Biomedical subjects

Michael Biehl

Publications and source records attributed to Michael Biehl.

6 recordsLinked to original sources

PubMed2006

Performance analysis of LVQ algorithms: a statistical physics approach.

Learning vector quantization (LVQ) constitutes a powerful and intuitive method for adaptive nearest prototype classification. However, original LVQ has been introduced based on heuristics and numerous modifications exist to achieve better convergence and stability. Recently, a mathematical foundation by means of a cost function has been proposed which, as a limiting case, yields a learning rule similar to classical LVQ2.1. It also motivates a modification which shows better stability. However, the exact dynamics as well as the generalization ability of many LVQ algorithms have not been thoroughly investigated so far. Using concepts from statistical physics and the theory of on-line learning, we present a mathematical framework to analyse the performance of different LVQ algorithms in a typical scenario in terms of their dynamics, sensitivity to initial conditions, and generalization ability. Significant differences in the algorithmic stability and generalization ability can be found already for slightly different variants of LVQ. We study five LVQ algorithms in detail: Kohonen's original LVQ1, unsupervised vector quantization (VQ), a mixture of VQ and LVQ, LVQ2.1, and a variant of LVQ which is based on a cost function. Surprisingly, basic LVQ1 shows very good performance in terms of stability, asymptotic generalization ability, and robustness to initializations and model parameters which, in many cases, is superior to recent alternative proposals.

Algorithms↗

PubMed2005

Efficient training of multilayer perceptrons using principal component analysis.

A training algorithm for multilayer perceptrons is discussed and studied in detail, which relates to the technique of principal component analysis. The latter is performed with respect to a correlation matrix computed from the example inputs and their target outputs. Typical properties of the training procedure are investigated by means of a statistical physics analysis in models of learning regression and classification tasks. We demonstrate that the procedure requires by far fewer examples for good generalization than traditional online training. For networks with a large number of hidden units we derive the training prescription which achieves, within our model, the optimal generalization behavior.

Journal Article↗

PubMed2004

Trends in private and public health insurance for adolescents.

CONTEXT: Previous studies (1984-1995) of adolescent health insurance have shown little change in the proportion with coverage. Federally mandated expansions in Medicaid were offset by declines in private coverage. Further expansions of Medicaid and implementation of the State Children's Health Insurance Program (SCHIP) have opened new avenues for increasing coverage rates. OBJECTIVES: To assess the current health insurance status of adolescents, the demographic and socioeconomic correlates of insurance coverage, and document recent changes in public and private coverage rates. DESIGN, SETTING, AND PARTICIPANTS: We analyzed data on 12 995 adolescents aged 10 to 18 years, who had been included in the 2002 National Health Interview Survey. We conducted multivariate analyses to assess the independent association of age, sex, race, poverty status, family structure, family size, and region on the likelihood of having insurance coverage. Results are compared with previously published findings on adolescent health insurance coverage spanning 1984 to 1995. MAIN OUTCOME MEASURE: Insurance coverage for adolescents. RESULTS: An estimated 12.2% of adolescents were uninsured in 2002, which is a decrease from 14.1% in 1995 (P<.003). The decrease occurred entirely because of an expansion of public coverage and is concentrated among children in poor (<100% of the federal poverty level) and near-poor (100%-199% of the federal poverty level) families. A substantial decrease in the differences between poor and higher-income groups occurred between 1995 and 2002 due to gains in coverage for adolescents in poor and near-poor families and losses in coverage among those in middle- and upper-income families (> or =200% of the federal poverty level). Specifically, the proportion of adolescents in poor families without coverage declined from 27.4% in 1995 to 19.7% in 2002 (P<.001). The proportion of adolescents in near-poor families without coverage declined from 24.8% in 1995 to 19.2% in 2002 (P<.002). In contrast, the proportion of adolescents in middle- and higher-income families without insurance increased from 4.1% in 1995 to 6.3% in 2002 because availability of insurance through the private market declined (P<.001). CONCLUSIONS: A modest but significant reduction in the percentage of adolescents without insurance has occurred since 1995, largely as a result of expansions in public coverage. An even larger reduction in the proportion of adolescents without coverage would have occurred, if not for a reduction in private coverage for adolescents in middle- and higher-income families.

Adolescent↗

PubMed2004

Perceived risks and benefits of smoking: differences among adolescents with different smoking experiences and intentions.

BACKGROUND: Explanations of adolescent smoking often make reference to adolescents' beliefs that they are invulnerable to harm. However, empirical examination of whether adolescents do acknowledge risks. Further, few studies have considered perceived benefits in adolescents' behavioral decisions. This study examined perceived smoking-related physical and social risks and benefits between adolescents who have vs. have not smoked and do vs. do not intend to smoke. METHODS: Three hundred and ninety-five students (mean age = 14.0) completed a survey concerning their smoking experiences, intentions, and perceived risks and benefits of smoking. RESULTS: Adolescent smokers and those who intend to smoke estimated their chance of experiencing a smoking-related negative outcome as less likely than did nonsmokers and non-intenders. Smokers and intenders also reported the chance of addiction as less likely than did others. In contrast, adolescent smokers and intenders perceived the chance of experiencing a smoking-related benefit as more likely than did nonsmokers and non-intenders. CONCLUSIONS: The data suggest that rather than solely focusing on health risks as a way to deter adolescent smoking, the role of perceived social risks and benefits in adolescents' smoking may be an additional critical focus for intervention. In addition, efforts should be made to increase adolescents' awareness of the addictive nature of cigarettes.

Adolescent↗

PubMed2003

Socioeconomic patterning of smoking, sedentary lifestyle, and overweight status among adolescents in Norway and the United States.

Increased efforts are needed to promote adolescent health in all areas investigated in this study. Differences both in prevalence and socioeconomic patterning between Norway and the United States indicate that the major challenges differ between the two countries. The results also emphasize the need to undertake joint efforts to use measurements that make national data sets comparable across countries.

Adolescent↗

PubMed2003

Receipt of psychological or emotional counseling by suicidal adolescents.

OBJECTIVE: This study examined utilization of psychological or emotional counseling by suicidal adolescents to answer questions about the extent to which health services can contribute to the prevention of adolescent suicide. METHOD: The study used data from Wave 1 of the National Longitudinal Study of Adolescent Health, which involved a household-based interview with a nationally representative sample of 15 483 adolescents from grades 7 to 12. Of these, 2482 adolescents were classified as suicidal, as indicated by an affirmative response to the question "During the past 12 months, did you ever seriously think about committing suicide?" For this group, the study asked the following questions: 1) What proportion receives psychological or emotional counseling? 2) What are the sources of this counseling? 3) What factors are associated with receipt of such counseling? RESULTS: Less than one third (28%) of suicidal adolescents received psychological or emotional counseling. The most common sources of care were private doctors' offices (37%) and schools (34%). Factors associated with receipt of counseling in the past 12 months included age, race, degree of suicidality, depression status, and having had a physical examination during the same period. CONCLUSIONS: Only one third of those who report suicidal ideation and behavior receive psychological or emotional counseling. Although not all of these young people may identify a need for counseling, this finding still suggests that many of those at risk of harming themselves do not receive professional help. However, on the positive side, those who do use counseling services tend to do so on the basis of their being in the greatest need, rather than their parents' capacity to pay for services. Counseling services have an important role to play in suicide prevention, and a variety of sources of care need to be available. Although counseling services are vital, a range of other strategies is necessary to reduce the youth suicide rate.

Adolescent↗