Showing papers by "University of Toronto published in 1995"

A Stakeholder Framework for Analyzing and Evaluating Corporate Social Performance

Abstract: This article presents conclusions from a 10-year research program, the purpose of which has been to develop a framework and methodology, grounded in the reality of corporate behavior, for analyzing and evaluating corporate social performance. There are three principal sections: (a) a summary of the approaches, models, and methodologies used in conducting more than 70 field studies of corporate social performance from 1983-1993; (b) a discussion of the principal conclusions derived from the data that (1) corporations manage relationships with stakeholder groups rather than with society as a whole, (2) it is important to distinguish between social issues and stakeholder issues, and (3) it is necessary to identify the appropriate level of analysis in order to evaluate CSP; and (c) a discussion of propositions and areas for further research.

Cloning of a gene bearing missense mutations in early-onset familial Alzheimer's disease

Abstract: Some cases of Alzheimer's disease are inherited as an autosomal dominant trait. Genetic linkage studies have mapped a locus (AD3) associated with susceptibility to a very aggressive form of Alzheimer's disease to chromosome 14q24.3. We have defined a minimal cosegregating region containing the AD3 gene, and isolated at least 19 different transcripts encoded within this region. One of these transcripts (S182) corresponds to a novel gene whose product is predicted to contain multiple transmembrane domains and resembles an integral membrane protein. Five different missense mutations have been found that cosegregate with early-onset familial Alzheimer's disease. Because these changes occurred in conserved domains of this gene, and are not present in normal controls, they are likely to be causative of AD3.

Failure of blood-island formation and vasculogenesis in Flk-1-deficient mice.

Abstract: The receptor tyrosine kinase Flk-1 (ref. 1) is believed to play a pivotal role in endothelial development. Expression of the Flk-1 receptor is restricted to endothelial cells and their embryonic precursors, and is complementary to that of its ligand, vascular endothelial growth factor (VEGF), which is an endothelial-specific mitogen. Highest levels of flk-1 expression are observed during embryonic vasculogenesis and angiogenesis, and during pathological processes associated with neovascularization, such as tumour angiogenesis. Because flk-1 expression can be detected in presumptive mesodermal yolk-sac blood-island progenitors as early as 7.0 days postcoitum, Flk-1 may mark the putative common embryonic endothelial and haematopoietic precursor, the haemangioblast, and thus may also be involved in early haematopoiesis. Here we report the generation of mice deficient in Flk-1 by disruption of the gene using homologous recombination in embryonic stem (ES) cells. Embryos homozygous for this mutation die in utero between 8.5 and 9.5 days post-coitum, as a result of an early defect in the development of haematopoietic and endothelial cells. Yolk-sac blood islands were absent at 7.5 days, organized blood vessels could not be observed in the embryo or yolk sac at any stage, and haematopoietic progenitors were severely reduced. These results indicate that Flk-1 is essential for yolk-sac blood-island formation and vasculogenesis in the mouse embryo.

Changing Physician Performance: A Systematic Review of the Effect of Continuing Medical Education Strategies

Abstract: Objective. —To review the literature relating to the effectiveness of education strategies designed to change physician performance and health care outcomes. Data Sources. —We searched MEDLINE, ERIC, NTIS, the Research and Development Resource Base in Continuing Medical Education, and other relevant data sources from 1975 to 1994, using continuing medical education (CME) and related terms as keywords. We manually searched journals and the bibliographies of other review articles and called on the opinions of recognized experts. Study Selection. —We reviewed studies that met the following criteria: randomized controlled trials of education strategies or interventions that objectively assessed physician performance and/or health care outcomes. These intervention strategies included (alone and in combination) educational materials, formal CME activities, outreach visits such as academic detailing, opinion leaders, patient-mediated strategies, audit with feedback, and reminders. Studies were selected only if more than 50% of the subjects were either practicing physicians or medical residents. Data Extraction. —We extracted the specialty of the physicians targeted by the interventions and the clinical domain and setting of the trial. We also determined the details of the educational intervention, the extent to which needs or barriers to change had been ascertained prior to the intervention, and the main outcome measure(s). Data Synthesis. —We found 99 trials, containing 160 interventions, that met our criteria. Almost two thirds of the interventions (101 of 160) displayed an improvement in at least one major outcome measure: 70% demonstrated a change in physician performance, and 48% of interventions aimed at health care outcomes produced a positive change. Effective change strategies included reminders, patient-mediated interventions, outreach visits, opinion leaders, and multifaceted activities. Audit with feedback and educational materials were less effective, and formal CME conferences or activities, without enabling or practice-reinforcing strategies, had relatively little impact. Conclusion. —Widely used CME delivery methods such as conferences have little direct impact on improving professional practice. More effective methods such as systematic practice-based interventions and outreach visits are seldom used by CME providers. ( JAMA . 1995;274:700-705)

Lymphoproliferative Disorders with Early Lethality in Mice Deficient in Ctla-4

Abstract: The role of the cell-surface molecule CTLA-4 in the regulation of T cell activation has been controversial. Here, lymph nodes and spleens of CTLA-4-deficient mice accumulated T cell blasts with up-regulated activation markers. These blast cells also infiltrated liver, heart, lung, and pancreas tissue, and amounts of serum immunoglobulin were elevated. The mice invariably became moribund by 3 to 4 weeks of age. Although CTLA-4-deficient T cells proliferated spontaneously and strongly when stimulated through the T cell receptor, they were sensitive to cell death induced by cross-linking of the Fas receptor and by gamma irradiation. Thus, CTLA-4 acts as a negative regulator of T cell activation and is vital for the control of lymphocyte homeostasis.

Role of the Flt-1 receptor tyrosine kinase in regulating the assembly of vascular endothelium

Abstract: The vascular endothelial growth factor (VEGF) and its high-affinity binding receptors, the tyrosine kinases Flt-1 and Flk-1, are thought to be important for the development of embryonic vasculature. Here we report that Flt-1 is essential for the organization of embryonic vasculature, but is not essential for endothelial cell differentiation. Mouse embryos homozygous for a targeted mutation in the flt-1 locus, flt-1lcz, formed endothelial cells in both embryonic and extra-embryonic regions, but assembled these cells into abnormal vascular channels and died in utero at mid-somite stages. At earlier stages, the blood islands of flt-1lcz homozygotes were abnormal, with angioblasts in the interior as well as on the periphery. We suggest that the Flt-1 signalling pathway may regulate normal endothelial cell-cell or cell-matrix interactions during vascular development.

American college of rheumatology preliminary definition of improvement in rheumatoid arthritis

Abstract: Objective. Trials of rheumatoid arthritis (RA) treatments report the average response in multiple outcome measures for treated patients. It is more clinically relevant to test whether individual patients improve with treatment, and this identifies a single primary efficacy measure. Multiple definitions of improvement are currently in use in different trials. The goal of this study was to promulgate a single definition for use in RA trials. Methods. Using the American College of Rheumatology (ACR) core set of outcome measures for RA trials, we tested 40 different definitions of improvement, using a 3-step process. First, we performed a survey of rheumatologists, using actual patient cases from trials, to evaluate which definitions corresponded best to rheumatologists' impressions of improvement, eliminating most candidate definitions of improvement. Second, we tested 20 remaining definitions to determine which maximally discriminated effective treatment from placebo treatment and also minimized placebo response rates. With 8 candidate definitions of improvement remaining, we tested to see which were easiest to use and were best in accord with rheumatologists' impressions of improvement. Results. The following definition of improvement was selected: 20% improvement in tender and swollen joint counts and 20% improvement in 3 of the 5 remaining ACR core set measures: patient and physician global assessments, pain, disability, and an acutephase reactant. Additional validation of this definition was carried out in a comparative trial, and the results suggest that the definition is statistically powerful and does not identify a large percentage of placebo-treated patients as being improved. Conclusion. We present a definition of improvement which we hope will be used widely in RA trials.

Bayesian Density Estimation and Inference Using Mixtures

Abstract: We describe and illustrate Bayesian inference in models for density estimation using mixtures of Dirichlet processes. These models provide natural settings for density estimation and are exemplified by special cases where data are modeled as a sample from mixtures of normal distributions. Efficient simulation methods are used to approximate various prior, posterior, and predictive distributions. This allows for direct inference on a variety of practical issues, including problems of local versus global smoothing, uncertainty about density estimates, assessment of modality, and the inference on the numbers of components. Also, convergence results are established for a general class of normal mixture models.

Multiple organ dysfunction score : a reliable descriptor of a complex clinical outcome

Abstract: ObjectiveTo develop an objective scale to measure the severity of the multiple organ dysfunction syndrome as an outcome in critical illness.DesignSystematic literature review; prospective cohort study.SettingSurgical intensive care unit (ICU) of a tertiary-level teaching hospital.PatientsAll patient

Familial Alzheimer's disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer's disease type 3 gene

Abstract: We report the cloning of a novel gene (E5-1) encoded on chromosome 1 which has substantial nucleotide and amino-acid sequence similarity to the S182 gene on chromosome 14q24.3. Mutations, including three new missense mutations in the S182 gene, are associated with the AD3 subtype of early-onset familial Alzheimer's disease (AD). Both the E5-1 and the S182 proteins are predicted to be integral membrane proteins with seven membrane-spanning domains, and a large exposed loop between the sixth and seventh transmembrane domains. Analysis of the nucleotide sequence of the open reading frame (ORF) of the E5-1 gene led to the discovery of two missense substitutions at conserved amino-acid residues in affected members of pedigrees with a form of familial AD that has a later age of onset than the AD3 subtype (50-70 years versus 30-60 years for AD3). These observations imply that the E5-1 gene on chromosome 1 and the S182 gene on chromosome 14q24.3 are members of a family of genes (presenilins) with related functions, and indicates that mutations in conserved residues of E5-1 could also play a role in the genesis of AD. Our results also indicate that still other AD susceptibility genes exist.

Factorial Hidden Markov Models

Abstract: Hidden Markov models (HMMs) have proven to be one of the most widely used tools for learning probabilistic models of time series data. In an HMM, information about the past is conveyed through a single discrete variable—the hidden state. We discuss a generalization of HMMs in which this state is factored into multiple state variables and is therefore represented in a distributed manner. We describe an exact algorithm for inferring the posterior probabilities of the hidden state variables given the observations, and relate it to the forward–backward algorithm for HMMs and to algorithms for more general graphical models. Due to the combinatorial nature of the hidden state representation, this exact algorithm is intractable. As in other intractable systems, approximate inference can be carried out using Gibbs sampling or variational methods. Within the variational framework, we present a structured approximation in which the the state variables are decoupled, yielding a tractable algorithm for learning the parameters of the model. Empirical comparisons suggest that these approximations are efficient and provide accurate alternatives to the exact methods. Finally, we use the structured approximation to model Bach‘s chorales and show that factorial HMMs can capture statistical structure in this data set which an unconstrained HMM cannot.

W/kit gene required for interstitial cells of Cajal and for intestinal pacemaker activity

Abstract: The pacemaker activity in the mammalian gut is responsible for generating anally propagating phasic contractions. The cellular basis for this intrinsic activity is unknown. The smooth muscle cells of the external muscle layers and the innervated cellular network of interstitial cells of Cajal, which is closely associated with the external muscle layers of the mammalian gut, have both been proposed to stimulate pacemaker activity. The interstitial cells of Cajal were identified in the last century but their developmental origin and function have remained unclear. Here we show that the interstitial cells of Cajal express the Kit receptor tyrosine kinase. Furthermore, mice with mutations in the dominant white spotting (W) locus, which have cellular defects in haematopoiesis, melanogenesis and gametogenesis as a result of mutations in the Kit gene, also lack the network of interstitial cells of Cajal associated with Auerbach's nerve plexus and intestinal pacemaker activity.

The epidemiology of social stress

Abstract: We examine the social distribution of exposure to stress to test the hypothesis that differences in stress exposure are one factor in sociodemographic variations in mental health. We make a more comprehensive effort to estimate stress exposure than has been typical, and present data that challenge the prevailing view that differences in exposure to stress are of only minimal significance for understanding variations in mental health. We report several findings, principal among which are: Differences in exposure to stress account for substantially more variability in depressive symptoms and major depressive disorder than previous reports have suggested; the distributions of stress exposure across sex, age, marital status, and occupational status precisely correspond to the distributions of depressive symptoms and major depressive disorder across the same factors; and differences in exposure to stress alone account for between 23 and 50 percent of observed differences in mental health by sex, marital status, and occupation. These findings contrast with the prevailing view that differences in vulnerability to stress across social statuses account for social status variations in mental health.

Cardiac malformation in neonatal mice lacking connexin43

Abstract: Gap junctions are made up of connexin proteins, which comprise a multigene family in mammals. Targeted mutagenesis of connexin43 (Cx43), one of the most prevalent connexin proteins, showed that its absence was compatible with survival of mouse embryos to term, even though mutant cell lines showed reduced dye coupling in vitro. However, mutant embryos died at birth, as a result of a failure in pulmonary gas exchange caused by a swelling and blockage of the right ventricular outflow tract from the heart. This finding suggests that Cx43 plays an essential role in heart development but that there is functional compensation among connexins in other parts of the developing fetus.

Gaussian Processes for Regression

Abstract: The Bayesian analysis of neural networks is difficult because a simple prior over weights implies a complex prior distribution over functions. In this paper we investigate the use of Gaussian process priors over functions, which permit the predictive Bayesian analysis for fixed values of hyperparameters to be carried out exactly using matrix operations. Two methods, using optimization and averaging (via Hybrid Monte Carlo) over hyperparameters have been tested on a number of challenging problems and have produced excellent results.

The geometry of graphs and some of its algorithmic applications

Abstract: In this paper we explore some implications of viewing graphs asgeometric objects. This approach offers a new perspective on a number of graph-theoretic and algorithmic problems. There are several ways to model graphs geometrically and our main concern here is with geometric representations that respect themetric of the (possibly weighted) graph. Given a graphG we map its vertices to a normed space in an attempt to (i) keep down the dimension of the host space, and (ii) guarantee a smalldistortion, i.e., make sure that distances between vertices inG closely match the distances between their geometric images. In this paper we develop efficient algorithms for embedding graphs low-dimensionally with a small distortion. Further algorithmic applications include: Given faithful low-dimensional representations of statistical data, it is possible to obtain meaningful and efficientclustering. This is one of the most basic tasks in pattern-recognition. For the (mostly heuristic) methods used in the practice of pattern-recognition, see [20], especially chapter 6. Our studies of multicommodity flows also imply that every embedding of (the metric of) ann-vertex, constant-degree expander into a Euclidean space (of any dimension) has distortion Ω(logn). This result is tight, and closes a gap left open by Bourgain [12].

Quantitative Classification of Mammographic Densities and Breast Cancer Risk: Results From the Canadian National Breast Screening Study

Abstract: BACKGROUND The radiographic appearance of the female breast varies from woman to woman depending on the relative amounts of fat and connective and epithelial tissues present. Variations in the mammographic density of breast tissue are referred to as the parenchymal pattern of the breast. Fat is radiologically translucent or clear (darker appearance), and both connective and epithelial tissues are radiologically dense (lighter appearance). Previous studies have generally supported an association between parenchymal patterns and breast cancer risk (greater risk with increasing densities), but there has been considerable heterogeneity in risk estimates reported. PURPOSE Our objective was to determine the level of breast cancer risk associated with varying mammographic densities by quantitatively classifying breast density with conventional radiological methods and novel computer-assisted methods. METHODS From the medical records of a cohort of 45,000 women assigned to mammography in the Canadian National Breast Cancer Screening Study (NBSS), a multicenter, randomized trial, mammograms from 354 case subjects and 354 control subjects were identified. Case subjects were selected from those women in whom histologically verified invasive breast cancer had developed 12 months or more after entering the trial. Control subjects were selected from those of similar age who, after a similar period of observation, had not developed breast cancer. The mammogram taken at the beginning of the NBSS was the image used for measurements. Mammograms were classified into six categories of density, either by radiologists or by computer-assisted measurements. All radiological classification and computer-assisted measurements were made using one craniocaudal view from the breast contralateral to the cancer site in case subjects and the corresponding breast of control subjects. All P values represent two-sided tests of statistical significance. RESULTS For all subjects, there was a 43% increase in the relative risk (RR) between the lower and the next higher category of density, as determined by radiologists, and there was a 32% increase as determined by the computer-assisted method. For all subjects, the RR in the most extensive category relative to the least was 6.05 (95% confidence interval [CI] = 2.82-12.97) for radiologists and 4.04 (95% CI = 2.12-7.69) for computer-assisted methods. Statistically significant increases in breast cancer risk associated with increasing mammographic density were found by both radiologists and computer-assisted methods for women in the age category 40-49 years (P = .005 for radiologists and P = .003 for computer-assisted measurements) and the age category 50-59 years (P = .002 for radiologists and P = .001 for computer-assisted measurements). CONCLUSION These results show that increases in the level of breast tissue density as assessed by mammography are associated with increases in risk for breast cancer.

The "Wake-Sleep" Algorithm for Unsupervised Neural Networks

Abstract: An unsupervised learning algorithm for a multilayer network of stochastic neurons is described. Bottom-up "recognition" connections convert the input into representations in successive hidden layers, and top-down "generative" connections reconstruct the representation in one layer from the representation in the layer above. In the "wake" phase, neurons are driven by recognition connections, and generative connections are adapted to increase the probability that they would reconstruct the correct activity vector in the layer below. In the "sleep" phase, neurons are driven by generative connections, and recognition connections are adapted to increase the probability that they would produce the correct activity vector in the layer above.

Bricks: laying the foundations for graspable user interfaces

Abstract: We introduce the concept of Graspable User Interfaces that allow direct control of electronic or virtual objects through physical handles for control. These physical artifacts, which we call "bricks," are essentially new input devices that can be tightly coupled or “attached” to virtual objects for manipulation or for expressing action (e.g., to set parameters or for initiating processes). Our bricks operate on top of a large horizontal display surface known as the "ActiveDesk." We present four stages in the development of Graspable UIs: (1) a series of exploratory studies on hand gestures and grasping; (2) interaction simulations using mock-ups and rapid prototyping tools; (3) a working prototype and sample application called GraspDraw; and (4) the initial integrating of the Graspable UI concepts into a commercial application. Finally, we conclude by presenting a design space for Bricks which lay the foundation for further exploring and developing Graspable User Interfaces.

The Molecular Basis of Turcot's Syndrome

Abstract: Background Turcot's syndrome is characterized clinically by the concurrence of a primary brain tumor and multiple colorectal adenomas. We attempted to define the syndrome at the molecular level. Methods Fourteen families with Turcot's syndrome identified in two registries and the family originally described by Turcot and colleagues were studied. Germ-line mutations in the adenomatous polyposis coli (APC) gene characteristic of familial adenomatous polyposis were evaluated, as well as DNA replication errors and germ-line mutations in nucleotide mismatch-repair genes characteristic of hereditary nonpolyposis colorectal cancer. In addition, a formal risk analysis for brain tumors in familial adenomatous polyposis was performed with a registry data base. Results Genetic abnormalities were identified in 13 of the 14 registry families. Germ-line APC mutations were detected in 10. The predominant brain tumor in these 10 families was medulloblastoma (11 of 14 patients, or 79 percent), and the relative risk of cer...

Observation of top quark production in p̄p collisions with the collider detector at fermilab

Abstract: We establish the existence of the top quark using a $67{\mathrm{pb}}^{\ensuremath{-}1}$ data sample of $\overline{p}p$ collisions at $\sqrt{s}\phantom{\rule{0ex}{0ex}}=\phantom{\rule{0ex}{0ex}}1.8\mathrm{TeV}$ collected with the Collider Detector at Fermilab (CDF). Employing techniques similar to those we previously published, we observe a signal consistent with $t\overline{t}$ decay to $\mathrm{WWb}\overline{b}$, but inconsistent with the background prediction by $4.8\ensuremath{\sigma}$. Additional evidence for the top quark is provided by a peak in the reconstructed mass distribution. We measure the top quark mass to be $176\ifmmode\pm\else\textpm\fi{}8(\mathrm{stat})\ifmmode\pm\else\textpm\fi{}10(\mathrm{syst})\mathrm{GeV}{/c}^{2}$, and the $t\overline{t}$ production cross section to be ${6.8}_{\ensuremath{-}2.4}^{+3.6}\mathrm{pb}$.

The Helmholtz machine

Abstract: Discovering the structure inherent in a set of patterns is a fundamental aim of statistical inference or learning. One fruitful approach is to build a parameterized stochastic generative model, independent draws from which are likely to produce the patterns. For all but the simplest generative models, each pattern can be generated in exponentially many ways. It is thus intractable to adjust the parameters to maximize the probability of the observed patterns. We describe a way of finessing this combinatorial explosion by maximizing an easily computed lower bound on the probability of the observations. Our method can be viewed as a form of hierarchical self-supervised learning that may relate to the function of bottom-up and top-down cortical processing pathways.

Realistic modeling for facial animation

Abstract: A major unsolved problem in computer graphics is the construction and animation of realistic human facial models. Traditionally, facial models have been built painstakingly by manual digitization and animated by ad hoc parametrically controlled facial mesh deformations or kinematic approximation of muscle actions. Fortunately, animators are now able to digitize facial geometries through the use of scanning range sensors and animate them through the dynamic simulation of facial tissues and muscles. However, these techniques require considerable user input to construct facial models of individuals suitable for animation. In this paper, we present a methodology for automating this challenging task. Starting with a structured facial mesh, we develop algorithms that automatically construct functional models of the heads of human subjects from laser-scanned range and reflectance data. These algorithms automatically insert contractile muscles at anatomically correct positions within a dynamic skin model and root them in an estimated skull structure with a hinged jaw. They also synthesize functional eyes, eyelids, teeth, and a neck and fit them to the final model. The constructed face may be animated via muscle actuations. In this way, we create the most authentic and functional facial models of individuals available to date and demonstrate their use in facial animation.

Lexical chains as representations of context for the detection and correction of malapropisms

Abstract: Natural language utterances are, in general, highlyambiguous, and a unique interpretationcan usuallybe determined only by taking into account the constraining influence of the context in which theutterance occurred. Much of the research in natural language understanding in the last twenty yearscan be thought of as attempts to characterize and represent context and then derive interpretationsthatfit best with that context. Typically, this research was heavy with AI, taking context to be nothing lessthan a complete conceptual understanding of the preceding utterances. This was reasonable, as suchan understanding of a text was often the main task anyway. However, there are many text-processingtasksthatrequireonlya partialunderstandingofthetext, andhencea ‘lighter’representationofcontextis sufficient. In this paper, we examine the idea oflexical chains as such a representation. We showhow they can be constructed by means of WordNet, and how they can be applied in one particularlinguistic task: the detection and correction of malapropisms.A malapropism is the confounding of an intended word with another word of similar sound orsimilar spelling that has a quite different and malapropos meaning, e.g., an ingenuous [for ingenious]machine forpeelingoranges. In thisexample, there isaone-letterdifference betweenthe malapropismand the correct word. Ignorance, or a simple typing mistake, might cause such errors. However, sinceingenuous is a correctly spelled word, traditional spelling checkers cannot detect this kind of mistake.In section 4, we will propose an algorithm for detecting and correcting malapropisms that is based onthe construction of lexical chains.