Describing Patterns of Weight Changes Using Principal Components Analysis: Results from the Action for Health in Diabetes (Look AHEAD) Research Group

Department of Biostatistical Sciences, Wake Forest University School of Medicine and Winston-Salem, NC 27157, USA.
Annals of epidemiology (Impact Factor: 2). 08/2009; 19(10):701-10. DOI: 10.1016/j.annepidem.2009.06.001
Source: PubMed


To demonstrate how principal components analysis can be used to describe patterns of weight changes in response to an intensive lifestyle intervention.
Principal components analysis was applied to monthly percent weight changes measured on 2,485 individuals enrolled in the lifestyle arm of the Action for Health in Diabetes (Look AHEAD) clinical trial. These individuals were 45 to 75 years of age, with type 2 diabetes and body mass indices greater than 25 kg/m(2). Associations between baseline characteristics and weight loss patterns were described using analyses of variance.
Three components collectively accounted for 97.0% of total intrasubject variance: a gradually decelerating weight loss (88.8%), early versus late weight loss (6.6%), and a mid-year trough (1.6%). In agreement with previous reports, each of the baseline characteristics we examined had statistically significant relationships with weight loss patterns. As examples, males tended to have a steeper trajectory of percent weight loss and to lose weight more quickly than women. Individuals with higher hemoglobin A(1c) (glycosylated hemoglobin; HbA(1c)) tended to have a flatter trajectory of percent weight loss and to have mid-year troughs in weight loss compared to those with lower HbA(1c).
Principal components analysis provided a coherent description of characteristic patterns of weight changes and is a useful vehicle for identifying their correlates and potentially for predicting weight control outcomes.

Download full-text


Available from: Lawrence J Cheskin,
  • Source
    • "In medical research, particularly in oncology diseases, PCA has been used for processing data to extract significant variables and to distinguish cancer patients and healthy subjects (Madsen et al., 2010; Cheng et al., 2005; Méndez et al., 2009; Selaru et al., 2004), as an example among various others [see e.g., (Polat and Günes, 2008; Reich et al., 2008; Polat and Günes¸, 2007; Latifo˘ glu et al., 2008; Espeland et al., 2009; Spratlin et al., 2009; Edefonti et al., 2010; Xie et al., 2011; Gualberto et al., 2010)] "
    [Show abstract] [Hide abstract]
    ABSTRACT: This work aims at characterizing rat mammary tumors induced by 7,12-dimethylbenz(a)anthracene (DMBA) and the respective malignancy potential, commonly graded with histopathology features grouped by intensity levels. Tumors were described over fourteen multiple ranged microscopic parameters and a comprehensive characterization of the histological patterns and their relation with tumor grade was carried out by principal component analysis (PCA). The number of histological patterns present on a tumor tends to correlate with malignant features. High grade tumors are characterized by the presence of several structural patterns, with cribriform prevalence and necrosis. The cribriform pattern correlates with grading, i.e., tumors having a higher predominance of the cribriform pattern are likely to be more malignant. The findings may represent a benchmark for similar characterization studies in other models.
    Experimental and toxicologic pathology: official journal of the Gesellschaft fur Toxikologische Pathologie 10/2013; 66(1). DOI:10.1016/j.etp.2013.09.001 · 1.86 Impact Factor
  • Source
    • "Note that preliminary data inspection is especially important when establishing medical diagnosis procedures . In this context, PCA should be the first choice to look for patterns , aggregates, trends and outliers in the data under scrutiny [26] [27] and it has been used for identifying significant variables (which are related to the main principal components) and distinguishing patients from healthy subjects in oncologic diseases [28], as an example among various others [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Clinical tests and epidemiological studies often produce large amounts of data, being multivariate in nature. The respective analysis is, in most cases, of importance comparable to the clinical and sampling tasks. Simple, easily interpretable techniques from chemometrics provide most of the ingredients to carry out this analysis. We have selected available data from different sources pertaining to cancer diagnosis and incidence: (1) cytological diagnosis of breast cancer, (2) classification of breast tissues through parameters obtained from impedance spectra and (3) distribution of new cancer cases in the United States. Hierarchical cluster analysis (HCA) is needed especially in cases where there is no a priori identification of classes, suggesting a structure of the data based on clusters. These clusters or the classes, are then further detailed and rationalized by principal component analysis (PCA). Partial least squares (PLS) and linear discriminant analysis (LDA) provide further insight into the systems. An additional step for understanding the data set is the removal of less characteristic data (NR) using a density-based approach, so as to make it more clearly defined. Results clearly reveal that breast cytology diagnosis relies on variables conveying mostly the same type of information, being thus interchangeable in nature. In the study on tissue characterization by electrical measurements, the distribution of the different types of tissues can be easily constructed. Finally, the distribution of new cancer cases possesses clear, easily unravelled, geographical patterns.
    Journal of Biomedical Informatics 09/2012; 46(1). DOI:10.1016/j.jbi.2012.09.005 · 2.19 Impact Factor

  • Obesity Facts 12/2009; 2(6):339-41. DOI:10.1159/000268077 · 2.25 Impact Factor
Show more