University of Saskatchewan

About: University of Saskatchewan is a education organization based out in Saskatoon, Saskatchewan, Canada. It is known for research contribution in the topics: Population & Health care. The organization has 25021 authors who have published 52579 publications receiving 1483049 citations. The organization is also known as: USask.

Is the Canadian childhood obesity epidemic related to physical inactivity

Abstract: OBJECTIVE: This study examined the relation among children's physical activity, sedentary behaviours, and body mass index (BMI), while controlling for sex, family structure, and socioeconomic status. DESIGN: Epidemiological study examining the relations among physical activity participation, sedentary behaviour (video game use and television (TV)/video watching), and BMI on a nationally representative sample of Canadian children. SUBJECTS: A representative sample of Canadian children aged 7–11 (N=7216) from the 1994 National Longitudinal Survey of Children and Youth was used in the analysis. MEASUREMENTS: Physical activity and sport participation, sedentary behaviour (video game use and TV/video watching), and BMI measured by parental report. RESULTS: Both organized and unorganized sport and physical activity are negatively associated with being overweight (10–24% reduced risk) or obese (23–43% reduced risk), while TV watching and video game use are risk factors for being overweight (17–44% increased risk) or obese (10–61% increased risk). Physical activity and sedentary behaviour partially account for the association of high socioeconomic status and two-parent family structure with the likelihood of being overweight or obese. CONCLUSION: This study provides evidence supporting the link between physical inactivity and obesity of Canadian children.

WILDS: A Benchmark of in-the-Wild Distribution Shifts

Abstract: Distribution shifts -- where the training distribution differs from the test distribution -- can substantially degrade the accuracy of machine learning (ML) systems deployed in the wild. Despite their ubiquity, these real-world distribution shifts are under-represented in the datasets widely used in the ML community today. To address this gap, we present WILDS, a curated collection of 8 benchmark datasets that reflect a diverse range of distribution shifts which naturally arise in real-world applications, such as shifts across hospitals for tumor identification; across camera traps for wildlife monitoring; and across time and location in satellite imaging and poverty mapping. On each dataset, we show that standard training results in substantially lower out-of-distribution than in-distribution performance, and that this gap remains even with models trained by existing methods for handling distribution shifts. This underscores the need for new training methods that produce models which are more robust to the types of distribution shifts that arise in practice. To facilitate method development, we provide an open-source package that automates dataset loading, contains default model architectures and hyperparameters, and standardizes evaluations. Code and leaderboards are available at this https URL.

A review of rapid prototyping technologies and systems

Abstract: Rapid Prototyping and Manufacturing (RP&M) technologies have emerged for quickly creating 3D products directly from computer-aided design systems. These technologies significantly improve the present prototyping practices in industry. This paper reviews the main technologies and applications of RP&M. The principles and the features of those RP&M technologies are presented. Some existing problems and research issues on these new technologies are introduced. We also include two current research and application examples in using rapid prototyping for further illustration.

Alarm signals in fishes

Abstract: the evolutionary questions surrounding alarm signalling remain unresolved, but we should now have a better understanding of the elements that must be considered in the balance sheet. The amplification that may occur in alarm signalling may be a key to understanding its evolution. The benefit to receivers will often go far beyond the response of a few nearby schoolmates over a few minutes, the response that has traditionally been measured. Distant fish may receive the signal by secondary transmission, and individuals that are not even present at the time may learn about predator stimuli through cultural transmission. These effects, such as learned response to predator odour or avoidance of an area, may persist for days or much longer, and work on invertebrates implies that there may be the potential for changes in morphology and life history. Thus one signal, such as release of alarm pheromone, could alter predation risk for many individuals over long periods of time. Anything that increases the total benefit to receivers will affect the evolutionary balance sheet. Increase in number of benefits and beneficiaries of a signal will increase the likelihood that the sender will receive adequate kin-selection benefits to drive the evolution of alarm displays. Likewise, to have many individuals avoiding predation would increase the post-signal benefits, such as reduced predation in the region (Trivers, 1971), to senders that survived. Similarly, anything that decreases the cost or increases the direct benefit to the sender will favour alarm signalling. Alarm signals that do double duty as predator deterrents, or aposematic displays, and distress signals that call in mobbers or secondary predators will have lower net cost than signals that only exist to warn others. It may be common for the sender's display to evolve primarily in response to direct benefits to the sender while the reaction of conspecific receivers is selected by their survival. Selection on receivers that reduces their response threshold will make signalling cheaper for the sender. The variety of life histories and biological adaptations in the fishes, combined with the potential of several different, independently evolved alarm signals should provide many avenues of approach and potential research subjects for examining the evolution of these systems. There have been many interesting effects reported in other groups of animals that may occur in fishes and which would extend both the biological interest of these systems and their generality. I have mentioned the morphological and life cycle responses found in invertebrates. Birds show deceitful alarm signalling (Munn, 1986; Moller, 1988), in which senders give false alarm calls to distract receivers from food or other resources. Audience effects occur in domestic chickens (Marler, 1986); they are more likely to give an alarm call if with a companion than when alone. Vervet monkeys assess the reliability of individual signallers and tend to ignore signals from untrustworthy individuals (Cheney and Seyfarth, 1988). Birds can acquire and transmit the identity of individual predators that prey on their species, in contrast to other individuals of the same predatory species that do not (Conover, 1987). The many effects of alarm signalling that have been documented or proposed in fishes or other organisms indicate that this phenomenon must be taken into account in any examination of foraging tactics or predator-prey interaction or any of the several areas of decision making that could be influenced by information on predation risk. Alarm signalling is probably much more widespread than was previously thought. Alarm pheromones are not just an obscure feature of the ostariophysans, although that group alone includes over 6000 species, but also occur in various forms in darters (150 species), gobies (2000 species), sculpins (300 species) and perhaps others. Distress sounds occur in over 24 families (Myrberg, 1981). Alarm calls occur in at least some holocentrids (60 species) and possibly in cods (only 55 species, but some economic value). Visual alarm signals have been reported in gobies (2000 species) and bioluminescent displays in a batrachoidid (65 species). Yet only a small fraction of fishes have been carefully examined for alarm or distress signalling. If we multiply the range of effects by the number of potential species involved, we have a subject area of some general importance in understanding the interactions between prey and predators. The prime requirement in this field, as in so many others, is for carefully designed studies, particularly in the wild, that take account of the whole suite of possible effects that may occur in alarm signalling. These studies should try to include all the participants in the system, including the predator(s), the signaller, and the various classes of receivers. They should also consider both the ultimate and proximate factors at work in each system. Very often proximate mechanisms can tell us important things about the ultimate factors that may be possible.