Suicide, statistics and the coroner: A comparative study of death investigations

Australia has a significantly higher suicide rate than England. Rather than accepting that this ‘statistical fact’ is a direct reflection of some positivist truth, this paper begins with the premise that how suicide is counted depends upon what counts as suicide. This study involves semi-structured interviews with coroners both in Australia and England, as well as observations at inquests. Important differences between the two coronial systems include: first, quite different logics of operation; second, the burden of proof for reaching a finding of suicide is significantly higher in England; and third, the presence of family members at English inquests results in far greater pressure being brought to bear upon coroners. These combined factors result in a reduced likelihood of English coroners reaching a finding of suicide. The conclusions are twofold. First, this research supports existing criticisms of comparative suicide statistics. Second, this research adds theoretical weight to criticisms of positivist analyses of social phenomena.

Predicting reduced driver alertness on monotonous highways : a driving simulator study

Impaired driver alertness increases the likelihood of drivers’ making mistakes and reacting too late to unexpected events while driving. This is particularly a concern on monotonous roads, where a driver’s attention can decrease rapidly. While effective countermeasures do not currently exist, the development of in-vehicle sensors opens avenues for monitoring driving behavior in real-time. The aim of this study is to predict drivers’ level of alertness through surrogate measures collected from in-vehicle sensors. Electroencephalographic activity is used as a reference to evaluate alertness. Based on a sample of 25 drivers, data was collected in a driving simulator instrumented with an eye tracking system, a heart rate monitor and an electrodermal activity device. Various classification models were tested from linear regressions to Bayesians and data mining techniques. Results indicated that Neural Networks were the most efficient model in detecting lapses in alertness. Findings also show that reduced alertness can be predicted up to 5 minutes in advance with 90% accuracy, using surrogate measures such as time to line crossing, blink frequency and skin conductance level. Such a method could be used to warn drivers of their alertness level through the development of an in-vehicle device monitoring, in real-time, drivers' behavior on highways.

Short-data recursive HMM parameter estimation for rapid vision-based aircraft heading estimation

Rapid recursive estimation of hidden Markov Model (HMM) parameters is important in applications that place an emphasis on the early availability of reasonable estimates (e.g. for change detection) rather than the provision of longer-term asymptotic properties (such as convergence, convergence rate, and consistency). In the context of vision- based aircraft (image-plane) heading estimation, this paper suggests and evaluates the short-data estimation properties of 3 recursive HMM parameter estimation techniques (a recursive maximum likelihood estimator, an online EM HMM estimator, and a relative entropy based estimator). On both simulated and real data, our studies illustrate the feasibility of rapid recursive heading estimation, but also demonstrate the need for careful step-size design of HMM recursive estimation techniques when these techniques are intended for use in applications where short-data behaviour is paramount.

Roundabout design and cycling safety

Roundabouts reduce the frequency and severity of motor vehicle crashes and therefore the number installed has increased dramatically in the last 20 years in many countries. However, the safety impacts of roundabouts for bicycle riders are a source of concern, with many studies reporting lower injury reductions for cyclists than car occupants. This paper summarises the results of a project undertaken to provide guidance on how cyclist safety could be improved at existing roundabouts in Queensland, Australia, where cyclist crashes have been increasing and legislation gives motor vehicles priority over cyclists and pedestrians at roundabouts. The review of international roundabout design guidelines identified two schools of design: tangential roundabouts (common in English-speaking countries, including Australia), which focus on minimising delay to motor vehicles, and radial roundabouts (common in continental Europe), which focus on speed reduction and safety. While it might be expected that radial roundabouts would be safer for cyclists, there have been no studies to confirm this view. Most guidelines expect cyclists to act as vehicle traffic in single-lane, typically low-speed, roundabouts. Some jurisdictions do not permit cyclists to travel on multi-lane roundabouts, and recommend segregated bicycle facilities because of their lowest crash risk for cyclists. Given that most bicycle-vehicle crashes at roundabouts involve an entering vehicle and a circulating cyclist, the greatest challenges appear to be reducing the speed of motor vehicles on the approach/entry to roundabouts and other ways of maximizing the likelihood that cyclists will be seen. Lower entry speeds are likely to underpin the greater safety of compact roundabouts for cyclists and, conversely, the higher than expected crash rates at two-lane roundabouts. European research discourages the use of bike lanes in roundabouts which position cyclists at the edge of the road and contributes to cyclists being less likely to be noticed by drivers.

Decision support for the efficient annotation of bioacoustic events

Acoustic sensors allow scientists to scale environmental monitoring over large spatiotemporal scales. The faunal vocalisations captured by these sensors can answer ecological questions, however, identifying these vocalisations within recorded audio is difficult: automatic recognition is currently intractable and manual recognition is slow and error prone. In this paper, a semi-automated approach to call recognition is presented. An automated decision support tool is tested that assists users in the manual annotation process. The respective strengths of human and computer analysis are used to complement one another. The tool recommends the species of an unknown vocalisation and thereby minimises the need for the memorization of a large corpus of vocalisations. In the case of a folksonomic tagging system, recommending species tags also minimises the proliferation of redundant tag categories. We describe two algorithms: (1) a “naïve” decision support tool (16%–64% sensitivity) with efficiency of O(n) but which becomes unscalable as more data is added and (2) a scalable alternative with 48% sensitivity and an efficiency ofO(log n). The improved algorithm was also tested in a HTML-based annotation prototype. The result of this work is a decision support tool for annotating faunal acoustic events that may be utilised by other bioacoustics projects.

Leptospirosis from water sources

Leptospirosis outbreaks have been associated with many common water events including water consumption, water sports, environmental disasters and occupational exposure. The ability of leptospires to survive in moist environments makes them a high risk agent for infection following contact with any contaminated water source. Water treatment processes reduce the likelihood of leptospirosis or other microbial agents causing infection provided they do not malfunction and the distribution networks are maintained. Notably, there are many differences in water treatment systems around the world, particularly between developing and developed countries. Detection of leptospirosis in water samples is uncommonly performed by molecular methods.

Statistical methods for electromyography data and associated problems

This thesis proposes three novel models which extend the statistical methodology for motor unit number estimation, a clinical neurology technique. Motor unit number estimation is important in the treatment of degenerative muscular diseases and, potentially, spinal injury. Additionally, a recent and untested statistic to enable statistical model choice is found to be a practical alternative for larger datasets. The existing methods for dose finding in dual-agent clinical trials are found to be suitable only for designs of modest dimensions. The model choice case-study is the first of its kind containing interesting results using so-called unit information prior distributions.

Pre-processing for approximate Bayesian computation in image analysis

Most of the existing algorithms for approximate Bayesian computation (ABC) assume that it is feasible to simulate pseudo-data from the model at each iteration. However, the computational cost of these simulations can be prohibitive for high dimensional data. An important example is the Potts model, which is commonly used in image analysis. Images encountered in real world applications can have millions of pixels, therefore scalability is a major concern. We apply ABC with a synthetic likelihood to the hidden Potts model with additive Gaussian noise. Using a pre-processing step, we fit a binding function to model the relationship between the model parameters and the synthetic likelihood parameters. Our numerical experiments demonstrate that the precomputed binding function dramatically improves the scalability of ABC, reducing the average runtime required for model fitting from 71 hours to only 7 minutes. We also illustrate the method by estimating the smoothing parameter for remotely sensed satellite imagery. Without precomputation, Bayesian inference is impractical for datasets of that scale.

Adsorption and mobility of metals in build-up on road surfaces

The study investigated the adsorption and bioavailability characteristics of traffic generated metals common to urban land uses, in road deposited solids particles. To validate the outcomes derived from the analysis of field samples, adsorption and desorption experiments were undertaken. The analysis of field samples revealed that metals are selectively adsorbed to different charge sites on solids. Zinc, copper, lead and nickel are adsorbed preferentially to oxides of manganese, iron and aluminium. Lead is adsorbed to organic matter through chemisorption. Cadmium and chromium form weak bonding through cation exchange with most of the particle sizes. Adsorption and desorption experiments revealed that at high metal concentrations, chromium, copper and lead form relatively strong bonds with solids particles while zinc is adsorbed through cation exchange with high likelihood of being released back into solution. Outcomes from this study provide specific guidance for the removal of metals from stormwater based on solids removal.

Long-tailed bats' use of a Pinus radiata stand in Kinleith Forest : Recommendations for monitoring

Targeted monitoring of threatened species within plantations is becoming more important due to forest certification programmes’ requirement to consider protection of threatened species, and to increase knowledge of the distribution of species. To determine patterns of long-tailed bat (Chalinolobus tuberculatus) activity in different habitat structures, with the aim of improving the likelihood of detection by targeting monitoring, we monitored one stand of 26 year-old Pinus radiata over seven months between December 2007 and June 2008 in Kinleith Forest, an exotic plantation forest centred around Tokoroa, South Waikato, New Zealand. Activity was determined by acoustic recording equipment, which is able to detect and record bats’ echolocation calls. We monitored activity from sunset to sunrise along a road through the stand, along stand edges, and in the interior of the stand. Bats were recorded on 80% of the 35 nights monitored. All activity throughout the monitoring period was detected on the edge of the stand or along the road. No bats were detected within the interior of the stand. Bat activity was highest along the road through the stand (40.4% of all passes), followed by an edge with stream running alongside (35.2%), along the road within a skidsite (19.8%), and along an edge without a stream (4.6%). There was a significant positive relationship between bat pass rate (bat passes h-1) and the feeding buzz rate (feeding buzzes h-1) indicating that bat activity was associated with feeding and not just commuting. Bat feeding activity was also highest along the road through the stand (59.2% of feeding buzzes), followed by the road within the skidsite (30.6%), and along the stream-side edge (10.2%). No feeding buzzes were recorded in either the interior or along the edge without the stream. Differences in overall feeding activity were significant only between the road and edge and between edges with and without a stream. Bat activity was detected each month and always by the second night of monitoring, and in this stand was highest during April. We recommend targeted monitoring for long-tailed bats be focused on road-side and stand edge habitat, and along streams, and that monitoring take place for at least three nights to maximise probability of detection.

Incorporation of road safety into road management systems

Road collisions negatively affect the lives of hundreds of Canadians per year. Unfortunately, safety has been typically neglected from management systems. It is common to find that a great deal of effort has been devoted to develop and implement systems capable of achieving and sustaining good levels of condition. It is relatively recent that road safety has become an important objective. Managing a network of roads is not an easy task; it requires long, medium and short term plans to maintain, rehabilitate and upgrade aging assets, reduce and mitigate accident exposure, likelihood and severity. This thesis presents a basis for incorporating road safety into road management systems; two case studies were developed; one limited by available data and another from sufficient information. A long term analysis was used to allocate improvements for condition and safety of roads and bridges, at the network level. It was confirmed that a safety index could be used to obtain a first cut model; meanwhile potential for improvement which is a difference between observed and predicted number of accidents was capable of capturing the degree of safety of individual segments. It was found that the completeness of the system resulted in savings because of the economies obtained from trade-off optimization. It was observed that safety improvements were allocated at the beginning of the analysis in order to reduce the extent of issues, which translated into a systematic reduction of potential for improvement up to a point of near constant levels, which were hypothesized to relate to those unavoidable collisions from human error or vehicle failure.

Is there a potential relationship between prior h amstring strain injury and increased risk for future a nterior cruciate ligament injury?

Hamstring strain injuries (HSIs) are the most prevalent injury in a number of sports, and while anterior cruciate ligament (ACL) injuries are less common, they are far more severe and have long-term implications, such as an increased risk of developing osteoarthritis later in life. Given the high incidence and severity of these injuries, they are key targets of injury preventive programs in elite sport. Evidence has shown that a previous severe knee injury (including ACL injury) increases the risk of HSI; however, whether the functional deficits that occur after HSI result in an increased risk of ACL injury has yet to be considered. In this clinical commentary, we present evidence that suggests that the link between previous HSI and increased risk of ACL injury requires further investigation by drawing parallels between deficits in hamstring function after HSI and in women athletes, who are more prone to ACL injury than men athletes. Comparisons between the neuromuscular function of the male and female hamstring has shown that women display lower hamstring-to-quadriceps strength ratios during isokinetic knee flexion and extension, increased activation of the quadriceps compared with the hamstrings during a stop-jump landing task, a greater time required to reach maximal isokinetic hamstring torque, and lower integrated myoelectrical hamstring activity during a sidestep cutting maneuver. Somewhat similarly, in athletes with a history of HSI, the previously injured limb, compared with the uninjured limb, displays lower eccentric knee flexor strength, a lower hamstrings-to-quadriceps strength ratio, lower voluntary myoelectrical activity during maximal knee flexor eccentric contraction, a lower knee flexor eccentric rate of torque development, and lower voluntary myoelectrical activity during the initial portion of eccentric contraction. Given that the medial and lateral hamstrings have different actions at the knee joint in the coronal plane, which hamstring head is previously injured might also be expected to influence the likelihood of future ACL. Whether the deficits in function after HSI, as seen in laboratory-based studies, translate to deficits in hamstring function during typical injurious tasks for ACL injury has yet to be determined but should be a consideration for future work.

Estimating the mean critical gap

The estimation of the critical gap has been an issue since the 1970s, when gap acceptance was introduced to evaluate the capacity of unsignalized intersections. The critical gap is the shortest gap that a driver is assumed to accept. A driver’s critical gap cannot be measured directly and a number of techniques have been developed to estimate the mean critical gaps of a sample of drivers. This paper reviews the ability of the Maximum Likelihood technique and the Probability Equilibrium Method to predict the mean and standard deviation of the critical gap with a simulation of 100 drivers, repeated 100 times for each flow condition. The Maximum Likelihood method gave consistent and unbiased estimates of the mean critical gap. Whereas the probability equilibrium method had a significant bias that was dependent on the flow in the priority stream. Both methods were reasonably consistent, although the Maximum Likelihood Method was slightly better. If drivers are inconsistent, then again the Maximum Likelihood method is superior. A criticism levelled at the Maximum Likelihood method is that a distribution of the critical gap has to be assumed. It was shown that this does not significantly affect its ability to predict the mean and standard deviation of the critical gaps. Finally, the Maximum Likelihood method can predict reasonable estimates with observations for 25 to 30 drivers. A spreadsheet procedure for using the Maximum Likelihood method is provided in this paper. The PEM can be improved if the maximum rejected gap is used.

A recommendation system for predicting risks across multiple business process instances

This paper proposes a recommendation system that supports process participants in taking risk-informed decisions, with the goal of reducing risks that may arise during process execution. Risk reduction involves decreasing the likelihood and severity of a process fault from occurring. Given a business process exposed to risks, e.g. a financial process exposed to a risk of reputation loss, we enact this process and whenever a process participant needs to provide input to the process, e.g. by selecting the next task to execute or by filling out a form, we suggest to the participant the action to perform which minimizes the predicted process risk. Risks are predicted by traversing decision trees generated from the logs of past process executions, which consider process data, involved resources, task durations and other information elements like task frequencies. When applied in the context of multiple process instances running concurrently, a second technique is employed that uses integer linear programming to compute the optimal assignment of resources to tasks to be performed, in order to deal with the interplay between risks relative to different instances. The recommendation system has been implemented as a set of components on top of the YAWL BPM system and its effectiveness has been evaluated using a real-life scenario, in collaboration with risk analysts of a large insurance company. The results, based on a simulation of the real-life scenario and its comparison with the event data provided by the company, show that the process instances executed concurrently complete with significantly fewer faults and with lower fault severities, when the recommendations provided by our recommendation system are taken into account.

Supervised latent dirichlet allocation models for efficient activity representation

Local spatio-temporal features with a Bag-of-visual words model is a popular approach used in human action recognition. Bag-of-features methods suffer from several challenges such as extracting appropriate appearance and motion features from videos, converting extracted features appropriate for classification and designing a suitable classification framework. In this paper we address the problem of efficiently representing the extracted features for classification to improve the overall performance. We introduce two generative supervised topic models, maximum entropy discrimination LDA (MedLDA) and class- specific simplex LDA (css-LDA), to encode the raw features suitable for discriminative SVM based classification. Unsupervised LDA models disconnect topic discovery from the classification task, hence yield poor results compared to the baseline Bag-of-words framework. On the other hand supervised LDA techniques learn the topic structure by considering the class labels and improve the recognition accuracy significantly. MedLDA maximizes likelihood and within class margins using max-margin techniques and yields a sparse highly discriminative topic structure; while in css-LDA separate class specific topics are learned instead of common set of topics across the entire dataset. In our representation first topics are learned and then each video is represented as a topic proportion vector, i.e. it can be comparable to a histogram of topics. Finally SVM classification is done on the learned topic proportion vector. We demonstrate the efficiency of the above two representation techniques through the experiments carried out in two popular datasets. Experimental results demonstrate significantly improved performance compared to the baseline Bag-of-features framework which uses kmeans to construct histogram of words from the feature vectors.