WWW.DISSERTATION.XLIBX.INFO
FREE ELECTRONIC LIBRARY - Dissertations, online materials
 
<< HOME
CONTACTS



Pages:   || 2 |

«13. Rodriguez FA. Cardiorespiratory and metabolic field testing in swimming and water polo: from physiological concepts to practical methods. In: ...»

-- [ Page 1 ] --

13. Rodriguez FA. Cardiorespiratory and metabolic field testing in swimming and water polo: from physiological

concepts to practical methods. In: Keskinen Kl, Komi PV, Hollander AP, editors. Biomechanics and Medicine

in Swimming VIII. University of Jyvaskyla, Finland: Gummerus Printing; 1999. p. 219-26.

14. Schuller T, Rodrfguez FA, lglesias X, Barrero A, Chaverri D, Hoffmann U. A new model for estimating peak

oxygen uptake based on post-exercise measurements and heart rate kinetics in swimming. 18th annual Congress of European College of Sport Science; 2013; Barcelona: INEFC, ECSS; 2013.

15. Drescher U, Essfeld D, Hoffmann U. Mode ling Muscular V'02-Kinetics on the Basis of Respiratory V'02 and Cardiac Output Measurements. In: Kurkusuz F, editor. 15th Annual Congress of the European College of Sport Science Antalya: Middle East Technical University Faculty of Education Physical Education & Sport Department; 2010.

16. Christie JL, SheldahllM, Tristani FE, Wann LS, Sagar KB, Levandoski SG, et al. Cardiovascular regulation during head-out water immersion exercise. J Appl Physiol. 1990; 69(2): 657-64.

17. Keskinen KL, Rodriguez FA, Keskinen OP. Respiratory snorkel and valve system for breath-by-breath gas analysis in swimming. Scand J Med Sci Sports. 2003; 13(5): 322-9.

18. Rodriguez FA, Keskinen KL, Kusch M, Hoffmann U. Validity of a swimming snorkel for metabolic testing. lnt J Sports Med. 2008; 29(2): 120-8.

19. Roberts AD, Morton AR. Total and alactic oxygen debts after supramaximal work. Eur J Appl Physiol Occup Physiol. 1978; 38(4): 281-9.

Comparing methods for summarising a training load in prediction models of swimming performance 1 23 4 12 Charlotte Scordia ', Marta Avalos ' ', Philippe Hellard 's 3 1 2 Univ. Bordeaux, !SPED, France, 1NSERM U897-Epidemiologie-Biostatistique, France, 1NRIA-SISTM, France, 4 5 Research Department, French Swimming Federation, France, 1RMES, lnstitut de Recherche bioMedicale et d'Epidemiologie du Sport, lnsep, France Keywords: elite swimming, longitudinal data, performance, statistical machine learning, training load intensities, training measurement Abstract Introduction. Training quantifications are valuable for monitoring and prescribing elite swimmers' training and are indispensable in mathematical models that attempt to accurately predict performance. Modelling the association of training with performance raises an important issue: how should we account for volumes at different training intensities? Mujika et al. (1996) constructed a training load by adding weighted (by a priori constants representing energetic intensities) volumes from each intensity. Avalos et al. (2003) computed a training load as the sum of normalised training intensities. Here we compared the predictive accuracy of these methods to others based on: a/ alternative normalisations, b/ summary scores derived from data, and c/ machine learning techniques, with recognised predictive qualities, such as PLS.

Methods. Training volumes at eight intensity levels (in kilometres and minutes per week, for in-water and dry-land workouts, respectively) and performances in competition of 138 professional French swimmers were collected during 20 seasons. Training intensities were determined using measurements of blood lactate concentrations. We assumed that swimmers may react differently to the same training and over time, thus we used mixed-effects models adjusted for sex, age, swimming distance and event specialty. The comparison criterion was the cross-validated prediction error.

Results. Summary scores for three training loads (low-intensity/high-intensity/dry-land workouts) with data derived weights showed the best results (mean cross-validated prediction error± SD were

0.60±0.89, 0.50±0.62 and 0.10±0.19 for sprint, mid- and long-distances, respectively). However, crossvalidated prediction errors were close relative to their variances, which were high.

511 BMS2014-PROCEEDINGS Conclusions. The use of complex machine learning techniques did not lead to more accuracy in predicting performance. Although data derived scores showed the lowest prediction error, the statistical variability was too high for being conclusive. A possible explanation is that the lactate sensitivity to extraneous factors (mode of exercise, technique quality of training, diet or sleep quality prior to test) and the subject-specific variations in lactate thresholds introduce not negligible measurement error. As practical recommendation, we suggest completing lactate measurements with athlete/coach questionnaires to better assess the physiological stress associated with the training load. Also, errors-in-variables models might be more appropriated.

Introduction Achieving optimal performance in swimming implies developing and improving metabolic processes (the phosphagen, glycolytic and aerobic systems), muscular strength and endurance qualities, and technical skills through training. Training quantifications (the frequency, duration, intensity and type of exercise) provide valuable information to coaches and swimmers for monitoring and prescribing training. On the other hand, quantifications are indispensable in mathematical models that attempt to accurately predict performance and, thus, be used as a tool in the organisation of the athlete's training program.

A few models have been proposed in the literature to study the training-performance relationship.





The Banister's model and its variations use antagonistic transfer functions representing the positive effects of training (leading to an increase in performance) and the negative effects (leading to shortor long-term fatigue and having a negative influence on performance). These models, close to those used in other fields that focus on the study of dose-response relationships (such as pharmacokinetics/pharmacodynamics) have been widely used in swimming (Banister et al. 1975;

Busso et al. 1994, 2003; Mujika et al. 1996). Fitting these models involve, however, some statistical problems: since there is a model per subject, a considerable number of observations per swimmer is needed to fit each model, which is unusual in non experimental studies. Also, the model structure (symmetric antagonistic functions all estimated from the same training information), leads to large confidence intervals and unidentifiable parameters (i.e., severe parameters correlation) avoiding their biological interpretation (Hellard et al. 2006). Lastly, estimation is performed assuming independence of observations, however, performances closer in time may correlate, in which case, inference may be invalid. To address these problems, Avalos et al. (2003) proposed using mixed effects regression models that include 1/the usual fixed effects for the dependent variables, thus, the effects common to the population are estimated using all the data; 11/ the random effects that allows taking into account individual heterogeneity; and Ill/the residual structure that allows taking into account that data may be correlated within subjects.

Modelling the association of training with performance in swimming raises another important issue:

how should we account for volumes at different training intensities? Mujika et al. (1996) constructed a training load by adding weighted intensity training volumes by a priori constants representing energetic intensities. Avalos et al. (2003) computed a training load as the sum of individually normalised (expressed as a percentage of the individual maximum) training intensities. Both approaches used measurements of blood lactate concentrations to determine training intensities.

These methods aimed to scale training volumes from different intensities or modes (in-water kilometres, dryland minutes) in such a way that their values lie within a comparable range. These approaches circumvent problems of dealing with a large number of correlated training variables.

Consequently, statistical models do not provide understanding of the effect of each training intensity on performance.

The objective of this study was to compare the predictive accuracy of these two normalisation approaches to others based on: a/ alternative normalisations, b/ summary scores derived from data, and c/ machine learning techniques, with recognised predictive qualities, such as Partial least squares or Lasso (Hastie et al. 2009). While alternative normalisations still leading to a single training variable,

–  –  –

Methods Data Training volumes at eight intensity levels and performances in competition of 138 professional French swimmers were collected during 20 seasons, between 1991 and 2011. Log-performance times in competition were related to the top 10 world log-performances (of the same specialty, distance, sex and year). Training intensities were determined from lactate thresholds, using measurements of blood lactate concentrations, updated several times throughout the season (Mujika et al. 1996;

Avalos et al. 2003). An incremental test to exhaustion was performed at the beginning of each season (repeated and adjusted four times per season) to determine the relationship between blood lactate concentration and swimming speed. Each subject swam 6x200-m at progressively higher percentages of their personal best competition time over this distance, until exhaustion. Lactate concentration was measured in blood samples collected from the fingertip during the 1-min recovery periods separating the 200-m swims. All swimming sessions were divided into five intensity levels according to the individual results obtained during this test: swimming speeds (1) below ~2 mmol.l-1; (2) at ~4 mmol-1, the onset of blood lactate accumulation; (3) just above ~6 mmol-1; (4) at ~10 mmol-1; and (5) at maximal swimming speed. In-water workouts were quantified in meters per week at each intensity level. Strength training included (6) dryland workouts of resistance training, (7) dryland workouts at maximal strength, and (8) general conditioning (involving cycling, running, cross-country skiing, team sports, etc.), and it was quantified in minutes of active exercise per week.

Longitudinal data modelling and criteria for evaluating models performance This is an observational study with no control on training programs. We assumed that swimmers may react differently to the same training load (interindividual differences) and over time (intra individual differences), thus we used mixed-effects models adjusted for sex, age, swim distance and specialty.

Non-linear effects were tested using fractional polynomials (Sauerbrei et al. 2007).

The comparison criterion was the cross-validated prediction error to avoid for over-optimistic results about the quality of the modelling procedures (Hastie et al. 2009). Cross-validation consists of removing each subject from the analysis, re-fitting the model on remaining participants, and then testing the prediction on the excluded subject using the new model. The procedure is repeated for all subjects and prediction rates averaged across all iterations. Analyses were stratified by distance.

Compared methods We adapted to our data the following methods: 1/the score proposed by Mujika et al. {1996), 2/the normalisation proposed by Avalos et al. (2003), 3/the normalisation 'sum of training intensities expressed as a percentage of the common maximum', 4/zero-mean and unit-variance individual normalisation, 5/zero-mean and unit-variance common normalisation, 6/a score for in-water and for dryland workouts based on univariate regressions, 7/a score for low-intensity and high-intensity inwater workouts, and for dryland workouts based on univariate regressions, 8/Partialleast squares (PLS) regression and 9/the Lasso (feast absolute shrinkage and selection operator) regression.

–  –  –

of subjects; j=1,...,n; the number of observation for the i-th subject and K=2,... 5 the in-water intensity levels. Low-intensity in-water score was computed from intensities 1 to 3, high-intensity in-waster score was computed from intensities 4 to 5.

513 BMS20 14-PROCEEDINGS The underlying assumption of Partial Least Squares (PLS) is that the observed data is generated by a system or process which is driven by a reduced number of latent (not directly observed or measured) variables. PLS creates orthogonal components by maximising the covariance between sets of independent variables and the dependent variable. Then these components serve as a new representation of the set of independent variables and the dependent variable is regressed on these new predictors. We used R (R Development Core Team, 2011) version 2.13.1 for all computations. The R package nlme and pls were used to fit mixed models and PLS, respectively. We used a two-step procedure: first the PLS is fitted assuming independent data, second the PLS components are introduced in a mixed-effects model (Guyon et al. 2011).

Finally, the Lasso is a shrinkage method that can be applied to address estimate variance inflation or convergence problems, which arise in complex regression situations, such as in the presence of a large number of variables (Hastie et al. 2009). The Lasso maximises the likelihood function, with a bound on the sum of the absolute values of the coefficients, which ensures that unstable estimates are shrunk more than stable ones. As a result, the Lasso may completely delete certain covariates, those showing no association with the dependent variable, performing both estimation and variable selection simultaneously. The R packages lmmlasso (Schelldorfer et al. 2011) and glmmLasso {Groll et al. 2011) were tested to compute the Lasso estimates for longitudinal data.

Results Summary scores for three training loads (low-intensity/high-intensity/dry-land workouts) with data derived weights showed the best results. However, cross-validated prediction errors were close relative to their variances, which were high. Figure 1 shows mean prediction errors and standard deviation estimated through cross-validation for all the tested methods excepting the Lasso. For this last method, convergence problems appeared whatever the available algorithm used.

–  –  –

Note: 1,249, 1,455 and 254 weeks-person are used in sprint, mid- and long-distances analyses, respectively. Values are means+5D.



Pages:   || 2 |


Similar works:

«SITUATION ANALYSIS AND RECOMMENDATIONS Antibiotic Use and Resistance in Tanzania The GARP-Tanzania Working Group Said Aboud, MD, PhD, Chairman Robinson Mdegela, BVSc, PhD, Vice-chairman June 2015 1 GARP TANZANIA WORKING GROUP Said Aboud, MD, PhD, Chairman, Associate Professor of Microbiology and Immunology, Muhimbili University of Health and Allied Sciences (MUHAS) Robinson Mdegela, BVSc, PhD, Vice-chairman, Professor of Veterinary Medicine, Sokoine University of Agriculture (SUA) Pastory...»

«CURRICULUM VITAE David D. Dore, PharmD, PhD CONTACT INFORMATION Office: Optum Epidemiology Home: 20 Rebecca Road 950 Winter Street S. Dartmouth, MA 02748 Suite 3800 Waltham, MA Tel: 401-741-4154 david_dore@brown.edu david.dore@optum.com EDUCATION 2008 PhD, Epidemiology (Advisor: Kate L. Lapane, PhD) Brown Medical School, Providence, RI Dissertation: “Benefits and Consequences of Drug Therapy in the Elderly” 2006−2007 Post-doctoral Fellowship (Mentor: Vincent Mor, PhD) Center for...»

«Self-Compassion in Clinical Practice Christopher K. Germer1 and Kristin D. Neff2 1 Harvard Medical School/Cambridge Health Alliance 2 University of Texas, Austin Self-compassion is conceptualized as containing 3 core components: self-kindness versus selfjudgment, common humanity versus isolation, and mindfulness versus overidentification, when relating to painful experiences. Research evidence demonstrates that self-compassion is related to psychological flourishing and reduced...»

«Package leaflet: Information for the user Ibuprofen 100mg/5ml oral solution Ibuprofen Read all of this leaflet carefully before you start taking this medicine because it contains important information for you.Keep this leaflet. You may need to read it again.If you have any further questions, ask your doctor or pharmacist.This medicine has been prescribed for you. Do not pass it on to others. It may harm them, even if their signs of illness are the same as yours. If you get any side effects,...»

«The Illustrated Guide to Assistive Technology and Devices This page intentionally left blank The Illustrated Guide to Assistive Technology and Devices Tools and Gadgets for Living Independently Suzanne Robitaille New York Visit our web site at www.demosmedpub.com © 2010 Demos Medical Publishing, LLC. All rights reserved. This book is protected by copyright. No part of it may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical,...»

«Public Health Service DEPARTMENT OF HEALTH & HUMAN SERVICES Food and Drug Administration Silver Spring, MD 20993 Cheryl Rini Director, Regulatory Affairs Vertex Pharmaceuticals, Inc. 130 Waverly St. Cambridge, MA 02139 RE: DA 201917 N INCIVEKTM (telaprevir) Film Coated Tablets MA# 54 Dear Ms. Rini, The Office of Prescription Drug Promotion (OPDP), Division of Consumer Drug Promotion (DCDP) of the U.S. Food and Drug Administration (FDA) has reviewed the James JP M. Branded Story (branded...»

«Gene expression profiling of oral cancer cells chronic exposed to areca nut extract Yi-Chen Li (李宜珍)1, Ann-Joy Cheng (鄭恩加)2 1 Graduate Institute of Biomedical Sciences, Chang Gung University, Taoyuan, 333, Taiwan 2 Department of Medical Biotechnology and Laboratory Science, Chang Gung University, Taoyuan 333, Taiwan Abstract Oral cancer is the 6th most frequent cancer in Taiwan. The habit of areca nut chewing is the main etiological factor of oral cancer. To shed light on molecular...»

«The Benefits of Medical Research and the Role of the NIH May 2000 Office of the Chairman, Connie Mack http://jec.senate.gov THE BENEFITS OF MEDICAL RESEARCH AND THE ROLE OF THE NIH May 17, 2000 THE BENEFITS OF MEDICAL RESEARCH AND THE ROLE OF THE NIH EXECUTIVE SUMMARY The NIH Leads the Battle Against Disease ! Leading the battle against disease. As the world’s leading medical research institution, the NIH funds more than 35,000 research grants each year to scientists across the country making...»

«Policy Directive Ministry of Health, NSW 73 Miller Street North Sydney NSW 2060 Locked Mail Bag 961 North Sydney NSW 2059 Telephone (02) 9391 9000 Fax (02) 9391 9101 http://www.health.nsw.gov.au/policies/ space space Maternity Management of Early Pregnancy Complications space Document Number PD2012_022 Publication date 02-May-2012 Functional Sub group Clinical/ Patient Services Maternity Clinical/ Patient Services Critical care Summary Provides policy direction for Early Pregnancy Assessment...»

«PEER CERTIFICATION: CHAIRPERSON Cindy Clafin EXECUTIVE OFFICER Jane Adcock WHAT ARE WE WAITING FOR?  Advocacy  Evaluation  Inclusion MS 2706 PO Box 997413 Sacramento, CA 95899Examining the Opportunities, Barriers, and Precedents for the Official 7413 916.323.4501 Recognition and Certification of Peer Specialists in California. fax 916.319.8030 February 2015 California Mental Health Planning Council February 2015 “When you talk to people who have been through these programs and ask...»

«B uilding Healthy Texans: A Guide to lower health care costs and more productive employees Table of Contents 1. Letter from Commissioner Worksite Wellness—Reap the Benefits of Health 2. Why Invest in Worksite Wellness?3. What Do Worksite Wellness Programs Cost?4. Choosing the Right Type of Worksite Wellness Program 5. Success Story -Public Sector 6. Success Story School District 7. How Do I Get Started? (Executive Summary of how-to) 8. Developing Management Support • Wellness Advisory Team...»

«The Health Centre Mission Statement ‘Healthy children get the most out of their education' Our aim is to enhance the educational potential of all pupils by promoting wellness and addressing any health issues which may create barriers to learning. In collaboration with parents, teachers and community resources such as the National Healthy Schools Programme (NHSP) we aim to encourage pupils to develop positive attitudes to health, promote the development of healthy life choices and ultimately...»





 
<<  HOME   |    CONTACTS
2016 www.dissertation.xlibx.info - Dissertations, online materials

Materials of this site are available for review, all rights belong to their respective owners.
If you do not agree with the fact that your material is placed on this site, please, email us, we will within 1-2 business days delete him.