminences for recent H. sapiens and H. erectus, there is no statistical difference between African and Asian H. erectus, and between the on average thicker H. sapiens populations and H. erectus, based on published data. Future work will investigate the presence or absence of thickness patterns in these fossil species.

Proportional Solenoid Valves (PSV) have been successfully used in the hydraulic industry for many years due to the benefits associated with higher accuracy compared to on/off solenoid valves, and the robustness and cost compared to servo valves. Because the PSV plays an important role in the performance of a hydraulic system, a technique commonly referred to as Condition Monitoring Scheme (CMS) has been used extensively to monitor the progress of faults in the PSV. But before any CMS can be implemented on a system, it needs to be thoroughly tested for its reliability of fault detection since, a failure of the CMS to detect any potential fault can be economically disastrous, and dangerous in terms of the safety of personnel. The motivation of this research was to develop a fault simulator which could reliably and repeatedly induce user defined faults in the PSV and thereby aid in testing the efficacy of the CMS for monitoring such simulated faults.<p>Industry research has revealed that the most common mode of failure in spool valves is an increase in the friction between the spool and valve, due to wear, contamination and dirt, which renders the valve inoperable. In this research, a non-destructive fault simulator was developed which induced artificial friction faults in the PSV. The PSV consisted of two solenoids on the opposite sides of the valve spool by virtue of which, bi-directional position control could be achieved. The PSV with the spool and one of the solenoids was used as the system in which the faults were simulated, and the second solenoid was used an a fault simulator for inducing the desired friction characteristics in the system. <p>The friction characteristics induced in the valve were similar to those in the classical friction curve, i.e., stiction at low velocities and Coulomb and viscous friction at higher velocities. By employing a closed loop position control scheme, one of the solenoids was used to generate a linearly increasing velocity profile by virtue of which the desired friction characteristics could be induced in different velocity regimes. The other solenoid was used to generate the desired friction force. A closed loop force control strategy, which used the feedback from a force transducer, allowed for the accurate control of the friction characteristics. stiction was induced at low velocities by passing the required current in both the solenoids that resulted in no net force on the valve spool. Due to the absence of any driving force the spool was stalled at the desired location, thus achieving the same effect of stiction at low velocities. The coulomb and viscous friction were induced at higher velocities by employing an algorithm which was a function of the spool velocity. Different magnitudes of static, coulomb and viscous friction were induced to achieve the friction characteristics represented by the classical friction curve. Since the change in force characteristics of the valve results in a corresponding change in the current drawn by the position control solenoid, a rudimentary CMS for monitoring the current characteristics is presented. Based on the experimental results and validation using the CMS it was concluded that the fault simulator was able to accurately produce the desired frictional loading on the valve spool and was able to do so with a high degree of repeatability. Proportional Solenoid Valves (PSV) have been successfully used in the hydraulic industry for many years due to the benefits associated with higher accuracy compared to on/off solenoid valves, and the robustness and cost compared to servo valves. Because the PSV plays an important role in the performance of a hydraulic system, a technique commonly referred to as Condition Monitoring Scheme (CMS) has been used extensively to monitor the progress of faults in the PSV. But before any CMS can be implemented on a system, it needs to be thoroughly tested for its reliability of fault detection since, a failure of the CMS to detect any potential fault can be economically disastrous, and dangerous in terms of the safety of personnel. The motivation of this research was to develop a fault simulator which could reliably and repeatedly induce user defined faults in the PSV and thereby aid in testing the efficacy of the CMS for monitoring such simulated faults. Industry research has revealed that the most common mode of failure in spool valves is an increase in the friction between the spool and valve, due to wear, contamination and dirt, which renders the valve inoperable. In this research, a non-destructive fault simulator was developed which induced artificial friction faults in the PSV. The PSV consisted of two solenoids on the opposite sides of the valve spool by virtue of which, bi-directional position control could be achieved.The PSV with the spool and one of the solenoids was used as the system in which the faults were simulated, and the second solenoid was used an a fault simulator for inducing the desired friction characteristics in the system. The friction characteristics induced in the valve were similar to those in the classical friction curve, i.e., stiction at low velocities and Coulomb and viscous friction at higher velocities. By employing a closed loop position control scheme, one of the solenoids was used to generate a linearly increasing velocity profile by virtue of which the desired friction characteristics could be induced in different velocity regimes. The other solenoid was used to generate the desired friction force. A closed loop force control strategy, which used the feedback from a force transducer, allowed for the accurate control of the friction characteristics. stiction was induced at low velocities by passing the required current in both the solenoids that resulted in no net force on the valve spool. Due to the absence of any driving force the spool was stalled at the desired location, thus achieving the same effect of stiction at low velocities. The coulomb and viscous friction were induced at higher velocities by employing an algorithm which was a function of the spool velocity. Different magnitudes of static, coulomb and viscous friction were induced to achieve the friction characteristics represented by the classical friction curve. Since the change in force characteristics of the valve results in a corresponding change in the current drawn by the position control solenoid, a rudimentary CMS for monitoring the current characteristics is presented. Based on the experimental results and validation using the CMS it was concluded that the fault simulator was able to accurately produce the desired frictional loading on the valve spool and was able to do so with a high degree of repeatability.

Proportional Solenoid Valves (PSV) have been successfully used in the hydraulic industry for many years due to the benefits associated with higher accuracy compared to on/off solenoid valves, and the robustness and cost compared to servo valves. Because the PSV plays an important role in the performance of a hydraulic system, a technique commonly referred to as Condition Monitoring Scheme (CMS) has been used extensively to monitor the progress of faults in the PSV. But before any CMS can be implemented on a system, it needs to be thoroughly tested for its reliability of fault detection since, a failure of the CMS to detect any potential fault can be economically disastrous, and dangerous in terms of the safety of personnel. The motivation of this research was to develop a fault simulator which could reliably and repeatedly induce user defined faults in the PSV and thereby aid in testing the efficacy of the CMS for monitoring such simulated faults.<p>Industry research has revealed that the most common mode of failure in spool valves is an increase in the friction between the spool and valve, due to wear, contamination and dirt, which renders the valve inoperable. In this research, a non-destructive fault simulator was developed which induced artificial friction faults in the PSV. The PSV consisted of two solenoids on the opposite sides of the valve spool by virtue of which, bi-directional position control could be achieved. The PSV with the spool and one of the solenoids was used as the system in which the faults were simulated, and the second solenoid was used an a fault simulator for inducing the desired friction characteristics in the system. <p>The friction characteristics induced in the valve were similar to those in the classical friction curve, i.e., stiction at low velocities and Coulomb and viscous friction at higher velocities. By employing a closed loop position control scheme, one of the solenoids was used to generate a linearly increasing velocity profile by virtue of which the desired friction characteristics could be induced in different velocity regimes. The other solenoid was used to generate the desired friction force. A closed loop force control strategy, which used the feedback from a force transducer, allowed for the accurate control of the friction characteristics. stiction was induced at low velocities by passing the required current in both the solenoids that resulted in no net force on the valve spool. Due to the absence of any driving force the spool was stalled at the desired location, thus achieving the same effect of stiction at low velocities. The coulomb and viscous friction were induced at higher velocities by employing an algorithm which was a function of the spool velocity. Different magnitudes of static, coulomb and viscous friction were induced to achieve the friction characteristics represented by the classical friction curve. Since the change in force characteristics of the valve results in a corresponding change in the current drawn by the position control solenoid, a rudimentary CMS for monitoring the current characteristics is presented. Based on the experimental results and validation using the CMS it was concluded that the fault simulator was able to accurately produce the desired frictional loading on the valve spool and was able to do so with a high degree of repeatability. Proportional Solenoid Valves (PSV) have been successfully used in the hydraulic industry for many years due to the benefits associated with higher accuracy compared to on/off solenoid valves, and the robustness and cost compared to servo valves. Because the PSV plays an important role in the performance of a hydraulic system, a technique commonly referred to as Condition Monitoring Scheme (CMS) has been used extensively to monitor the progress of faults in the PSV. But before any CMS can be implemented on a system, it needs to be thoroughly tested for its reliability of fault detection since, a failure of the CMS to detect any potential fault can be economically disastrous, and dangerous in terms of the safety of personnel. The motivation of this research was to develop a fault simulator which could reliably and repeatedly induce user defined faults in the PSV and thereby aid in testing the efficacy of the CMS for monitoring such simulated faults. Industry research has revealed that the most common mode of failure in spool valves is an increase in the friction between the spool and valve, due to wear, contamination and dirt, which renders the valve inoperable. In this research, a non-destructive fault simulator was developed which induced artificial friction faults in the PSV. The PSV consisted of two solenoids on the opposite sides of the valve spool by virtue of which, bi-directional position control could be achieved.The PSV with the spool and one of the solenoids was used as the system in which the faults were simulated, and the second solenoid was used an a fault simulator for inducing the desired friction characteristics in the system. The friction characteristics induced in the valve were similar to those in the classical friction curve, i.e., stiction at low velocities and Coulomb and viscous friction at higher velocities. By employing a closed loop position control scheme, one of the solenoids was used to generate a linearly increasing velocity profile by virtue of which the desired friction characteristics could be induced in different velocity regimes. The other solenoid was used to generate the desired friction force. A closed loop force control strategy, which used the feedback from a force transducer, allowed for the accurate control of the friction characteristics. stiction was induced at low velocities by passing the required current in both the solenoids that resulted in no net force on the valve spool. Due to the absence of any driving force the spool was stalled at the desired location, thus achieving the same effect of stiction at low velocities. The coulomb and viscous friction were induced at higher velocities by employing an algorithm which was a function of the spool velocity. Different magnitudes of static, coulomb and viscous friction were induced to achieve the friction characteristics represented by the classical friction curve. Since the change in force characteristics of the valve results in a corresponding change in the current drawn by the position control solenoid, a rudimentary CMS for monitoring the current characteristics is presented. Based on the experimental results and validation using the CMS it was concluded that the fault simulator was able to accurately produce the desired frictional loading on the valve spool and was able to do so with a high degree of repeatability.

Considerable amounts of event history data are collected through longitudinal surveys. These surveys have many particularities or features that are the results of the dynamic nature of the population under study and of the fact that data collected through longitudinal surveys involve the use of complex survey designs, with clustering and stratification. These particularities include: attrition, seam-effect, censoring, left-truncation and complications in the variance estimation due to the use of complex survey designs. This thesis focuses on the last two points. Statistical methods based on the stratified Cox proportional hazards model that account for intra-cluster dependence, when the sampling design is uninformative, are proposed. This is achieved using the theory of estimating equations in conjunction with empirical process theory. Issues concerning analytic inference from survey data and the use of weighted versus unweighted procedures are also discussed. The proposed methodology is applied to data from the U. S. Survey of Income and Program Participation (SIPP) and data from the Canadian Survey of Labour and Income Dynamics (SLID). Finally, different statistical methods for handling left-truncated sojourns are explored and compared. These include the conditional partial likelihood and other methods, based on the Exponential or the Weibull distributions.

Recent experiments are bringing to the fore more and more information about the effects of different treatments on the gene expression of different genes. The results obtained from these experiments show that some definite trends are observed in different genes in the Human Embryonic Kidney and Human Colon Adenocarcinoma Grade II cell lines. The difference in the gene expressions of the two cell lines motivates the problem in this thesis. The thesis provided intervention methods to make the colon cancer cell line genes behave more like their Human Embryonic Kidney cell line counterparts. Two methods of intervention were introduced. The first method was the simpler on-off control intervention while the second method used a more advanced proportional integral control to meet the goal. A comparison of these two intervention methods showed the clear implementational advantages of proportional integral control over on-off control.

The purpose of this study was to determine how well the Monte Carlo transport code FLUKA can simulate a tissue-equivalent proportional counter (TEPC) and produce the expected delta ray events when exposed to high energy heavy ions (HZE) like in the galactic cosmic ray (GCR) environment. Accurate transport codes are desirable because of the high cost of beam time, the inability to measure the mixed field GCR on the ground and the flexibility they offer in the engineering and design process. A spherical TEPC simulating a 1 um site size was constructed in FLUKA and its response was compared to experimental data for an 56Fe beam at 360 MeV/nucleon. The response of several narrow beams at different impact parameters were used to explain the features of the response of the same detector exposed to a uniform field of radiation. Additionally, an investigation was made into the effect of the wall thickness on the response of the TEPC and the range of delta rays in the tissue-equivalent (TE) wall material. A full impact parameter test (from IP = 0 to IP = detector radius) was performed to show that FLUKA produces the expected wall effect. That is, energy deposition in the gas volume can occur even when the primary beam does not pass through the gas volume. A final comparison to experimental data was made for the simulated TEPC exposed to various broad beams in the energy range of 200 - 1000 MeV/nucleon. FLUKA overestimated energy deposition in the gas volume in all cases. The FLUKA results differed from the experimental data by an average of 25.2 % for yF and 12.4 % for yD. It is suggested that this difference can be reduced by adjusting the FLUKA default ionization potential and density correction factors.

In Wideband DS-CDMA system, increasing the utilization of Orthogonal Variable Spreading Factor (OVSF) code tree can serve more users. As Dynamic Code Assignment (DCA) does, the allocated codes will be reassigned when a new call can¡¦t be served even if the capacity is enough. However, reassigning occupied codes is expensive. In this thesis, we have proposed a state classification code assignment principle and a proportional reservation policy. In the proposed algorithm, the OVSF code tree is classified into four states. Then, the system will decide whether to reassign codes or not by checking the current state of the tree. In other words, the occupied codes will be reallocated in an appropriate occasion. Besides, we reserve vacant and available codes corresponding to the probability of requests for each supported rate when reassignment occurs. Therefore, the reassignment is reduced. It is also proved on the basis of the property of Markov Chain. Taking advantage of the proposed states classification code assignment and the policy of proportional code reservation, the call blocking rate and the number of reassignment can be reduced.