Showing papers by "University of Toronto published in 1992"

Atoms, molecules, solids, and surfaces: Applications of the generalized gradient approximation for exchange and correlation.

Abstract: Generalized gradient approximations (GGA's) seek to improve upon the accuracy of the local-spin-density (LSD) approximation in electronic-structure calculations. Perdew and Wang have developed a GGA based on real-space cutoff of the spurious long-range components of the second-order gradient expansion for the exchange-correlation hole. We have found that this density functional performs well in numerical tests for a variety of systems: (1) Total energies of 30 atoms are highly accurate. (2) Ionization energies and electron affinities are improved in a statistical sense, although significant interconfigurational and interterm errors remain. (3) Accurate atomization energies are found for seven hydrocarbon molecules, with a rms error per bond of 0.1 eV, compared with 0.7 eV for the LSD approximation and 2.4 eV for the Hartree-Fock approximation. (4) For atoms and molecules, there is a cancellation of error between density functionals for exchange and correlation, which is most striking whenever the Hartree-Fock result is furthest from experiment. (5) The surprising LSD underestimation of the lattice constants of Li and Na by 3--4 % is corrected, and the magnetic ground state of solid Fe is restored. (6) The work function, surface energy (neglecting the long-range contribution), and curvature energy of a metallic surface are all slightly reduced in comparison with LSD. Taking account of the positive long-range contribution, we find surface and curvature energies in good agreement with experimental or exact values. Finally, a way is found to visualize and understand the nonlocality of exchange and correlation, its origins, and its physical effects.

Derivation of the SLEDAI. A disease activity index for lupus patients. The Committee on Prognosis Studies in SLE.

Abstract: Objective. To standardize outcome measures in systemic lupus erythematosus (SLE). Three indices were identified which could adequately describe outcome (disease activity, damage from disease, and health status); we describe here the development of the Disease Activity Index. Methods. Twenty-four variables were identified as important factors in a disease activity index. These were used to generate 574 patient profiles, which were rated on a disease activity scale of 0–10 by 14 rheumatologists. A second rating of 10 of the profiles yielded scores that were not significantly different from the first, indicating that experienced clinicians can reliably make global estimates of disease activity. Multiple regression models were used to estimate the relative importance of the 24 clinical variables in the physicians' global rating of disease activity. These were estimated on a ‘training set’ of 75% of physicians' ratings, and then validated on a ‘testing set,’ consisting of the remaining 25% of physicians' ratings. Results. The explanatory power of the models in the training set was high (R2 = 0.93). The models' regression coefficients for the organ systems were simplified for easier use in clinical practice. This generated a ‘weighted’ index of 9 organ systems for disease activity in SLE, the SLEDAI, as follows: 8 for central nervous system and vascular, 4 for renal and musculoskeletal, 2 for serosal, dermal, immunologic, and 1 for constitutional and hematologic. The maximum theoretical score is 105, but in practice, few patients have scores greater than 45. The SLEDAI predicted well the physicians' ratings in the testing set (Pearson's correlation coefficients = 0.64–0.79). Conclusion. The SLEDAI is a validated model of experienced clinicians' global assessments of disease activity in lupus. It represents the consensus of a group of experts in the field of lupus research.

Timeline Follow-Back A Technique for Assessing Self-Reported Alcohol Consumption

Abstract: Although the measurement of drinking is necessary for assessing and evaluating the treatment of alcohol problems, this key dependent variable has not always been reported in outcome studies.1, 2, 3 Today, the issue is not whether to measure drinking, but how to measure drinking. Concerns about how best to measure drinking patterns and problems date back to at least 1926, when Pearl stressed the importance of separating steady daily drinkers from occasional heavy drinkers.4

Revision of the Physical Activity Readiness Questionnaire (PAR-Q).

Abstract: The original Physical Activity Readiness Questionnaire (PAR-Q) offers a safe preliminary screening of candidates for exercise testing and prescription, but it screens out what seems an excessive proportion of apparently healthy older adults. To reduce unnecessary exclusions, an expert committee established by Fitness Canada has now revised the questionnaire wording. The present study compares responses to the original and the revised PAR-Q questionnaire in 399 men and women attending 40 accredited fitness testing centres across Canada. The number of subjects screened out by the revised test decreased significantly (p < .05), from 68 to 48 of the 399 subjects. The change reflects in part the inclusion of individuals who had made an erroneous positive response to the original question regarding high blood pressure. There is no simple gold standard to provide an objective evaluation of the sensitivity and specificity of either questionnaire format, but the revised wording has apparently had the intended effect of reducing positive responses, particularly to the question regarding an elevation of blood pressure.

A new method for solving hard satisfiability problems

Abstract: We introduce a greedy local search procedure called GSAT for solving propositional satisfiability problems. Our experiments show that this procedure can be used to solve hard, randomly generated problems that are an order of magnitude larger than those that can be handled by more traditional approaches such as the Davis-Putnam procedure or resolution. We also show that GSAT can solve structured satisfiability problems quickly. In particular, we solve encodings of graph coloring problems, N-queens, and Boolean induction. General application strategies and limitations of the approach are also discussed. GSAT is best viewed as a model-finding procedure. Its good performance suggests that it may be advantageous to reformulate reasoning tasks that have traditionally been viewed as theorem-proving problems as model-finding tasks.

The Use of β-Agonists and the Risk of Death and near Death from Asthma

Abstract: Background. Morbidity and mortality from asthma appear to be increasing, and it has been suggested that medications used to treat asthma are contributing to this trend. We investigated a possible association between death or near death from asthma and the regular use of β2-agonist bronchodilators. Methods. Using linked health insurance data bases from Saskatchewan, Canada, we conducted a matched case–control study of subjects drawn from a cohort of 12,301 patients for whom asthma medications had been prescribed between 1978 and 1987. We matched 129 case patients who had fatal or near-fatal asthma with 655 controls (who had received medications for asthma but had not had fatal or near-fatal events) with respect to region of residence, age, receipt of social assistance, and previous hospitalization for asthma. Results. The use of β-agonists administered by a metered-dose inhaler was associated with an increased risk of death from asthma (odds ratio, 2.6 per canister per month; 95 percent confidence...

Representing and using nonfunctional requirements: a process-oriented approach

Abstract: A comprehensive framework for representing and using nonfunctional requirements during the development process is proposed. The framework consists of five basic components which provide the representation of nonfunctional requirements in terms of interrelated goals. Such goals can be refined through refinement methods and can be evaluated in order to determine the degree to which a set of nonfunctional requirements is supported by a particular design. Evidence for the power of the framework is provided through the study of accuracy and performance requirements for information systems. >

Hard and easy distributions of SAT problems

Abstract: We report results from large-scale experiments in satisfiability testing. As has been observed by others, testing the satisfiability of random formulas often appears surprisingly easy. Here we show that by using the right distribution of instances, and appropriate parameter values, it is possible to generate random formulas that are hard, that is, for which satisfiability testing is quite difficult. Our results provide a benchmark for the evaluation of satisfiability-testing procedures.

Association of the Shc and Grb2/Sem5 SH2-containing proteins is implicated in activation of the Ras pathway by tyrosine kinases.

Abstract: THE mammalian shc gene encodes two overlapping, widely expressed proteins of 46 and 52K, with a carboxy-terminal SH2 domain that binds activated growth factor receptors, and a more amino-terminal glycine/proline-rich region1. These shc gene products (Shc) are transforming when overexpressed in fibroblasts1. Shc proteins become phosphorylated on tyrosine in cells stimulated with a variety of growth factors1, and in cells transformed by v-src (ref. 2), suggesting that they are tyrosine kinase targets that control a mitogenic signalling pathway. Here we report that tyrosine-phosphorylated She proteins form a specific complex with a non-phosphorylated 23K polypeptide encoded by the grb2/sem-5 gene3,4. The grb2/sem-5 gene product itself contains an SH2 domain, which mediates binding to Shc, and is implicated in activation of the Ras guanine nucleotide-binding protein by tyrosine kinases in both Caenorhabditis elegans and mammalian cells3,4. Consistent with a role in signalling through Ras, shc overexpression induced Ras-dependent neurite outgrowth in PC 12 cells. These results suggest that She tyrosine phosphorylation can couple tyrosine kinases to Grb2/Sem-5, through formation of a Shc-Grb2/Sem-5 complex, and thereby regulate the mammalian Ras signalling pathway.

A variable step size LMS algorithm

Abstract: A least-mean-square (LMS) adaptive filter with a variable step size is introduced. The step size increases or decreases as the mean-square error increases or decreases, allowing the adaptive filter to track changes in the system as well as produce a small steady state error. The convergence and steady-state behavior of the algorithm are analyzed. The results reduce to well-known results when specialized to the constant-step-size case. Simulation results are presented to support the analysis and to compare the performance of the algorithm with the usual LMS algorithm and another variable-step-size algorithm. They show that its performance compares favorably with these existing algorithms. >

Multiple dopamine D4 receptor variants in the human population

Abstract: THE dopamine D4 receptor structurally and pharmacologically resembles the dopamine D2 and D3 receptors1–5. Clozapine, an atypical antipsychotic that is relatively free of the adverse effects of drug-induced parkinsonism and tardive dyskinesia6,7, binds to the D4 receptor with an affinity 10 times higher than to the D2 and D3 receptors1. This may explain clozapine's atypical properties. Here we report the existence of at least three polymorphic variations in the coding sequence of the human D4 receptor. A 48-base-pair sequence in the putative third cytoplasmic loop of this receptor exists either as a direct-repeat sequence (D4.2), as a fourfold repeat (D4.4) or as a sevenfold repeat (D4.7). Two more variant alleles were detected in humans. Expression of the complementary DNA for the three cloned receptor variants showed different properties for the long form (D4.7) and the shorter forms (D4.2, D4.4) with respect to clozapine and spiperone binding. To our knowledge, this is the first report of a receptor in the catecholamine receptor family that displays polymorphic variation in the human population. Such variation among humans may underlie individual differences in susceptibility to neuropsychiatric disease and in responsiveness to antipsychotic medication.

Memory and working-with-memory: A component process model based on modules and central systems

Abstract: A neuropsychological model of memory is proposed that incorporates Fodor's (1983) idea of modules and central systems. The model has four essential components: (1) a non-frontal neocortical component that consists of perceptual (and perhaps interpretative semantic) modules that mediate performance on item-specific, implicit tests of memory, (2) a modular medial temporal/hippocampal component that mediates encoding, storage, and retrieval on explicit, episodic tests of memory that are associative/cue dependent, (3) a central system, frontal-lobe component that mediates performance on explicit tests that are strategic and on procedural tests that are rule-bound, and (4) a basal ganglia component that mediates performance on sensorimotor, procedural tests of memory. The usefulness of the modular/central system construct is explored and evidence from studies of normal, amnesic, agnosic, and demented people is provided to support the model.

Canadian National Breast Screening Study: 1. Breast cancer detection and death rates among women aged 40 to 49 years

Abstract: OBJECTIVES: To evaluate the efficacy of the combination of annual screening with mammography, physical examination of the breasts and the teaching of breast self-examination in reducing the rate of death from breast cancer among women aged 40 to 49 years on entry. DESIGN: Individually randomized controlled trial. SETTING: Fifteen urban centres in Canada with expertise in the diagnosis and treatment of breast cancer. PARTICIPANTS: Women with no history of breast cancer and no mammography in the previous 12 months were randomly assigned to undergo either annual mammography and physical examination (MP group) or usual care after an initial physical examination (UC group). The 50,430 women enrolled from January 1980 through March 1985 were followed for a mean of 8.5 years. DATA COLLECTION: Derived from the participants by initial and annual self-administered questionnaires, from the screening examinations, from the patients9 physicians, from the provincial cancer registries and by record linkage to the Canadian National Mortality Data Base. Expert panels evaluated histologic and death data. MAIN OUTCOME MEASURES: Rates of referral from screening, rates of detection of breast cancer from screening and from community care, nodal status, tumour size, and rates of death from all causes and from breast cancer. RESULTS: Over 90% of the women in each group attended the screening sessions or returned the annual questionnaires, or both, over years 2 to 5. The characteristics of the women in the two groups were similar. Compared with the Canadian population, the participants were more likely to be married, have fewer children, have more education, be in a professional occupation, smoke less and have been born in North America. The rate of screen-detected breast cancer on first examination was 3.89 per 1000 in the MP group and 2.46 per 1000 in the UC group; more node-positive tumours were found in the MP group than in the UC group. During years 2 through 5 the ratios of observed to expected cases of invasive breast cancer were 1.26 in the MP group and 1.02 in the UC group. Of the women with invasive breast cancer through to 7 years, 191 and 157 women in the MP and UC groups respectively had no node involvement, 55 and 43 had one to three nodes involved, 47 and 23 had four or more nodes involved, and 38 and 49 had an unknown nodal status. There were 38 deaths from breast cancer in the MP group and 28 in the UC group. The ratio of the proportions of death from breast cancer in the MP group compared with those in the UC group was 1.36 (95% confidence interval 0.84 to 2.21). The survival rates were similar in the two groups. The highest survival rate occurred among women whose cancer had been detected by mammography alone. CONCLUSION: The study was internally valid, and there was no evidence of randomization bias. Screening with yearly mammography and physical examination of the breasts detected considerably more node-negative, small tumours than usual care, but it had no impact on the rate of death from breast cancer up to 7 years9 follow-up from entry.

Does thoracic irradiation improve survival and local control in limited-stage small-cell carcinoma of the lung? A meta-analysis.

Abstract: PURPOSEOur main purpose was to determine whether the addition of thoracic radiation therapy to systemic chemotherapy improves 2-year survival, improves local (intrathoracic) tumor control, and affects treatment-related mortality in patients with limited-stage small-cell carcinoma of the lung.DESIGNEleven randomized trials addressing this issue were identified using a computerized literature search (Medline and Cancerline) and by polling senior investigators in the field. A meta-analysis was then performed and the results of the trials were analyzed in two ways, the odds ratio (OR) (Peto) method and the risk difference method (Dersimonian and Laird).RESULTSThe overall OR for benefit of thoracic radiation on 2-year survival (ie, the odds of surviving 2 years among patients allocated to radiation compared with the odds of surviving 2 years among patients allocated to control) is 1.53 (95% confidence interval [CI], 1.30 to 1.76; chi 2 = 12.76; P less than .001). The risk difference method showed that radiatio...

Simplifying neural networks by soft weight-sharing

Abstract: One way of simplifying neural networks so they generalize better is to add an extra term to the error function that will penalize complexity. Simple versions of this approach include penalizing the sum of the squares of the weights or penalizing the number of nonzero weights. We propose a more complicated penalty term in which the distribution of weight values is modeled as a mixture of multiple gaussians. A set of weights is simple if the weights have high probability density under the mixture model. This can be achieved by clustering the weights into subsets with the weights in each cluster having very similar values. Since we do not know the appropriate means or variances of the clusters in advance, we allow the parameters of the mixture model to adapt at the same time as the network learns. Simulations on two different problems demonstrate that this complexity term is more effective than previous complexity terms.

Feudal Reinforcement Learning

Abstract: One way to speed up reinforcement learning is to enable learning to happen simultaneously at multiple resolutions in space and time. This paper shows how to create a Q-learning managerial hierarchy in which high level managers learn how to set tasks to their submanagers who, in turn, learn how to satisfy them. Submanagers need not initially understand their managers' commands. They simply learn to maximise their reinforcement in the context of the current command. We illustrate the system using a simple maze task. As the system learns how to get around, satisfying commands at the multiple levels, it explores more efficiently than standard, flat, Q-learning and builds a more comprehensive map.

Cloning of cDNAs for Fanconi's anaemia by functional complementation.

Abstract: Fanconi's anaemia is a rare autosomal recessive disorder characterized by progressive pancytopaenia and a cellular hypersensitivity to DNA crosslinking agents. Four genetic complementation groups have been identified so far, and here we use a functional complementation method to clone complementary DNAs that correct the defect of group C cells. The cDNAs encode alternatively processed transcripts of a new gene, designated FACC, which is mutated in group C patients. The predicted FACC polypeptide does not contain any motifs common to other proteins and so represents a new gene involved in the cellular response to DNA damage.